Ethernet Networks Design Implementation Organiz..

Ethernet Networks Design Implementation Organiz..
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
ethernet
networks
Fourth Edition
Books by Gilbert Held, published by Wiley
Quality of Service in a Cisco Networking Environment
0 470 84425 6 (April 2002)
Bulletproofing TCP/IP-Based Windows NT/2000 Networks
0 471 49507 7 (April 2001)
Understanding Data Communications: From Fundamentals to Networking,
Third Edition
0 471 62745 3 (October 2000)
High Speed Digital Transmission Networking: Covering T/E-Carrier
Multiplexing, SONET and SDH, Second Edition
0 471 98358 6 (April 1999)
Data Communications Networking Devices: Operation, Utilization and LAN
and WAN Internetworking, Fourth Edition
0 471 97515 X (November 1998)
Dictionary of Communications Technology: Terms, Definitions and
Abbreviations, Third Edition
0 471 97517 6 (May 1998)
Internetworking LANs and WANs: Concepts, Techniques and Methods,
Second Edition
0 471 97514 1 (May 1998)
LAN Management with SNMP and RMON
0 471 14736 2 (September 1996)
ethernet
networks
Fourth Edition
♦
♦
♦
♦
Design
Implementation
Operation
Management
GILBERT HELD
4-Degree Consulting, Macon, Georgia, USA
Copyright  2003
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,
West Sussex PO19 8SQ, England
Telephone (+44) 1243 779777
Email (for orders and customer service enquiries): [email protected]
Visit our Home Page on www.wileyeurope.com or www.wiley.com
All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval
system or transmitted in any form or by any means, electronic, mechanical, photocopying,
recording, scanning or otherwise, except under the terms of the Copyright, Designs and
Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency
Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of
the Publisher. Requests to the Publisher should be addressed to the Permissions Department,
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ,
England, or emailed to [email protected], or faxed to (+44) 1243 770571.
This publication is designed to provide accurate and authoritative information in regard to
the subject matter covered. It is sold on the understanding that the Publisher is not engaged
in rendering professional services. If professional advice or other expert assistance is
required, the services of a competent professional should be sought.
Other Wiley Editorial Offices
John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA
Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA
Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany
John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia
John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark,
Singapore 129809
John Wiley & Sons Canada Ltd, 22 Worcester Road, Etobicoke, Ontario, Canada M9W 1L1
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN 0-470-84476-0
Typeset in 10.5/13pt Melior by Laserwords Private Limited, Chennai, India
Printed and bound in Great Britain by Biddles Ltd, Guildford and King’s Lynn
This book is printed on acid-free paper responsibly manufactured from sustainable forestry
in which at least two trees are planted for each one used for paper production.
For the past decade I have been most fortunate in being able to teach graduate
courses that were truly enjoyable to teach. In doing so I have been able to
tailor my presentation of technical information covering LAN performance
and other data communications topics, providing a two-way learning facility
and enhancing my presentation skills. Thus, I would be remiss if I did not
thank the students at Georgia College and State University as well as Dr Harvey
Glover for providing me with the opportunity to teach. In doing so I would
like to dedicate this book to those who pursue higher education.
contents
Preface xv
Acknowledgments
Chapter 1
Introduction to Networking Concepts
1.1
1.2
1.3
Chapter 2
xix
WIDE AREA NETWORKS 2
COMPUTER-COMMUNICATIONS EVOLUTION 2
REMOTE BATCH TRANSMISSION 2
IBM 3270 INFORMATION DISPLAY SYSTEM 3
NETWORK CONSTRUCTION 5
NETWORK CHARACTERISTICS 8
LOCAL AREA NETWORKS 8
COMPARISON TO WANS 9
TECHNOLOGICAL CHARACTERISTICS 14
TRANSMISSION MEDIUM 22
ACCESS METHOD 29
WHY ETHERNET 33
Networking Standards
2.1
2.2
1
37
STANDARDS ORGANIZATIONS 37
NATIONAL STANDARDS ORGANIZATIONS 38
INTERNATIONAL STANDARDS ORGANIZATIONS 39
THE ISO REFERENCE MODEL 40
LAYERED ARCHITECTURE 41
OSI LAYERS 42
DATA FLOW 46
vii
viii
contents
2.3
2.4
2.5
Chapter 3
IEEE 802 STANDARDS 48
802 COMMITTEES 48
DATA LINK SUBDIVISION 51
INTERNET STANDARDS 55
RFC EVOLUTION 56
TYPES AND SUBMISSION 56
OBTAINING RFCS 57
CABLING STANDARDS 57
EIA/TIA-568 58
UTP CATEGORIES 59
CABLE SPECIFICATIONS 60
OTHER METRICS 61
CAT 5E AND CAT 6 63
Ethernet Networks
3.1
3.2
3.3
3.4
65
ETHERNET 65
EVOLUTION 66
NETWORK COMPONENTS 66
THE 5-4-3 RULE 73
IEEE 802.3 NETWORKS 74
NETWORK NAMES 74
10BASE-5 75
10BASE-2 79
10BROAD-36 87
1BASE-5 89
10BASE-T 90
USE OF FIBER-OPTIC TECHNOLOGY 100
FOIRL 100
OPTICAL TRANSCEIVER 101
FIBER HUBS 101
FIBER ADAPTER 102
WIRE AND FIBER DISTANCE LIMITS 102
HIGH-SPEED ETHERNET 108
ISOCHRONOUS ETHERNET 108
FAST ETHERNET 110
100VG-ANYLAN 133
contents
3.5
3.6
Chapter 4
GIGABIT ETHERNET 138
COMPONENTS 138
MEDIA SUPPORT 141
10 GIGABIT ETHERNET 149
RATIONALE 149
ARCHITECTURE 150
OPERATING RATES 153
Frame Operations
4.1
4.2
4.3
4.4
4.5
155
FRAME COMPOSITION 155
PREAMBLE FIELD 156
START-OF-FRAME DELIMITER FIELD 157
DESTINATION ADDRESS FIELD 157
SOURCE ADDRESS FIELD 159
TYPE FIELD 164
LENGTH FIELD 166
DATA FIELD 168
FRAME CHECK SEQUENCE FIELD 168
INTERFRAME GAP 169
MEDIA ACCESS CONTROL 169
TRANSMIT MEDIA ACCESS MANAGEMENT 171
SERVICE PRIMITIVES 175
PRIMITIVE OPERATIONS 175
HALF- VERSUS FULL-DUPLEX OPERATION 176
LOGICAL LINK CONTROL 177
TYPES AND CLASSES OF SERVICE 179
SERVICE PRIMITIVES 181
OTHER ETHERNET FRAME TYPES 181
ETHERNET-802.3 181
ETHERNET-SNAP 182
IEEE 802.1Q FRAME 183
FRAME DETERMINATION 184
FAST ETHERNET 185
START-OF-STREAM DELIMITER 186
END-OF-STREAM DELIMITER 186
ix
x
contents
4.6
4.7
Chapter 5
GIGABIT ETHERNET 186
CARRIER EXTENSION 186
FRAME BURSTING 189
10 GIGABIT ETHERNET 190
Networking Hardware and Software
5.1
5.2
5.3
5.4
191
WIRED NETWORK HARDWARE COMPONENTS 192
REPEATERS 192
BRIDGES 195
ROUTERS 205
BROUTERS 210
GATEWAY 212
FILE SERVERS 214
WIRE HUBS 218
INTELLIGENT HUBS 219
SWITCHING HUBS 219
WIRELESS NETWORK HARDWARE COMPONENTS 221
NETWORK TOPOLOGIES 221
ACCESS POINT 221
WIRELESS ROUTER 222
WIRELESS BRIDGE 223
NETWORKING SOFTWARE 224
DOS 224
NETWORK SOFTWARE COMPONENTS 225
NETWORK OPERATING SYSTEMS 227
APPLICATION SOFTWARE 242
THE TCP/IP PROTOCOL SUITE 243
OVERVIEW 244
PROTOCOL DEVELOPMENT 244
THE TCP/IP STRUCTURE 245
DATAGRAMS VERSUS VIRTUAL CIRCUITS 247
ICMP 249
ARP 252
TCP 254
UDP 259
IP 260
contents
DOMAIN NAME SERVICE 269
NAME SERVER 272
TCP/IP CONFIGURATION 272
OPERATING MULTIPLE STACKS 275
Chapter 6
Bridging and Switching Methods and Performance
Issues 279
6.1
6.2
6.3
6.4
6.5
BRIDGING METHODS 279
ADDRESS ISSUES 280
TRANSPARENT BRIDGING 280
SPANNING TREE PROTOCOL 283
PROTOCOL DEPENDENCY 291
SOURCE ROUTING 292
SOURCE ROUTING TRANSPARENT BRIDGES 297
BRIDGE NETWORK UTILIZATION 299
SERIAL AND SEQUENTIAL BRIDGING 300
PARALLEL BRIDGING 301
STAR BRIDGING 302
BACKBONE BRIDGING 302
BRIDGE PERFORMANCE ISSUES 302
TRAFFIC FLOW 303
NETWORK TYPES 304
TYPE OF BRIDGE 304
ESTIMATING NETWORK TRAFFIC 304
PREDICTING THROUGHPUT 310
LAN SWITCHES 312
RATIONALE 313
BOTTLENECKS 314
CONGESTION-AVOIDANCE OPTIONS 314
LAN SWITCH OPERATIONS 318
SWITCH BASIC ARCHITECTURE 332
COMPONENTS 332
SWITCH FEATURES 334
SWITCHED-BASED VIRTUAL LANS 348
SWITCH USAGE 360
LAYER 3 AND LAYER 4 SWITCHING 364
xi
xii
contents
Chapter 7
Routers 365
7.1
7.2
7.3
7.4
7.5
7.6
Chapter 8
ROUTER OPERATION 365
IP SUPPORT OVERVIEW 365
BASIC OPERATION AND USE OF ROUTING TABLES 368
NETWORKING CAPABILITY 370
COMMUNICATION, TRANSPORT, AND ROUTING PROTOCOLS 371
COMMUNICATION PROTOCOL 371
ROUTING PROTOCOL 371
HANDLING NONROUTABLE PROTOCOLS 372
TRANSPORT PROTOCOL 373
ROUTER CLASSIFICATIONS 374
PROTOCOL-DEPENDENT ROUTERS 374
PROTOCOL-INDEPENDENT ROUTERS 377
ROUTING PROTOCOLS 381
TYPES OF ROUTING PROTOCOLS 381
INTERIOR DOMAIN ROUTING PROTOCOLS 381
EXTERIOR DOMAIN ROUTING PROTOCOLS 382
TYPES OF INTERIOR DOMAIN ROUTING PROTOCOLS 383
ROUTING INFORMATION PROTOCOL 386
CONFIGURATION EXAMPLE 389
ROUTING TABLE MAINTENANCE PROTOCOL 392
INTERIOR GATEWAY ROUTING PROTOCOL 393
LINK STATE PROTOCOLS 394
FILTERING 397
FILTERING EXPRESSIONS 400
FILTERING EXAMPLES 401
ROUTER ACCESS LISTS 402
PERFORMANCE CONSIDERATIONS 404
Wireless Ethernet
8.1
407
OVERVIEW 407
NETWORK TOPOLOGY 409
ROAMING 411
PHYSICAL LAYER OPERATIONS 412
HIGH-SPEED WIRELESS LANS 415
ACCESS METHOD 418
contents
8.2
8.3
Chapter 9
FRAME FORMATS 420
DATA FRAME 421
CONTROL FIELD 422
CONTROL FRAMES 428
MANAGEMENT FRAMES 429
PHYSICAL PROTOCOL DATA UNITS 432
DEPLOYMENT 434
WIRELESS PC NETWORK ADAPTER CARDS 434
ACCESS POINT 435
COMBINED ROUTER/ACCESS POINT 436
WIRELESS BRIDGE 439
ROUTER/ACCESS POINT CONFIGURATION 439
CLIENT CONFIGURATION 441
Security
9.1
9.2
9.3
447
THE SECURITY ROLE OF THE ROUTER 447
ACCESS CONTROL 448
ACCESS LISTS 457
STANDARD IP ACCESS LISTS 459
EXTENDED IP ACCESS LISTS 462
ANTI-SPOOFING STATEMENTS 471
NAMED ACCESS LISTS 472
DYNAMIC ACCESS LISTS 474
REFLEXIVE ACCESS LISTS 478
TIME-BASED ACCESS LISTS 482
CONTEXT BASED ACCESS CONTROL 483
THE ROLE OF THE FIREWALL 494
ACCESS-LIST LIMITATIONS 494
PROXY SERVICES 496
FIREWALL LOCATION 498
THE TECHNOLOGIC INTERCEPTOR 504
CHECKPOINT FIREWALL-1 510
THE ROLE OF THE VIRUS SCANNER
AND ENCRYPTION 516
VIRUS OVERVIEW 516
TYPES OF VIRUSES 517
INFECTION PREVENTION 518
xiii
xiv
contents
DESKTOP SCANNING 519
EMAIL SCANNING 524
RECOGNIZING INFECTION SYMPTOMS
Chapter 10
Managing the Network 531
10.1
10.2
10.3
10.4
Chapter 11
11.2
581
SNMP 531
BASIC COMPONENTS 532
OPERATION 533
REMOTE MONITORING 535
OPERATION 535
THE RMON MIB 536
MANAGING REMOTE NETWORKS 539
OTHER NETWORK MANAGEMENT FUNCTIONS 541
CONFIGURATION MANAGEMENT 542
PERFORMANCE MANAGEMENT 543
FAULT MANAGEMENT 543
ACCOUNTING MANAGEMENT 543
SECURITY MANAGEMENT 544
REPRESENTATIVE NETWORK MANAGEMENT PROGRAMS 544
TRITICOM ETHERVISION 545
CINCO NETWORK’S WEBXRAY 554
WILDPACKETS ETHERPEEK 559
The Future of Ethernet
11.1
Index
528
567
ETHERNET TRENDS 567
NETWORK ADAPTER CARD COST 567
FUTURE PRICE DIRECTION 568
NETWORK PERFORMANCE CONSIDERATIONS 570
SUPPLEMENTING AN EXISTING NETWORK 571
SUMMARY 579
preface
In a prior edition of this book the preface commenced with the paraphrase of
an old adage in an era of evolving local area networking technology: Ethernet
is dead — long live Ethernet!
Although advances in communications technology continue to occur at a
rapid pace, that paraphrase continues to be valid. Within the past decade, the
bandwidth of 10 Mbps Ethernet was advanced by a factor of one thousand with
the introduction of a series of enhancements to the original Ethernet specification. First, Fast Ethernet resulted in the bandwidth of Ethernet increasing
by a factor of 10 to 100 Mbps. The introduction of Gigabit Ethernet resulted in
another order of magnitude increase in bandwidth to 1 Gbps. Although many
persons felt that a transmission capacity of 1 Gbps would be more than sufficient for the foreseeable future, another adage states that many applications
will grow to use all available bandwidth. While most organizations may be
hard pressed to use 1 Gbps of bandwidth, other organizations, including Internet Service Providers and corporations and universities with large backbone
LANs, were able to literally fill the 1 Gbps pipe, resulting in the development
of 10 Gbps Ethernet. Thus, over the past decade Ethernet’s 10 Mbps operation
has increased by a factor of 1000 to 10 Gbps.
This new edition provides a significant amount of additional material to
most of the chapters of this book’s previous edition. New information added
includes coverage of the transmission of Gigabit over copper conductors, the
evolution of cabling standards that facilitate support of higher Ethernet operating rates, and the manner by which LAN switches operate on Ethernet frames
transporting information at higher layers in the Open System Interconnection
Reference Model.
Recognizing the importance of networking without wires, a new chapter
is focused upon wireless Ethernet. This chapter describes and discusses the
series of IEEE 802.11 standards and provides practical information concerning
the setup and operation of a wireless LAN. Recognizing the importance
of security in the modern era of networking resulted in the movement of
most security related topics to a new chapter focused on this topic. This
chapter considerably expands the prior disparate coverage of security by
adding information covering the use of firewalls in both a wired and wireless
xv
xvi
preface
environment. In addition, information concerning the use of router access lists
is considerably expanded, while new information covering authentication,
authorization and accounting has been added to the chapter.
Other topics that have been added or significantly revised in this new edition
include the operation of new versions of Windows on Ethernet LANs, the
operation and utilization of LAN switches above layer 2 in the ISO Reference
Model, new gateway methods you can consider to connect workstation users to
mainframes, and the use of both copper and fiber optic to transport high-speed
Ethernet. Thus, the scope and depth of material have been significantly revised
and updated to continue to provide you with detailed information concerning
the design, implementation, operation and management of different types of
Ethernet networks.
This book incorporates into one reference source the material you will
need to understand how Ethernet networks operate, the constraints and performance issues that affect their design and implementation, and how their
growth and use can be managed both locally and as part of an enterprise network. Assuming readers have varied backgrounds in communications terms
and technology, the first two chapters were written to provide a common foundation of knowledge. Those chapters cover networking concepts and network
standards — two topics on which material in succeeding chapters is based.
Succeeding chapters examine Ethernet concepts: frame operations; network
construction; the use of bridges, routers, hubs, switches, and gateways; Internet connectivity; network backbone construction; Wireless Ethernet; Security;
and the management of Ethernet networks.
In writing this book, my primary goal was to incorporate practical information you can readily use in designing, operating, implementing, and managing
an Ethernet network. Although Ethernet had its origins in the 1970s and
can be considered a relatively ‘‘old’’ technology, in reality, the technology
is anything but old. Only a few years ago, the standardization of what is
now known as 10BASE-T (a twisted-wire version of Ethernet) resulted in a
considerable expansion in the use of this type of local area network. By 1994
the use of intelligent switches greatly enhanced the operational capability of
10BASE-T networks, providing multiple simultaneous 10 Mbps connectivity.
During 1995 high-speed Ethernet technology in the form of Fast Ethernet
products provided users with the ability to upgrade their Ethernet networks
to satisfy emerging multimedia requirements. Within a few years industry
realized that emerging applications, as well as the growth in the use of the
Internet, required higher-speed backbone LANs as a mechanism to support
Internet access and similar high-speed networking requirements. This realization resulted in the deployment of Gigabit Ethernet hubs and switches during
preface
xvii
1997, which was quickly followed by 10 Gbps operations a few years later.
Thus, Ethernet technology can be expected to continue to evolve to satisfy the
communications requirements of business, government, and academia.
For over 30 years I have worked as a network manager responsible for the
design, operation, and management of an enterprise network in which local
area networks throughout the United States are interconnected through the
use of different wide area network transmission facilities. This challenging
position has provided me with the opportunity to obtain practical experience
in designing, operating, and interconnecting Ethernet networks to TokenRing, SNA, the Internet, and other types of networks — experience which
I have attempted to share with you. This book will help you consider the
practicality of different types of routing protocols, LAN switches, and gateway
methods. These and other network design issues are crucial to the efficient
and effective expansion of a local Ethernet so that users on that network can
access resources on other networks.
As a professional author, I very much value readers’ comments. Those
comments provide me with feedback necessary to revise future editions so
that they better reflect the information requirements of readers. I look forward
to receiving your comments, as well as suggestions for information you would
like to see in a future edition of this book. You can write to me directly or
through my publisher, whose address you can find on page 4 of this book or
communicate with me directly via electronic mail at gil− [email protected]
Gilbert Held
Macon, GA
acknowledgments
This book would not have been possible without the work of two people
whose foresight and pioneering efforts were instrumental in the development
of the technology upon which Ethernet is based.
One of the key concepts behind Ethernet — that of allocating the use of a
shared channel — can be traced to the pioneering efforts of Dr Norman Abramson and his colleagues at the University of Hawaii during the early 1970s. The
actual development of Ethernet is due to the foresight of Dr Robert Metcalfe.
Working at the Xerox Palo Alto Research Center in Palo Alto, California,
Dr Metcalfe headed a development team that interconnected over 100 computers on a 1-km cable using a carrier sense multiple access collision detection
(CSMA/CD) protocol. In addition to pioneering the technical development of
Ethernet, Dr Metcalfe coined its name, after the luminiferous ether through
which electromagnetic radiation was once thought to propagate. I would be
remiss if I did not thank Dr Abramson, Dr Metcalfe, and their colleagues for
their visionary efforts in developing the technology through which hundreds
of millions of people now communicate.
Writing and producing a book about technology requires not only the
technology itself, but also the efforts of many individuals. First and foremost,
I would like to thank my family for their understanding for the nights and
weekends I disappeared to write this book. Once again, I am indebted to
Mrs Linda Hayes and Mrs Susan Corbitt for taking my notes and drawings
and converting them into a manuscript. Last, but not least, I would like to
thank Birgit Gruber and Ann-Marie Halligan as well as the production staff
at John Wiley & Sons for backing the new edition of this book, as well as in
facilitating the conversion of my manuscript into the book you are reading.
xix
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter one
Introduction to Networking
Concepts
One of the most logical assumptions an author can make is that readers will
have diverse backgrounds of knowledge and experience. Making this book
as valuable as possible to persons with different backgrounds requires an
introductory chapter that covers basic networking concepts. Unfortunately,
basic concepts for one person may not be the same as basic concepts for
another person, which presents an interesting challenge for an author.
To meet this challenge, this book takes two courses of action. First, it assumes
that some readers will have limited knowledge about the different types of
communications systems available for transporting information, the relationship between wide area networks (WANs) and local area networks (LANs),
and the relationships among different types of local area networks. Thus, this
introductory chapter was written to provide those readers with a basic level
of knowledge concerning these important topics. Secondly, readers who are
already familiar with these basic concepts may wish to consult individual
chapters separately, rather than reading through the entire book. To satisfy
those readers, each chapter was written to be as independent as possible from
preceding and succeeding chapters. Thus, readers who are familiar with wide
and local area networking concepts, as well as the technical characteristics of
LANs, may elect to skim or bypass this chapter. For other readers, information
contained in this chapter will provide a level of knowledge that will make
succeeding chapters more meaningful.
In this introductory chapter, we will first focus our attention on the key concepts behind the construction of wide area networks and local area networks.
In doing so, we will examine each type of network to obtain an understanding
of its primary design goal. Next, we will compare and contrast their operations
and utilizations to obtain an appreciation for the rationale behind the use of
different types of local area networks.
1
2
chapter one
Although this book is about Ethernet networks, there are other types of local
area networks that provide a viable data transportation highway for millions
of users. By reviewing the technological characteristics of different types of
LANs, we will obtain an appreciation for the governing characteristics behind
the use of different local area networks. In addition, because many local
area networks are connected to other LANs and WANs, we will conclude this
chapter by focusing on the technological characteristics of local area networks.
This will form a foundation for discussing a variety of Ethernet networking
issues in succeeding chapters of this book.
1.1 Wide Area Networks
The evolution of wide area networks can be considered to have originated
in the mid- to late 1950s, commensurate with the development of the first
generation of computers. Based on the use of vacuum tube technology, the first
generation of computers were large, power-hungry devices whose placement
resulted in a focal point for data processing and the coinage of the term
data center.
Computer-Communications Evolution
Originally, access to the computational capability of first-generation computers was through the use of punched cards. After an employee of the
organization used a keypunch to create a deck of cards, that card deck was
submitted to a window in the data center, typically labeled input/output (I/O)
control. An employee behind the window would accept the card deck and
complete a form that contained instructions for running the submitted job.
The card deck and instructions would then be sent to a person in production control, who would schedule the job and turn it over to operations for
execution at a predefined time. Once the job was completed, the card deck
and any resulting output would be sent back to I/O control, enabling the job
originator to return to the window in the data center to retrieve his or her
card deck and the resulting output. With a little bit of luck, programmers
might see the results of their efforts on the same day that they submitted
their jobs.
Because the computer represented a considerable financial investment for
most organizations, it was understandable that these organizations would
be receptive to the possibility of extending their computers’ accessibility.
By the mid-1960s, several computer manufacturers had added remote access
capabilities to one or more of their computers.
introduction to networking concepts
3
Remote Batch Transmission
One method of providing remote access was the installation of a batch
terminal at a remote location. That terminal was connected via a telephone
company–supplied analog leased line and a pair of modems to the computer
in the corporate data center.
The first type of batch terminal developed to communicate with a data
center computer contained a card reader, a printer, a serial communications
adapter, and hard-wired logic in one common housing. The serial communications adapter converted the parallel bits of each internal byte read from the
card reader into a serial data stream for transmission. Similarly, the adapter
performed a reverse conversion process, converting a sequence of received
serial bits into an appropriate number of parallel bits to represent a character
internally within the batch terminal. Because the batch terminal was located
remotely from the data center, it was often referred to as a remote batch
terminal, while the process of transmitting data was referred to as remote
batch transmission. In addition, the use of a remote terminal as a mechanism
for grouping card decks of individual jobs, all executed at the remote data
center, resulted in the term remote job entry terminal being used as a name
for this device.
Figure 1.1 illustrates in schematic form the relationships between a batch
terminal, transmission line, modems, and the data center computer. Because
the transmission line connects a remote batch terminal in one geographic area
to a computer located in a different geographic area, Figure 1.1 represents one
of the earliest types of wide area data communications networks.
Paralleling the introduction of remote batch terminals was the development
of a series of terminal devices, control units, and specialized communications equipment, which resulted in the rapid expansion of interactive
computer applications. One of the most prominent collections of products was
introduced by the IBM Corporation under the trade name 3270 Information
Display System.
Batch
terminal
Transmission
line
Modem
Modem
Computer
Figure 1.1 Remote batch transmission. The transmission of data from a
remote batch terminal represents one of the first examples of wide area data
communications networks.
4
chapter one
IBM 3270 Information Display System
The IBM 3270 Information Display System was a term originally used to
describe a collection of products ranging from interactive terminals that
communicate with a computer, referred to as display stations, through several types of control units and communications controllers. Later, through
the introduction of additional communications products from IBM and
numerous third-party vendors and the replacement of previously introduced products, the IBM 3270 Information Display System became more
of a networking architecture and strategy rather than a simple collection
of products.
First introduced in 1971, the IBM 3270 Information Display System was
designed to extend the processing power of the data center computer to
remote locations. Because the data center computer typically represented the
organization’s main computer, the term mainframe was coined to refer to a
computer with a large processing capability. As the mainframe was primarily
designed for data processing, its utilization for supporting communications
degraded its performance.
Communications Controller
To offload communications functions from the mainframe, IBM and other
computer manufacturers developed hardware to sample communications
lines for incoming bits, group bits into bytes, and pass a group of bytes
to the mainframe for processing. This hardware also performed a reverse
function for data destined from the mainframe to remote devices. When
first introduced, such hardware was designed using fixed logic circuitry,
and the resulting device was referred to as a communications controller.
Later, minicomputers were developed to execute communications programs,
with the ability to change the functionality of communications support by
the modification of software — a considerable enhancement to the capabilities of this series of products. Because both hard-wired communications
controllers and programmed minicomputers performing communications
offloaded communications processing from the mainframe, the term frontend processor evolved to refer to this category of communications equipment.
Although most vendors refer to a minicomputer used to offload communications processing from the mainframe as a front-end processor, IBM
has retained the term communications controller, even though their fixed
logic hardware products were replaced over 20 years ago by programmable
minicomputers.
introduction to networking concepts
5
Control Units
To reduce the number of controller ports required to support terminals, as
well as the amount of cabling between controller ports and terminals, IBM
developed poll and select software to support its 3270 Information Display
System. This software enabled the communications controller to transmit
messages from one port to one or more terminals in a predefined group
of devices. To share the communications controller port, IBM developed
a product called a control unit, which acts as an interface between the
communications controller and a group of terminals.
In general terms, the communications controller transmits a message to the
control unit. The control unit examines the terminal address and retransmits
the message to the appropriate terminal. Thus, control units are devices that
reduce the number of lines required to link display stations to mainframe computers. Both local and remote control units are available; the key differences
between them are the method of attachment to the mainframe computer and
the use of intermediate devices between the control unit and the mainframe.
Local control units are usually attached to a channel on the mainframe,
whereas remote control units are connected to the mainframe’s front-end
processor, which is also known as a communications controller in the IBM
environment. Because a local control unit is within a limited distance of the
mainframe, no intermediate communications devices, such as modems or data
service units, are required to connect a local control unit to the mainframe.
In comparison, a remote control unit can be located in another building or
in a different city; it normally requires the utilization of intermediate communications devices, such as a pair of modems or a pair of data service
units, for communications to occur between the control unit and the communications controller. The relationship of local and remote control units to
display stations, mainframes, and a communications controller is illustrated
in Figure 1.2.
Network Construction
To provide batch and interactive access to the corporate mainframe from
remote locations, organizations began to build sophisticated networks. At
first, communications equipment such as modems and transmission lines was
obtainable only from AT&T and other telephone companies. Beginning in
1974 in the United States with the well-known Carterphone decision, competitive non–telephone company sources for the supply of communications
equipment became available. The divestiture of AT&T during the 1980s and
6
chapter one
Terminal
Communications
controller
Modem*
Modem*
Remote
control
unit
•
•
•
Terminal
Mainframe
Terminal
Local
control
unit
•
•
Terminal
*Note: Modems replaced by data service units when a digital transmission facility used.
Figure 1.2
Relationship of 3270 information display products.
the emergence of many local and long-distance communications carriers
paved the way for networking personnel to be able to select from among
several or even hundreds of vendors for transmission lines and communications equipment.
As organizations began to link additional remote locations to their mainframes, the cost of providing communications began to escalate rapidly.
This, in turn, provided the rationale for the development of a series of linesharing products referred to as multiplexers and concentrators. Although
most organizations operated separate data and voice networks, in the mid1980s communications carriers began to make available for commercial use
high-capacity circuits known as T1 in North America and E1 in Europe.
Through the development of T1 and E1 multiplexers, voice, data, and video
transmission can share the use of common high-speed circuits. Because the
interconnection of corporate offices with communications equipment and
facilities normally covers a wide geographical area outside the boundary
of one metropolitan area, the resulting network is known as a wide area
network (WAN).
Figure 1.3 shows an example of a wide area network spanning the continental United States. In this example, regional offices in San Francisco and New
York are connected with the corporate headquarters, located in Atlanta, via T1
multiplexers and T1 transmission lines operating at 1.544 Mbps. Assuming
that each T1 multiplexer is capable of supporting the direct attachment of
a private branch exchange (PBX), both voice and data are carried by the T1
circuits between the two regional offices and corporate headquarters. The
three T1 circuits can be considered the primary data highway, or backbone,
of the corporate network.
introduction to networking concepts
7
New York, NY
PBX
San Francisco, CA
T1
MUX
PBX
Computer
T
T
T
T1
MUX
Computer
T
T1
1.544 Mbps
digital
circuits
New Haven, CT
T
19.2 Kbps
Sacramento, CA
STDM
Analog circuit
T
T
19.2 Kbps
STDM
Analog circuit
T
Atlanta, GA
PBX
T1
MUX
Macon, GA
T
56 Kbps digital circuit
STDM
T
Legend:
T
= Data Terminal
PBX = Private Branch Exchange
STDM = Statistical Time Division Multiplexer
Figure 1.3 Wide area network example. A WAN uses telecommunications
lines obtained from one or more communications carriers to connect geographically dispersed locations.
In addition to the three major corporate sites that require the ability to route
voice calls and data between locations, let us assume that the corporation
also has three smaller area offices located in Sacramento, California; Macon,
Georgia; and New Haven, Connecticut. If these locations only require data
terminals to access the corporate network for routing to the computers located
8
chapter one
in San Francisco and New York, one possible mechanism to provide network
support is obtained through the use of tail circuits. These tail circuits could
be used to connect a statistical time division multiplexer (STDM) in each area
office, each serving a group of data terminals to the nearest T1 multiplexer,
using either analog or digital circuits. The T1 multiplexer would then be
configured to route data terminal traffic over the corporate backbone portion
of the network to its destination.
Network Characteristics
There are certain characteristics we can associate with wide area networks.
First, the WAN is typically designed to connect two or more geographical areas.
This connection is accomplished by the lease of transmission facilities from
one or more communications vendors. Secondly, most WAN transmission
occurs at or under a data rate of 1.544 Mbps or 2.048 Mbps, which are the
operating rates of T1 and E1 transmission facilities.
A third characteristic of WANs concerns the regulation of the transmission
facilities used for their construction. Most, if not all, transmission facilities
marketed by communications carriers are subject to a degree of regulation at
the federal, state, and possibly local government levels. Even though we now
live in an era of deregulation, carriers must seek approval for many offerings
before making new facilities available for use. In addition, although many
of the regulatory controls governing the pricing of services were removed,
the communications market is still not a truly free market. Thus, regulatory
agencies at the federal, state, and local levels still maintain a degree of
control over both the offering and pricing of new services and the pricing of
existing services.
1.2 Local Area Networks
The origin of local area networks can be traced, in part, to IBM terminal equipment introduced in 1974. At that time, IBM introduced a series of terminal
devices designed for use in transaction-processing applications for banking
and retailing. What was unique about those terminals was their method of connection: a common cable that formed a loop provided a communications path
within a localized geographical area. Unfortunately, limitations in the data
transfer rate, incompatibility between individual IBM loop systems, and other
problems precluded the widespread adoption of this method of networking.
The economics of media sharing and the ability to provide common access
introduction to networking concepts
9
to a centralized resource were, however, key advantages, and they resulted
in IBM and other vendors investigating the use of different techniques to
provide a localized communications capability between different devices. In
1977, Datapoint Corporation began selling its Attached Resource Computer
Network (ARCNet), considered by most people to be the first commercial local
area networking product. Since then, hundreds of companies have developed
local area networking products, and the installed base of terminal devices
connected to such networks has increased exponentially. They now number
in the hundreds of millions.
Comparison to WANs
Local area networks can be distinguished from wide area networks by geographic area of coverage, data transmission and error rates, ownership,
government regulation, and data routing — and, in many instances, by the
type of information transmitted over the network.
Geographic Area
The name of each network provides a general indication of the scope of
the geographic area in which it can support the interconnection of devices.
As its name implies, a LAN is a communications network that covers a
relatively small local area. This area can range in scope from a department
located on a portion of a floor in an office building, to the corporate staff
located on several floors in the building, to several buildings on the campus
of a university.
Regardless of the LAN’s area of coverage, its geographic boundary will be
restricted by the physical transmission limitations of the local area network.
These limitations include the cable distance between devices connected to the
LAN and the total length of the LAN cable. In comparison, a wide area network
can provide communications support to an area ranging in size from a town
or city to a state, country, or even a good portion of the entire world. Here,
the major factor governing transmission is the availability of communications
facilities at different geographic areas that can be interconnected to route data
from one location to another.
To better grasp the differences between LANs and WANs, today we can view
the LAN as being analogous to our local telephone company, while the WAN
can be compared with the long-distance communications carrier. However,
this may not be true in the future when local telephone companies obtain
permission to offer long-distance service and long-distance communications
10
chapter one
carriers obtain regulatory approval to offer local telephone service. However,
for the present we will presume that telephone support in different cities is
provided by the local telephone company in each city. Thus, for calls between
cities, the local telephone companies must connect to the long-distance carrier.
Similarly, we can have separate LANs in different cities or within different
buildings in the same city; however, to interconnect those LANs we would
normally require a wide area network.
Until the turn of the millennium these differences between LANs and
WANs with the respect to geographic area of coverage were distinct and
easy to recognize. However, within the past two years a new role has been
proposed for Gigabit Ethernet that could enable this technology to function
as a miniature WAN. As we will note later in this book, the ability to
transmit Gigabit Ethernet over optical fiber makes it possible to transmit this
technology over extended distances. In fact, by the time you read this book
Gigabit Ethernet may provide a low-cost alternative to synchronous optical
networking (SONET) transmission.
Data Transmission and Error Rates
Two additional areas that differentiate LANs from WANs and explain the
physical limitation of the LAN geographic area of coverage are the data
transmission rate and error rate for each type of network. Older LANs, such
as the original version of Ethernet and Token-Ring, normally operate at a
low megabit-per-second rate, typically ranging from 4 Mbps to 16 Mbps. More
modern high-speed Ethernet networks, such as Fast Ethernet that operates
at 100 Mbps, Gigabit Ethernet and 10 Gigabit Ethernet provide transmission
rates of 1 Gbps and 10 Gbps, respectively. In comparison, the communications
facilities used to construct a major portion of most WANs provide a data
transmission rate at or under the T1 and E1 data rates of 1.544 Mbps and
2.048 Mbps.
Although some readers may have encountered references to WAN transmission rates of 10 and 40 Gbps in various trade literature, the use of optical
carriers (OCs) at those data rates is primarily by communications carriers
whose transmission facilities are shared by tens of thousands to millions of
users. Thus, in considering the data transmission rate with respect to LANs
and WANs on a non-communications carrier basis, we can say that LANs
provide a transmission capacity up to 10 Gbps while WANs are limited to a
low Mbps data rate.
Because LAN cabling is primarily within a building or over a small geographical area, it is relatively safe from natural phenomena, such as thunderstorms
and lightning. This safety enables transmission at a relatively high data rate,
introduction to networking concepts
11
resulting in a relatively low error rate. In comparison, because wide area
networks are based on the use of communications facilities that are much
farther apart and always exposed to the elements, they have a much higher
probability of being disturbed by changes in the weather, electronic emissions generated by equipment, or such unforeseen problems as construction
workers accidentally causing harm to a communications cable. Because of
these factors, the error rate on WANs is considerably higher than the rate
experienced on LANs. On most WANs you can expect to experience an error
rate between 1 in a million and 1 in 10 million (1 × 106 to 1 × 107 ) bits. In
comparison, the error rate on a typical LAN may exceed that range by one or
more orders of magnitude, resulting in an error rate from 1 in 10 million to 1
in 100 million bits.
Ownership
The construction of a wide area network requires the leasing of transmission facilities from one or more communications carriers. Although your
organization can elect to purchase or lease communications equipment, the
transmission facilities used to connect diverse geographical locations are
owned by the communications carrier. In comparison, an organization that
installs a local area network normally owns all of the components used to
form the network, including the cabling used to form the transmission path
between devices.
Regulation
Because wide area networks require transmission facilities that may cross
local, state, and national boundaries, they may be subject to a number of
governmental regulations at the local, state, and national levels. Most of those
regulations govern the services that communications carriers can provide customers and the rates (tariff ) they can charge for those services. In comparison,
regulations affecting local area networks are primarily in the areas of building codes. Such codes regulate the type of wiring that can be installed in a
building and whether the wiring must run in a conduit.
Data Routing and Topology
In a local area network data is routed along a path that defines the network.
That path is normally a bus, ring, tree, or star structure, and data always
flows on that structure. The topology of a wide area network can be much
more complex. In fact, many wide area networks resemble a mesh structure,
12
chapter one
including equipment to reroute data in the event of communications circuit
failure or excessive traffic between two locations. Thus, the data flow on a
wide area network can change, while the data flow on a local area network
primarily follows a single basic route.
Type of Information Carried
The last major difference between local and wide area networks is the type
of information carried by each network. Many wide area networks support
the simultaneous transmission of voice, data, and video information. In
comparison, most local area networks are currently limited to carrying data.
In addition, although all wide area networks can be expanded to transport
voice, data, and video, many local area networks are restricted by design to
the transportation of data. An exception to the preceding is asynchronous
transfer mode (ATM), which can provide both a local and wide area network
transmission capability. Asynchronous transfer mode was designed to support
voice, data, and video from end-to-end, enabling different types of data to be
transported from one LAN to another via an ATM WAN. Unfortunately, the
cost of ATM network adapters considerably exceeded the cost of other types
of LAN equipment used to include different types of Ethernet adapters. As the
base of Ethernet expanded, the cost associated with establishing an Ethernet
infrastructure decreased, widening the price gap between ATM and Ethernet,
making it difficult for asynchronous transfer mode to establish a viable market
for local area networking. Today the vast majority of ATM equipment is used
by communications carriers in the wide area network. Table 1.1 summarizes
the similarities and differences between local and wide area networks.
Utilization Benefits
In its simplest form, a local area network is a cable that provides an electronic
highway for the transportation of information to and from different devices
connected to the network. Because a LAN provides the capability to route
data between devices connected to a common network within a relatively
limited distance, numerous benefits can accrue to users of the network. These
can include the ability to share the use of peripheral devices, thus obtaining
common access to data files and programs, the ability to communicate with
other people on the LAN by electronic mail, and the ability to access the
larger processing capability of mainframes through common gateways that
link a local area network to larger computer systems. Here, the gateway can
be directly cabled to the mainframe if it resides at the same location, or it may
be connected remotely via a corporate wide area network.
introduction to networking concepts
TABLE 1.1
13
Comparing LANs and WANs
Characteristic
Local Area Network
Wide Area Network
Geographic area of
coverage
Localized to a building,
group of buildings, or
campus
Can span an area ranging in size
from a city to the globe
Data transmission
rate
Typically 4 Mbps to
16 Mbps, with
high-speed Ethernet and
fiber optic-based
networks operating at
100 Mbps and 1 and
10 Gbps
Normally operate at or below T1
and E1 transmission rates of
1.544 Mbps and 2.048 Mbps
Error rate
1 in 107 to 1 in 108
1 in 106 to 1 in 107
Ownership
Usually with the
implementor
Communications carrier retains
ownership of line facilities
Data routing
Normally follows fixed
route
Switching capability of network
allows dynamic alteration of
data flow
Topology
Usually limited to bus,
ring, tree, or star
Virtually unlimited design
capability
Type of information
carried
Primarily data
Voice, data, and video
commonly integrated
Peripheral sharing allows network users to access color laser printers, CDROM jukebox systems, and other devices that may be needed only a small
portion of the time a workstation is in operation. Thus, users of a LAN can
obtain access to resources that would probably be too expensive to justify for
each individual workstation user.
The ability to access data files and programs from multiple workstations
can substantially reduce the cost of software. In addition, shared access to
database information allows network users to obtain access to updated files
on a real-time basis.
One popular type of application program used on LANs enables users to
transfer messages electronically. Commonly referred to as electronic mail or
e-mail, this type of application program can be used to supplement and, in
many cases, eliminate the need for paper memoranda.
For organizations with a mainframe, a local area network gateway can
provide a common method of access. Without the use of a LAN gateway, each
14
chapter one
personal computer requiring access to a mainframe would require a separate
method of access. This might increase both the complexity and the cost of
providing access.
Perhaps the most popular evolving use of LANs is to provide a group of
computer users with economical access to the Internet. Instead of having
a large number of employees obtain individual modem dial-up or ISDN
dial access accounts with an Internet service provider (ISP), it is often more
economical to connect an organizational LAN to the Internet via a single
connection to an ISP. In addition, the connection to the Internet will usually
provide a higher transmission capability than obtainable on an individual
user basis. Later in this book we will turn our attention to methods that can
be used to connect organizational LANs to the Internet, as well as the use of
different products to protect organizational computer facilities from Internet
users that have no business accessing those facilities.
Technological Characteristics
Although a local area network is a limited distance transmission system,
the variety of options available for constructing such networks is anything
but limited. Many of the options available for the construction of local
area networks are based on the technological characteristics that govern
their operation. These characteristics include different topologies, signaling methods, transmission media, access methods used to transmit data
on the network, and the hardware and software required to make the network operate.
Topology
The topology of a local area network is the structure or geometric layout of
the cable used to connect stations on the network. Unlike conventional data
communications networks, which can be configured in a variety of ways with
the addition of hardware and software, most local area networks are designed
to operate based on the interconnection of stations that follow a specific
topology. The most common topologies used in LANs include the loop, bus,
ring, star, and tree, as illustrated in Figure 1.4.
Loop As previously mentioned in this chapter, IBM introduced a series of
transaction-processing terminals in 1974 that communicated through the use
of a common controller on a cable formed into a loop. This type of topology
is illustrated at the top of Figure 1.4.
introduction to networking concepts
15
Loop
Controller
Bus
Ring
Star
Tree
Figure 1.4 Local area network topology. The five most common geometric
layouts of LAN cabling form a loop, bus, ring, star, or tree structure.
Because the controller employed a poll-and-select access method, terminal
devices connected to the loop required a minimum of intelligence. Although
this reduced the cost of terminals connected to the loop, the controller lacked
the intelligence to distribute the data flow evenly among terminals. A lengthy
exchange between two terminal devices or between the controller and a
terminal would thus tend to bog down this type of network structure. A
second problem associated with this network structure was the centralized
16
chapter one
placement of network control in the controller. If the controller failed, the
entire network would become inoperative. Due to these problems, the use
of loop systems is restricted to several niche areas, and they are essentially
considered a derivative of a local area network.
Bus In a bus topology structure, a cable is usually laid out as one long
branch, onto which other branches are used to connect each station on
the network to the main data highway. Although this type of structure
permits any station on the network to talk to any other station, rules are
required for recovering from such situations as when two stations attempt
to communicate at the same time. Later in this chapter, we will examine
the relationships among the network topology, the method employed to
access the network, and the transmission medium employed in building
the network.
Ring In a ring topology, a single cable that forms the main data highway is
shaped into a ring. As with the bus topology, branches are used to connect
stations to one another via the ring. A ring topology can thus be considered
to be a looped bus. Typically, the access method employed in a ring topology
requires data to circulate around the ring, with a special set of rules governing
when each station connected to the network can transmit data.
Star The fourth major local area network topology is the star structure,
illustrated in the lower portion of Figure 1.4. In a star network, each station
on the network is connected to a network controller. Then, access from
any one station on the network to any other station can be accomplished
through the network controller. Here, the network controller functions like
a telephone switchboard, because access from one station to another station
on the network can occur only through the central device. In fact, you can
consider a telephone switchboard or PBX as representing a star-structured
LAN whose trunks provide connections to the wide area network telephone
infrastructure.
Tree A tree network structure represents a complex bus. In this topology,
the common point of communications at the top of the structure is known
as the headend. From the headend, feeder cables radiate outward to nodes,
which in turn provide workstations with access to the network. There may
also be a feeder cable route to additional nodes, from which workstations gain
access to the network. One common example of a tree structure topology is
the cable TV network many readers use on a daily basis. With the upgrade
introduction to networking concepts
17
of many cable TV systems to two-way amplifiers and the support of digital
transmission, the local cable TV infrastructure can be considered to represent
an evolving type of tree-structured local area network.
Mixed Topologies Some networks are a mixture of topologies. For example,
as previously discussed, a tree structure can be viewed as a series of interconnected buses. Another example of the mixture of topologies is a type
of Ethernet known as 10BASE-T, which is described in detail in Chapter 3.
That network can actually be considered a star-bus topology, because up to
16 or 24 devices known as stations are first connected to a common device
known as a hub, which in turn can be connected to other hubs to expand
the network.
Comparison of Topologies
Although there are close relationships among the topology of the network,
its transmission media, and the method used to access the network, we
can examine topology as a separate entity and make several generalized
observations. First, in a star network, the failure of the network controller will
render the entire network inoperative. This is because all data flow on the
network must pass through the network controller. On the positive side, the
star topology normally consists of telephone wires routed to a LAN switch. A
local area network that can use in-place twisted-pair telephone wires in this
way is simple to implement and usually very economical.
In a ring network, the failure of any node connected to the ring normally
inhibits data flow around the ring. Due to the fact that data travels in a circular
path on a ring network, any cable break has the same effect as the failure of the
network controller in a star-structured network. Because each network station
is connected to the next network station, it is usually easy to install the cable
for a ring network. In comparison, a star network may require cabling each
section to the network controller if existing telephone wires are not available,
and this can result in the installation of very long cable runs.
In a bus-structured network, data is normally transmitted from a single
station to all other stations located on the network, with a destination address
included within each transmitted data block. As part of the access protocol,
only the station with the destination address in the transmitted data block
will respond to the data. This transmission concept means that a break in
the bus affects only network stations on one side of the break that wish to
communicate with stations on the other side of the break. Thus, unless a
network station functioning as the primary network storage device becomes
inoperative, a failure in a bus-structured network is usually less serious than a
18
chapter one
failure in a ring network. However, some local area networks, such as TokenRing and FDDI, were designed to overcome the effect of certain types of cable
failures. Token-Ring networks include a backup path which, when manually
placed into operation, may be able to overcome the effect of a cable failure
between hubs (referred to as multistation access units or MAUs). In an FDDI
network, a second ring can be activated automatically as part of a self-healing
process to overcome the effect of a cable break.
A tree-structured network is similar to a star-structured network in that
all signals flow through a common point. In the tree-structured network,
the common signal point is the headend. Failure of the headend renders
the network inoperative. This network structure requires the transmission
of information over relatively long distances. For example, communications
between two stations located at opposite ends of the network would require
a signal to propagate twice the length of the longest network segment. Due
to the propagation delay associated with the transmission of any signal, the
use of a tree structure may result in a response time delay for transmissions
between the nodes that are most distant from the headend.
Although the first type of Ethernet network was based on a bus-structured
topology, other types of Ethernet networks incorporate the use of different
topologies. Today you can select bus-based, star-bus, or tree-structured Ethernet networks. Thus, you can select a particular type of Ethernet network to
meet a particular topology requirement.
Signaling Methods
The signaling method used by a local area network refers to both the way
data is encoded for transmission and the frequency spectrum of the media.
To a large degree, the signaling method is related to the use of the frequency
spectrum of the media.
Broadband versus Baseband
Two signaling methods used by LANs are broadband and baseband. In broadband signaling, the bandwidth of the transmission medium is subdivided by
frequency to form two or more subchannels, with each subchannel permitting
data transfer to occur independently of data transfer on another subchannel.
In baseband signaling, only one signal is transmitted on the medium at any
point in time.
Broadband is more complex than baseband, because it requires information
to be transmitted via the modulation of a carrier signal, thus requiring the use
of special types of modems, as discussed later in this chapter.
introduction to networking concepts
19
Single channel
Baseband
Frequency
Multiple channels
Broadband
A
B
N
Frequency
Figure 1.5 Baseband versus broadband signaling. In baseband signaling the
entire frequency bandwidth is used for one channel. In comparison, in broadband signaling the channel is subdivided by frequency into many subchannels.
Figure 1.5 illustrates the difference between baseband and broadband
signaling with respect to channel capacity. It should be noted that although a
twisted-pair wire system can be used to transmit both voice and data, the data
transmission is baseband, because only one channel is normally used for data.
In comparison, a broadband system on coaxial cable can be designed to carry
voice and several subchannels of data, as well as fax and video transmission.
Broadband Signaling A broadband local area network uses analog technology, in which high-frequency (HF) modems operating at or above 4 kHz place
carrier signals onto the transmission medium. The carrier signals are then
modified — a process known as modulation, which impresses information
onto the carrier. Other modems connected to a broadband LAN reconvert
the analog signal block into its original digital format — a process known as
demodulation.
Figure 1.6 illustrates the three primary methods of data encoding used by
broadband analog systems: amplitude, frequency, and phase modulation. The
most common modulation method used on broadband LANs is frequency shift
keying (FSK), in which two different frequencies are used, one to represent a
binary 1 and another frequency to represent a binary 0.
Another popular modulation method uses a combination of amplitude
and phase shift changes to represent pairs of bits. Referred to as amplitude
modulation phase shift keying (AM PSK), this method of analog signaling is
also known as duobinary signaling because each analog signal represents a
pair of digital bits.
Because it is not economically feasible to design amplifiers that boost signal
strength to operate in both directions, broadband LANs are unidirectional.
20
chapter one
Digital signal
1
0
1
1
0
Amplitude
modulation
Frequency
modulation
Phase
modulation
Figure 1.6 Modulation methods. Baseband signaling uses amplitude, frequency, or phase modulation, or a combination of modulation techniques to
represent digital information.
To provide a bidirectional information transfer capability, a broadband LAN
uses one channel for inbound traffic and another channel for outbound traffic.
These channels can be defined by differing frequencies or obtained by the use
of a dual cable.
Baseband Signaling In comparison to broadband local area networks,
which use analog signaling, baseband LANs use digital signaling to convey
information.
To understand the digital signaling methods used by most baseband LANs,
let us first review the method of digital signaling used by computers and
terminal devices. In that signaling method, a positive voltage is used to
represent a binary 1, while the absence of voltage (0 volts) is used to represent
a binary 0. If two successive 1 bits or 0 bits occur, two successive bit positions
then have a similar positive voltage level or a similar zero voltage level.
introduction to networking concepts
21
Because the signal goes from 0 to some positive voltage and does not return
to 0 between successive binary 1s, it is referred to as a unipolar nonreturn
to zero signal (NRZ). This signaling technique is illustrated at the top of
Figure 1.7.
Although unipolar NRZ signaling is easy to implement, its use for transmission has several disadvantages. One of the major disadvantages associated
with this signaling method involves determining where one bit ends and
another begins. For example, if you examine the top portion of Figure 1.7 you
will note that the bit sequences ‘‘00’’ and ‘‘111’’ remain at distinct voltage
levels. Thus, the ability to recognize that a sequence of two or more pulses of
the same value occurred requires synchronization between a transmitter and
receiver by the use of clocking circuitry, which can be relatively expensive.
To overcome the need for clocking, baseband LANs use Manchester or
Differential Manchester encoding. In Manchester encoding, a timing transition
always occurs in the middle of each bit, while an equal amount of positive
and negative voltage is used to represent each bit. This coding technique
provides a good timing signal for clock recovery from received data, due
to its timing transitions. In addition, because the Manchester code always
maintains an equal amount of positive and negative voltage, it prevents direct
current (DC) voltage buildup, enabling repeaters to be spaced farther apart
from one another.
Unipolar nonreturn to zero
1
0
0
1
0
1
1
1
Manchester coding
+V
Line
signal
−V
Figure 1.7 Unipolar nonreturn to zero signaling and Manchester coding. In
Manchester coding, a timing transition occurs in the middle of each bit and
the line code maintains an equal amount of positive and negative voltage.
22
chapter one
The lower portion of Figure 1.7 illustrates an example of Manchester coding.
Note that a low to high voltage transition represents a binary 1, while a high
to low voltage transition represents a binary 0. A second type of Manchester
coding is referred to as Differential Manchester encoding. Under Differential
Manchester encoding, the voltage transition is used only to provide clocking.
The encoding of a binary 0 or 1 is represented by the presence or absence
of a transition at the beginning of each bit period. Refer to Chapter 4 for
specific information concerning Manchester encoding on different types of
Ethernet networks.
Along with providing users with a choice of topologies, Ethernet also
provides a choice of signaling methods. Although most types of Ethernet
networks use baseband signaling, a broadband Ethernet is also available. In
fact, you can connect baseband- and broadband-based Ethernet networks to
satisfy different organizational requirements. Refer to Chapter 3 for specific
information concerning the signaling methods used by different Ethernet
networks and the hardware components used to construct and interconnect
such networks.
Transmission Medium
The transmission medium used in a local area network can range in scope from
twisted-pair wire, such as is used in conventional telephone lines, to coaxial
cable, fiber-optic cable, and electromagnetic waves such as those used by FM
radio and infrared. Each transmission medium has a number of advantages
and disadvantages. The primary differences between media are their cost and
ease of installation; the bandwidth of the cable, which may or may not permit
several transmission sessions to occur simultaneously; the maximum speed
of communications permitted; and the geographic scope of the network that
the medium supports.
Twisted-Pair Wire
In addition to being the most inexpensive medium available for LAN installations, twisted-pair wire is very easy to install. Since this wiring uses the same
RJ11 and RJ45 modular connectors as a telephone system, once a wire is cut
and a connector fastened, the attachment of the connector to network devices
is extremely simple. Normally, a screwdriver and perhaps a pocket knife are
the only tools required for the installation of twisted-pair wire. Anyone who
has hooked up a pair of speakers to a stereo set has the ability to install this
transmission medium.
introduction to networking concepts
23
Although inexpensive and easy to install, unshielded twisted-pair (UTP)
wire is very susceptible to noise generated by fluorescent light ballasts and
electrical machinery. In addition, a length of twisted-pair wire acts as an
antenna; however, the twists serve as a mechanism to partially counteract
this antenna effect. Unfortunately, due to the law of physics, the longer the
wire length, the greater the noise it gathers. At a certain length, the received
noise will obliterate the signal, which attenuates or decreases in strength as
it propagates along the length of the wire. This noise can affect the error rate
of data transmitted on the network, although lead-shielded twisted-pair (STP)
cable can be employed to provide the cable with a high degree of immunity
to the line noise and enable extended transmission distances. In Chapter 3
we will examine a building cabling standard and the various categories of
twisted-pair that can support different transmission rates which, in turn,
enable different types of Ethernet networks to be supported.
Because the bandwidth of twisted-pair cable is considerably less than
coaxial or fiber-optic cable, normally only one signal is transmitted on this
cable at a time. As previously explained, this signaling technique is known
as baseband signaling and should be compared with the broadband signaling
capability of coaxial and fiber-optic cable.
Although a twisted-pair wire system can be used to transmit both voice and
data, the data transmission is baseband because only one channel is normally
used for data. In comparison, a broadband system on coaxial or fiber-optic
cable can be designed to carry voice and several subchannels of data, as well
as fax and video transmission. Other constraints of unshielded twisted-pair
wire are the rate at which data can flow on the network and the distance it can
flow. Although data rates up to 1 gigabit per second (Gbps) can be achieved,
normally local area networks employing UTP wiring operate at a lower data
rate. In addition, UTP systems normally cover a limited distance, measured in
terms of several hundred to a few thousand feet, while coaxial and fiber-optic
cable–based systems may be limited in terms of miles. Extending transmission
distances over twisted-pair wire requires the periodic insertion of repeaters
into the cable. A repeater receives a digital signal and then regenerates it;
hence, it is also known as a data regenerator.
Coaxial Cable
At the center of a coaxial cable is a copper wire, which is covered by
an insulator known as a dielectric. An overlapping woven copper mesh
surrounds the dielectric, and the mesh, in turn, is covered by a protective
jacket consisting of polyethylene or aluminum. Figure 1.8 illustrates the
24
chapter one
Protective polyethylene jacket
Polyethylene
dielectric
Steel strength
member
Overlapped-seam
copper outer
conductor
Figure 1.8
Copper inner
conductor
Coaxial cable.
composition of a typical coaxial cable; however, it should be noted that
over 100 types of coaxial cable are currently marketed. The key differences
between such cables involve the number of conductors contained in the
cable, the dielectric employed, and the type of protective jacket and material
used to provide strength to the cable so it can be pulled through conduits
without breaking.
Two basic types of coaxial cable are used in local area networks. The type
of cable used is based on the transmission technique employed: baseband
or broadband signaling. Both cable types are much more expensive than
twisted-pair wire; however, the greater frequency bandwidth of coaxial cable
permits higher data rates for longer distances than you can obtain over
twisted-pair wire.
Normally, 50-ohm coaxial cable is used in baseband networks, while 75ohm cable is used in broadband networks. The latter coaxial is identical to
that used in cable television (CATV) applications, including the coaxial cable
used in a home. Data rates on baseband networks using coaxial cable range
from 50 to 100 Mbps. With broadband transmissions, data rates up to and
including 400 Mbps are obtainable.
A coaxial cable with a polyethylene jacket is normally used for baseband
signaling. Data is transmitted from stations on the network to the baseband cable in a digital format, and the connection from each station to the
cable is accomplished by the use of a simple coaxial T-connector. Because
data on a baseband network travels in a digital form, those signals can be
introduction to networking concepts
25
easily regenerated by the use of a device known as a line driver or data
regenerator. The line driver or data regenerator is a low-cost device that is
constructed to look for a pulse rise, and upon detecting the occurrence of
the rise, it will disregard the entire pulse and regenerate an entirely new
pulse. Thus, you can install low-cost line drivers into a baseband coaxial network to extend the distance over which transmission can occur on
the cable. Typically, a coaxial cable baseband system can cover an area
of several miles, and may contain hundreds to thousands of stations on
the network.
Obtaining independent subchannels defined by separate frequencies on
coaxial cable broadband transmission requires the translation of the digital signals from workstations into appropriate frequencies. This translation
process is accomplished by the use of radio-frequency (RF) modems, which
modulate the digital data into analog signals and then convert or demodulate
received analog signals into digital signals. Because signals are transmitted at
one frequency and received at a different frequency, a headend or frequency
translator is also required for broadband transmission on coaxial cable. This
device is also known as a remodulator, as it simply converts the signals from
one subchannel to another subchannel.
Fiber-Optic Cable
Fiber-optic cable is a transmission medium for light energy, and as such,
provides a very high bandwidth, permitting data rates ranging up to billions
of bits per second. The fiber-optic cable has a thin core of glass or plastic,
which is surrounded by a protective shield. Several of these shielded fibers
are bundled in a jacket, with a central member of aluminum or steel employed
for tensile strength.
Digital data represented by electrical energy must be converted into light
energy for transmission on a fiber-optic cable. This is normally accomplished by a low-power laser, or through the use of a light-emitting diode
and appropriate circuitry. At the receiver, light energy must be reconverted into electrical energy. Normally, a device known as a photo detector,
as well as appropriate circuitry to regenerate the digital pulses and an
amplifier, are used to convert the received light energy into its original
digital format.
Figure 1.9 provides an illustration of the cross section of a single-strand
fiber cable. The cladding that surrounds the core of the fiber can be considered
to represent a cylindrical mirror whose job is to ensure light stays in the core
as it flows along the fiber. The kevlar fibers add strength to the cable, while the
26
chapter one
Cladding
Core
Buffer
coating
Outer jacket
Kelvar fibers
Figure 1.9
Horizontal cross section of a single-strand fiber cable.
outer jacket, which is commonly colored orange, represents a polymer-based
shield that protects the cable from the elements.
There are two key factors that govern the manner by which light flows
through a fiber-optic cable. Those factors are the diameter of the core and the
light source. The first type of fiber-optic cable developed had a relatively large
diameter that ranged from 50 to 140 microns, where a micron is a millionth
of a meter. The original light source used to transmit information was a
light-emitting diode (LED). The coupling of an LED to a large-diameter optical
fiber results in photons flowing along multiple paths through the optical fiber,
resulting in the transmission referred to as multimode, which is also the same
name used to reference the type of optical fiber.
There are two types of multimode fiber, referred to as step-index and gradedindex. A step-index fiber has a core with a uniform refractive index, resulting
in the different components of a light signal in the form of modes or rays
flowing in a non-uniform manner through the optical cable. The top portion of
Figure 1.10 illustrates the flow of light through a step-index, multimode fiber.
In a graded-index multimode fiber, the refractive index is varied from the
center to the edge of the core to minimize modal dispersion. The middle
portion of Figure 1.10 illustrates the flow of light through a graded-index,
multimode fiber. This type of fiber minimizes model dispersion and supports
higher data rates than a step-index multimode optical fiber.
A third type of optical fiber has a relatively small core diameter, typically
between 7 and 12 microns (10−6 meters). This type of optical fiber permits
only one path for the flow of light due to the small diameter of the core. As
introduction to networking concepts
27
Step–index, multimode
LED
Graded–index, multimode
LED
Single mode
LASER
Figure 1.10
Light flow in multimode and single-mode optical fiber.
a result of the lack of modal dispersion, single mode supports a much higher
data rate than multimode fiber. Because of the small diameter of single-mode
fiber, lasers are used as the light source instead of LEDs.
Both the core thickness and the cladding of an optical fiber are measured
in microns. The three major core thicknesses used in optical fiber are 50, 62
and 100 microns. The associated claddings for those core diameters are 125,
125 and 140 microns, respectively. Table 1.2 summarizes the relationships
between core and cladding diameters for single-mode and multimode fiber.
In trade literature you will note what appears as a ratio in the form of x/y
when the metrics associated with an optical fiber are referenced. In actuality,
this is not a ratio but instead references the core and the cladding thickness
of an optical fiber in microns. Thus, 50/125 would, for example, reference an
optical fiber whose core diameter is 50 microns and whose cladding diameter
is 125 microns.
In addition to their high bandwidth, fiber-optic cables offer users several
additional advantages over conventional transmission media. Because data
travels in the form of light, it is immune to electrical interference and to the
building codes that often require expensive conduits for conventional cables.
28
chapter one
TABLE 1.2 Common Core and Cladding
Diameters of Optical Fiber in Microns
Mode
Single-mode
Multimode
Glass Core
Diameter
Glass
Cladding
2–8
20
50
125
62
125
100
140
Similarly, fiber-optic cable can be installed through areas where the flow of
electricity could be dangerous.
Because most fibers provide only a single, unidirectional transmission path,
a minimum of two cables is normally required to connect all transmitters to
all receivers on a network built using fiber-optic cable. Due to the higher cost
of fiber-optic cable, the dual cable requirement of fiber cables can make them
relatively expensive in comparison with other types of cable. In addition,
until recently it was very difficult to splice fiber-optic cable, and sophisticated
equipment and skilled installers were required to implement a fiber-opticbased network. Similarly, once this type of network was installed, until
recently it was difficult to modify the network. Recent advances in fiber
transmission through the use of wavelength division multiplexing enables two
or more transmission paths separated by the frequency of light to be carried
on a common optical cable. Although wavelength division multiplexing is
being used in the long-distance fiber backbones of communications carriers to
increase the transmission capacity of their infrastructure, the cost of electrooptical transmitter receivers to support this technology usually precludes its
use with local area networks.
Currently, the cost of the cable and the degree of difficulty of installation
and modification make the utilization of fiber-optic-based local area networks
impractical for many commercial applications. Today, the primary use of
fiber-optic cable is to extend the distance between stations on a network or
to connect two distant networks to one another. The device used to connect
a length of fiber-optic cable into the LAN or between LANs is a fiber-optic
repeater. The repeater converts the electrical energy of signals flowing on
the LAN into light energy for transmission on the fiber-optic cable. At the
end of the fiber-optic cable, a second repeater converts light energy back into
electrical energy. A key exception to the preceding involves Gigabit Ethernet,
introduction to networking concepts
29
whose data rate normally requires the use of optical fiber as a transport
medium. Although Gigabit Ethernet can operate on copper conductors, its
transmission distance is limited in comparison to when optical fiber is used.
With the declining cost of the fiber-optic cable and fiber-optic components,
and the continued introduction of improvements that simplify the installation
and modification of networks using this type of cable, the next few years may
bring a profound movement toward the utilization of fiber optics throughout
local area networks.
Ethernet can be categorized as a network for everyone, because of its
support of multiple topologies, signaling methods, and transmission media.
In addition to twisted-pair and coaxial cable-based Ethernet networks, you
can also extend the transmission distance of such networks by the use of
fiber-optic cable. Thus, the old adage from a presidential campaign, ‘‘a choice
not an echo,’’ is highly relevant to Ethernet networks.
Access Method
If the topology of a local area network can be compared with a data highway,
then the access method might be viewed as the set of rules that enable data from
one workstation to successfully reach its destination via the data highway.
Without such rules, it is quite possible for two messages sent by two different
workstations to collide, with the result that neither message reaches its destination. The three access methods primarily employed in local area networks are
carrier-sense multiple access/collision detection (CSMA/CD), carrier-sense
multiple access/collision avoidance (CSMA/CA), and token passing. Each
of these access methods is uniquely structured to address the previously
mentioned collision and data-destination problems. A fourth access method,
referred to as demand priority, is applicable to a high-speed Ethernet network
referred to as 100VG-AnyLAN. Through the use of a demand-priority access
method you can support either existing Ethernet CSMA/CD or token passing
networks at 100 Mbps.
Before discussing how access methods work, let us first examine the two
basic types of devices that can be attached to a local area network to gain an
appreciation for the work that the access method must accomplish.
Listeners and Talkers
We can categorize each device by its operating mode as being a listener or a
talker. Some devices, like printers, only receive data, and thus operate only
as listeners. Other devices, such as personal computers, can either transmit
or receive data and are capable of operating in both modes. In a baseband
30
chapter one
signaling environment where only one channel exists, or on an individual
channel on a broadband system, if several talkers wish to communicate at the
same time, a collision will occur. Therefore, a scheme must be employed to
define when each device can talk and, in the event of a collision, what must
be done to keep it from happening again.
For data to reach its destination correctly, each listener must have a unique
address, and its network equipment must be designed to respond to a message
on the net only when it recognizes its address. The primary goals in the
design of an access method are to minimize the potential for data collision,
to provide a mechanism for corrective action when data does collide, and to
ensure that an addressing scheme is employed to enable messages to reach
their destinations.
Carrier-Sense Multiple Access with Collision Detection (CSMA/CD)
CSMA/CD can be categorized as a listen-then-send access method. CSMA/CD
is one of the earliest developed access techniques and is the technique used
in Ethernet. Ethernet represents a local area network developed by Xerox
Corporation; its technology is now licensed to many companies, and was
standardized by the Institute of Electrical and Electronics Engineers (IEEE)
under its 802.3 standard.
Under the CSMA/CD concept, when a station has data to send, it first listens
to determine whether any other station on the network is talking. The fact that
the channel is idle is determined in one of two ways, based on whether the
network is broadband or baseband.
In a broadband network, the fact that a channel is idle is determined by
carrier sensing, or noting the absence of a carrier tone on the cable.
Ethernet, like other baseband systems, uses one channel for data transmission and does not employ a carrier. Instead, Ethernet encodes data using a
Manchester code, in which a timing transition always occurs in the middle
of each bit, as previously illustrated in Figure 1.7. Although Ethernet does
not transmit data via a carrier, the continuous transitions of the Manchester
code can be considered as equivalent to a carrier signal. Carrier sensing
on a baseband network is thus performed by monitoring the line for activity. Manchester coding is used by Ethernet at data rates up to 10 Mbps.
At operating rates above 10 Mbps, different encoding techniques are used,
which provide a higher level of efficiency than Manchester’s 50-percent level
of efficiency.
In a CSMA/CD network, if the channel is busy, the station will wait until it
becomes idle before transmitting data. Because it is possible for two stations to
listen at the same time and discover an idle channel, it is also possible that the
introduction to networking concepts
31
two stations could then transmit at the same time. When this situation arises,
a collision will occur. Upon sensing that a collision has occurred, a delay
scheme will be employed to prevent a repetition of the collision. Typically,
each station will use either a randomly generated or a predefined time-out
period before attempting to retransmit the message that collided. Because this
access method requires hardware capable of detecting the occurrence of a
collision, additional circuitry required to perform collision detection adds to
the cost of such hardware.
Figure 1.11 illustrates a CSMA/CD bus-based local area network. Each
workstation is attached to the transmission medium, such as coaxial cable, by
a device known as a bus interface unit (BIU). To obtain an overview of the
operation of a CSMA/CD network, assume that station A is currently using
the channel and stations C and D wish to transmit. The BIUs connecting
stations C and D to the network listen to the channel and note that it is
busy. Once station A completes its transmission, stations C and D attempt
to gain access to the channel. Because station A’s signal takes longer to
propagate down the cable to station D than to station C, C’s BIU notices that
the channel is free slightly before station D’s BIU. However, as station C gets
ready to transmit, station D now assumes that the channel is free. Within an
infinitesimal period of time, C starts transmission, followed by D, resulting in
a collision. Here, the collision is a function of the propagation delay of the
signal and the distance between two competing stations. CSMA/CD networks
therefore work better as the main cable length decreases.
The CSMA/CD access technique is best suited for networks with intermittent
transmission, because an increase in traffic volume causes a corresponding
increase in the probability of the cable being occupied when a station wishes
A
Bus-based
channel
B
BIU
BIU
BIU
C
BIU
D
BIU = Bus interface unit
Figure 1.11 CSMA/CD network operation. In a CSMA/CD network, as the
distance between workstations increases, the resulting increase in propagation
delay time increases the probability of collisions.
32
chapter one
to talk. In addition, as traffic volume builds under CSMA/CD, throughput may
decline, because there will be longer waits to gain access to the network, as
well as additional time-outs required to resolve collisions that occur. In spite
of the lower level of throughput, as we will note later in this book CSMA/CA
is used as the access method supported by many types of wireless LANs
including several types of wireless Ethernet LANs.
Carrier-Sense Multiple Access with Collision Avoidance (CSMA/CA)
CSMA/CA represents a modified version of the CSMA/CD access technique.
Under the CSMA/CA access technique, each of the hardware devices attached
to the talkers on the network estimates when a collision is likely to occur
and avoids transmission during those times. Because this technique eliminates the requirement for collision-detection hardware, the cost of hardware
to implement this access technique is usually less than CSMA/CD hardware. Unfortunately, time delays associated with collision avoidance usually
result in a lower level of throughput than that obtainable with CSMA/CDbased networks, and this limitation has made CSMA/CA-based networks
less popular.
Token Passing
In a token-passing access method, each time the network is turned on, a token
is generated. The token, consisting of a unique bit pattern, travels the length
of the network, either around a ring or along the length of a bus. When a
station on the network has data to transmit, it must first seize a free token.
On a Token-Ring network, the token is then transformed to indicate that it
is in use. Information is added to produce a frame, which represents data
being transmitted from one station to another. During the time the token is in
use, other stations on the network remain idle, eliminating the possibility of
collisions. Once the transmission is completed, the token is converted back
into its original form by the station that transmitted the frame, and becomes
available for use by the next station on the network.
Figure 1.12 illustrates the general operation of a token-passing Token-Ring
network using a ring topology. Because a station on the network can only
transmit when it has a free token, token passing eliminates the requirement
for collision detection hardware. Due to the dependence of the network on
the token, the loss of a station can bring the entire network down. To avoid
this, the design characteristics of Token-Ring networks include circuitry that
automatically removes a failed or failing station from the network, as well as
other self-healing features. This additional capability is costly: a Token-Ring
introduction to networking concepts
33
B
T
A
C
Workstation with data
to transmit looks for
free token
D
B
A
C
Workstation changes
token into a frame,
adding data to the
frame and its
destination address
F
D
B
A
Destination station
copies data addressed
to it
C
F
D
B
A
C
Frame flows back to
originator, which removes
the frame from the
network and generates a
new token
T
D
Figure 1.12
Token-Ring operation.
adapter card is typically priced at two to three times the cost of an Ethernet
adapter card.
Due to the variety of transmission media, network structures, and access
methods, there is no one best network for all users. Table 1.3 provides a
generalized comparison of the advantages and disadvantages of the technical
characteristics of local area networks, using the transmission medium as a
frame of reference.
34
chapter one
TABLE 1.3
Technical Characteristics of LANs
Transmission Medium
Characteristic
Twisted-pair
wire
Baseband
coaxial cable
Baseband
coaxial cable
Fiber-optic
cable
Topology
Bus, star, or
ring
Bus or ring
Bus or ring
Bus, star, or ring
Channels
Single
channel
Single
channel
Multichannel
Single,
multichannel
Data rate
Normally 4 to
16 Mbps; up
to 1000 Mbps
obtainable
Normally 2 to
10 Mbps; up
to 100 Mbps
obtainable
Up to 400 Mbps
Up to 40 Gbps
Maximum
nodes on net
Usually <255
Usually
<1024
Several
thousand
Several
thousand
Geographical
coverage
Thousands of
feet
Miles
Tens of miles
Tens of miles
Major
advantages
Low cost;
may be able
to use
existing
wiring
Low cost;
simple to
install
Supports voice,
data, video
applications
simultaneously
Supports voice,
data, video
applications
simultaneously
Major
disadvantages
Limited
bandwidth;
requires
conduits; low
immunity to
noise
Low
immunity to
noise
High cost;
difficult to
install; requires
RF modems and
headend
Cable cost;
difficult to
splice
1.3 Why Ethernet
After reviewing the basic characteristics of various local area networks, you
will note that they present a considerable range of technological characteristics. You can select a network based on its topology, access method, signaling
method, and/or support of a particular type of transmission medium. Although
there is no one best network for all users, Ethernet represents a diverse mixture of technological characteristics. As such, it provides many more potential
networking solutions to user communications requirements than other types
of local area networks.
introduction to networking concepts
35
In addition to supporting a broad range of technological characteristics,
Ethernet is a declining-cost network. Due to economies of manufacture, as
well as the recent support of twisted-pair wiring, Ethernet networks can be
established for a fraction of the cost of other types of local area networks.
In fact, by late 2002, adapter cards supporting 10 Mbps Ethernet that were
designed for use in IBM PCs and compatible personal computers could be
obtained for under $20, while 10/100 Mbps adapter cards could be obtained
for under $50. In addition, several notebook and laptop computer vendors,
as well as vendors of many business desktop systems, were incorporating Ethernet chip sets into their computers, enabling computer users who
want to transfer files to other computers and printers connected to the network to simply plug their computer into a telephone jack upon returning to
the office.
Although the previously mentioned reasons are significant, the scalability of Ethernet and the product development effort of industry provide two
additional reasons to use Ethernet. Concerning scalability, you can obtain
interoperable Ethernet equipment that enables you to operate a network
originally at 10 Mbps and have the option of using a variety of vendor
products to upgrade your network to 100 Mbps or even 1 or 10 Gbps.
Concerning product development, over the past decade a large number of
products reached the market that enable Ethernet users to boost the performance and management of their networks. Perhaps the most versatile
product is actually a series of Ethernet switches that enable multiple crossconnections between workstations and servers to occur simultaneously, as
well as enable the transmission on multiple ports to be aggregated, a technique referred to as a fat pipe. Over the past five years the capability of LAN
switches has been significantly enhanced. Today many LAN switch manufacturers support switching not only directly at the LAN frame layer but, in
addition, at higher layers in the International Standards Organization (ISO)
Open System Interconnection (OSI) Reference Model, which is described in
Chapter 2.
Although your organization may not currently require the ability to transmit
data at 100 Mbps or 1 or 10 Gbps nor require the ability to aggregate multiple
paths at those data rates, the capability is there if you should need it in
the future. For other organizations such as Internet Service Providers that
use a local area network as a gateway for subscribers to access the Internet,
the ability to upgrade to 1 or 10 Gbps can be a significant benefit. Due to
this capability Ethernet is truly a ‘‘people’s network,’’ able to satisfy the
networking requirements of the small business, large corporation, university,
and government.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter two
Networking Standards
Standards can be viewed as the ‘‘glue’’ that binds hardware and software from
different vendors so they can operate together. The importance of standards
and the work of standards organizations have proved essential for the growth
of both local and worldwide communications. In the United States and many
other countries, national standards organizations have defined physical and
operational characteristics that enable vendors to manufacture equipment
compatible with line facilities provided by communications carriers, as well
as equipment produced by other vendors. At the international level, standards
organizations have promulgated several series of communications-related recommendations. These recommendations, while not mandatory, have become
highly influential on a worldwide basis for the development of equipment
and facilities, and have been adopted by hundreds of public companies and
communications carriers.
In addition to national and international standards, a series of de facto
standards has evolved through the licensing of technology among companies.
These de facto standards have facilitated, for example, the development of
communications software for use on personal computers. Today, communications software can control modems manufactured by hundreds of vendors,
because most modems are now constructed to respond to a core set of uniform
control codes.
2.1 Standards Organizations
In this chapter, we will first focus our attention on two national and two
international standards organizations. The national standards organizations
we will briefly discuss in this section are the American National Standards
Institute (ANSI) and the Institute of Electrical and Electronics Engineers
(IEEE). The work of both organizations has been a guiding force in the rapid
expansion in the use of local area networks due to a series of standards they
37
38
chapter two
have developed. Due to the importance of the work of the IEEE in developing
LAN standards, we will examine those standards as a separate entity in the
next section in this chapter. In the international arena, we will discuss the
role of the International Telecommunications Union (ITU), formerly known
as the Consultative Committee for International Telephone and Telegraph
(CCITT), and the International Standards Organization (ISO), both of which
have developed numerous standards to facilitate the operation of local and
wide area networks.
Because of the importance of the ISO’s Open Systems Interconnection (OSI)
Reference Model and the IEEE’s 802 Committee lower layer standards, we
will examine each as a separate entity in this chapter. Because a series of
Internet standards define the manner by which the TCP/IP protocol suite
can transport data between LANs and WANs, we will also discuss what are
referred to as Requests For Comments (RFCs). Because we must understand
the OSI Reference Model before examining the effect of the efforts of the
IEEE and ANSI upon the lower layers of that model and the role of RFCs,
we will look at the OSI Reference Model before examining the role of other
standards.
National Standards Organizations
The two national standards organizations we will briefly discuss are the American National Standards Institute and the Institute of Electrical and Electronics
Engineers. In the area of local area networking standards, both ANSI and the
IEEE work in conjunction with the ISO to standardize LAN technology.
The ISO delegated the standardization of local area networking technology to ANSI. The American National Standards Institute, in turn, delegated
lower-speed LAN standards — initially defined as operating rates at and below
50 Mbps — to the IEEE. This resulted in ANSI’s developing standards for the
100-Mbps fiber distributed data interface (FDDI), while the IEEE developed
standards for Ethernet, Token-Ring, and other LANs. Because the IEEE developed standards for 10-Mbps Ethernet, that organization was tasked with the
responsibility for modifications to that LAN technology. This resulted in the
IEEE becoming responsible for the standardization of high-speed Ethernet
to include isoENET, 100BASE-T, and 100VG-AnyLAN, the latter two representing 100-Mbps LAN operating rates. Another series of IEEE standards
beginning with the prefix of 1000 defines the operation of Gigabit Ethernet
over different types of copper and optical fiber media. In addition, when this
book revision occurred the IEEE was in the process of finalizing a standard for
10 Gbps Ethernet.
networking standards
39
Once the IEEE develops and approves a standard, that standard is sent to
ANSI for review. If ANSI approves the standard, it is then sent to the ISO.
Then, the ISO solicits comments from all member countries to ensure that
the standard will work at the international level, resulting in an IEEE- or
ANSI-developed standard becoming an ISO standard.
ANSI
The principal standards-forming body in the United States is the American
National Standards Institute (ANSI). Located in New York City, this nonprofit,
nongovernmental organization was founded in 1918 and functions as the
representative of the United States to the ISO.
American National Standards Institute standards are developed through
the work of its approximately 300 Standards Committees, and from the
efforts of associated groups such as the Electronic Industry Association (EIA).
Recognizing the importance of the computer industry, ANSI established its
X3 Standards Committee in 1960. That committee consists of 25 technical
committees, each assigned to develop standards for a specific technical area.
One of those technical committees is the X3S3 committee, more formally
known as the Data Communications Technical Committee. This committee
was responsible for the ANSI X3T9.5 standard that governs FDDI operations,
and that is now recognized as the ISO 9314 standard.
IEEE
The Institute of Electrical and Electronics Engineers (IEEE) is a U.S.-based
engineering society that is very active in the development of data communications standards. In fact, the most prominent developer of local area networking
standards is the IEEE, whose subcommittee 802 began its work in 1980 before
they had even established a viable market for the technology.
The IEEE Project 802 efforts are concentrated on the physical interface
between network devices and the procedures and functions required to
establish, maintain, and release connections among them. These procedures
include defining data formats, error control procedures, and other control
activities governing the flow of information. This focus of the IEEE actually
represents the lowest two layers of the ISO model, physical and link, which
are discussed later in this chapter.
International Standards Organizations
Two important international standards organizations are the International
Telecommunications Union (ITU), formerly known as the Consultative
chapter two
40
Committee for International Telephone and Telegraph (CCITT), and the
International Standards Organization (ISO). The ITU can be considered a
governmental body, because it functions under the auspices of an agency of
the United Nations. Although the ISO is a nongovernmental agency, its work
in the field of data communications is well recognized.
ITU
The International Telecommunications Union (ITU) is a specialized agency of
the United Nations headquartered in Geneva, Switzerland. The ITU has direct
responsibility for developing data communications standards and consists
of 15 study groups, each with a specific area of responsibility. Although
the CCITT was renamed as the ITU in 1994, it periodically continues to be
recognized by its former mnemonic. Thus, the remainder of this book will refer
to this standards organization by its new set of commonly recognized initials.
The work of the ITU is performed on a four-year cycle known as a study
period. At the conclusion of each study period, a plenary session occurs.
During the plenary session, the work of the ITU during the previous four
years is reviewed, proposed recommendations are considered for adoption,
and items to be investigated during the next four-year cycle are considered.
The ITU’s eleventh plenary session met in 1996 and its twelfth session
occurred during 2000. Although approval of recommended standards is not
intended to be mandatory, ITU recommendations have the effect of law in
some Western European countries, and many of its recommendations have
been adopted by communications carriers and vendors in the United States.
Perhaps the best-known set of ITU recommendations is its V-series, which
describes the operation of many different modem features — for example, data
compression and transmission error detection and correction.
ISO
The International Standards Organization (ISO) is a nongovernmental entity
that has consultative status within the UN Economic and Social Council. The
goal of the ISO is to ‘‘promote the development of standards in the world with
a view to facilitating international exchange of goods and services.’’
The membership of the ISO consists of the national standards organizations of most countries. There are approximately 100 countries currently
participating in its work.
Perhaps the most notable achievement of the ISO in the field of communications is its development of the seven-layer Open Systems Interconnection
(OSI) Reference Model.
networking standards
41
2.2 The ISO Reference Model
The International Standards Organization (ISO) established a framework for
standardizing communications systems called the Open Systems Interconnection (OSI) Reference Model. The OSI architecture defines the communications
process as a set of seven layers, with specific functions isolated and associated
with each layer. Each layer, as illustrated in Figure 2.1, covers lower layer
processes, effectively isolating them from higher layer functions. In this way,
each layer performs a set of functions necessary to provide a set of services to
the layer above it.
Layer isolation permits the characteristics of a given layer to change without
impacting the remainder of the model, provided that the supporting services
remain the same. One major advantage of this layered approach is that users
can mix and match OSI-conforming communications products, and thus tailor
their communications systems to satisfy particular networking requirements.
The OSI Reference Model, while not completely viable with many current
network architectures, offers the potential to connect networks and networking
devices together to form integrated networks, while using equipment from
different vendors. This interconnectivity potential will be of substantial benefit
to both users and vendors. For users, interconnectivity will remove the
shackles that in many instances tie them to a particular vendor. For vendors,
the ability to easily interconnect their products will provide them with access
to a larger market. The importance of the OSI model is such that it was adopted
by the ITU as Recommendation X.200.
Layered Architecture
As previously discussed, the OSI Reference Model is based on the establishment of a layered, or partitioned, architecture. This partitioning effort is
Application
Layer 7
Presentation
Layer 6
Session
Layer 5
Transport
Layer 4
Network
Layer 3
Data Link
Layer 2
Physical
Layer 1
Figure 2.1
ISO Reference Model.
42
chapter two
derived from the scientific process, in which complex problems are subdivided
into several simpler tasks.
As a result of the application of a partitioning approach to communications
network architecture, the communications process was subdivided into seven
distinct partitions, called layers. Each layer consists of a set of functions
designed to provide a defined series of services. For example, the functions
associated with the physical connection of equipment to a network are referred
to as the physical layer.
With the exception of layers 1 and 7, each layer is bounded by the layers
above and below it. Layer 1, the physical layer, is bound below by the
interconnecting medium over which transmission flows, while layer 7 is the
upper layer and has no upper boundary. Within each layer is a group of
functions that provide a set of defined services to the layer above it, resulting
in layer n using the services of layer n − 1. Thus, the design of a layered
architecture enables the characteristics of a particular layer to change without
affecting the rest of the system, assuming that the services provided by the
layer do not change.
OSI Layers
The best way to gain an understanding of the OSI layers is to examine
a network structure that illustrates the components of a typical wide area
network. Figure 2.2 illustrates a network structure that is typical only in the
sense that it will be used for a discussion of the components upon which
networks are constructed.
The circles in Figure 2.2 represent nodes, which are points where data
enters or exits a network or is switched between two networks connected by
A
B
D
C
Legend:
E
= Node
Lines represent paths
Figure 2.2
Generic network structure.
networking standards
43
one or more paths. Nodes are connected to other nodes via communications
cables or circuits and can be established on any type of communications
medium, such as cable, microwave, or radio.
From a physical perspective, a node can be based on any of several types of
computers, including a personal computer, minicomputer, mainframe computer, or specialized computer, such as a front-end processor. Connections to
network nodes into a wide area network can occur via terminal devices, such
as PCs and fixed logic devices, directly connected to computers, terminals
connected to a node via one or more intermediate communications devices,
or paths linking one network to another network. In fact, a workstation on
an Ethernet local area network that provides access into a wide area network
can be considered a network node. In this situation, the workstation can be a
bridge, router, or gateway, and provides a connectivity mechanism between
other stations on the Ethernet local area network and the wide area network.
The routes between two nodes — such as C-E-A, C-D-A, C-A, and C-B-A, all
of which can be used to route data between nodes A and C — are information
paths. Due to the variability in the flow of information through a wide area
network, the shortest path between nodes may not be available for use,
or may be inefficient in comparison to other possible paths. A temporary
connection between two nodes that is based on such parameters as current
network activity is known as a logical connection. This logical connection
represents the use of physical facilities, including paths and temporary nodeswitching capability.
The major functions of each of the seven OSI layers are described in the
following seven paragraphs.
Layer 1 — The Physical Layer
At the lowest or most basic level, the physical layer (level 1) is a set of rules
that specifies the electrical and physical connection between devices. This
level specifies the cable connections and the electrical rules necessary to
transfer data between devices. Typically, the physical link corresponds to
previously established interface standards, such as the RS-232/V.24 interface.
This interface governs the attachment of data terminal equipment, such as the
serial port of personal computers, to data communications equipment, such
as modems.
Layer 2 — The Data Link Layer
The next layer, which is known as the data link layer (level 2), denotes
how a device gains access to the medium specified in the physical layer.
44
chapter two
It also defines data formats, including the framing of data within transmitted
messages, error control procedures, and other link control activities. Because
it defines data formats, including procedures to correct transmission errors,
this layer becomes responsible for the reliable delivery of information. An
example of a data link control protocol that can reside at this layer is the ITU’s
High-Level Data Link Control (HDLC).
Because the development of OSI layers was originally targeted toward wide
area networking, its applicability to local area networks required a degree of
modification. Under the IEEE 802 standards, the data link layer was initially
divided into two sublayers: logical link control (LLC) and media access control
(MAC). The LLC layer is responsible for generating and interpreting commands
that control the flow of data and perform recovery operations in the event of
errors. In comparison, the MAC layer is responsible for providing access to
the local area network, which enables a station on the network to transmit
information.
With the development of high-speed local area networks designed to operate
on a variety of different types of media, an additional degree of OSI layer
subdivision was required. First, the data link layer required the addition
of a reconciliation layer (RL) to reconcile a medium-independent interface
(MII) signal added to a version of high-speed Ethernet, commonly referred
to as Fast Ethernet. Next, the physical layer used for Fast Ethernet required
a subdivision into three sublayers. One sublayer, known as the physical
coding sublayer (PCS) performs data encoding. A physical medium attachment
sublayer (PMA) maps messages from the physical coding sublayer to the
transmission media, while a medium-dependent interface (MDI) specifies the
connector for the media used. Similarly, Gigabit Ethernet implements a gigabit
media-independent interface (GMII), which enables different encoding and
decoding methods to be supported that are used with different types of media.
Later in this chapter, we will examine the IEEE 802 subdivision of the data
link and physical layers, as well as the operation of each resulting sublayer.
Layer 3 — The Network Layer
The network layer (level 3) is responsible for arranging a logical connection
between the source and destination nodes on the network. This responsibility
includes the selection and management of a route for the flow of information
between source and destination, based on the available data paths in the
network. Services provided by this layer are associated with the movement
of data packets through a network, including addressing, routing, switching,
sequencing, and flow control procedures. In a complex network, the source
and destination may not be directly connected by a single path, but instead
networking standards
45
require a path that consists of many subpaths. Thus, routing data through the
network onto the correct paths is an important feature of this layer.
Several protocols have been defined for layer 3, including the ITU X.25
packet switching protocol and the ITU X.75 gateway protocol. X.25 governs
the flow of information through a packet network, while X.75 governs the flow
of information between packet networks. Other popular examples of layer 3
protocols include the Internet Protocol (IP) and Novell’s Internet Packet
Exchange (IPX), both of which represent layers in their respective protocol
suites that were defined before the ISO Reference Model was developed. In
an Ethernet environment the transport unit is a frame. As we will note later
in this book when we examine Ethernet frame formats in Chapter 4, the frame
on a local area network is used as the transport facility to deliver such layer 3
protocols as IP and IPX, which in turn represent the vehicles for delivering
higher-layer protocols in the IP and IPX protocol suites.
Layer 4 — The Transport Layer
The transport layer (level 4) is responsible for guaranteeing that the transfer
of information occurs correctly after a route has been established through the
network by the network level protocol. Thus, the primary function of this layer
is to control the communications session between network nodes once a path
has been established by the network control layer. Error control, sequence
checking, and other end-to-end data reliability factors are the primary concern
of this layer, and they enable the transport layer to provide a reliable endto-end data transfer capability. Examples of popular transport layer protocols
include the Transmission Control Protocol (TCP) and the User Datagram
Protocol (UDP), both of which are part of the TCP/IP protocol suite, and
Novell’s Sequence Packet Exchange (SPX).
Layer 5 — The Session Layer
The session layer (level 5) provides a set of rules for establishing and terminating data streams between nodes in a network. The services that this session
layer can provide include establishing and terminating node connections,
message flow control, dialogue control, and end-to-end data control.
Layer 6 — The Presentation Layer
The presentation layer (level 6) services are concerned with data transformation, formatting, and syntax. One of the primary functions performed by the
presentation layer is the conversion of transmitted data into a display format
46
chapter two
appropriate for a receiving device. This can include any necessary conversion
between ASCII and EBCDIC codes. Data encryption/decryption and data compression/decompression are additional examples of the data transformation
that can be handled by this layer.
Layer 7 — The Application Layer
Finally, the application layer (level 7) acts as a window through which the
application gains access to all of the services provided by the model. Examples
of functions performed at this level include file transfers, resource sharing,
and database access. While the first four layers are fairly well defined, the
top three layers may vary considerably, depending on the network protocol
used. For example, the TCP/IP protocol, which predates the OSI Reference
Model, groups layer 5 through layer 7 functions into a single application
layer. In Chapter 5 when we examine Internet connectivity, we will also
examine the relationship of the TCP/IP protocol stack to the seven-layer OSI
Reference Model.
Figure 2.3 illustrates the OSI model in schematic format, showing the
various levels of the model with respect to a terminal device, such as a personal
computer accessing an application on a host computer system. Although
Figure 2.3 shows communications occurring via a modem connection on
a wide area network, the OSI model schematic is also applicable to local
area networks. Thus, the terminal shown in the figure could be replaced
by a workstation on an Ethernet network while the front-end processor
(FEP) would, via a connection to that network, become a participant on
that network.
Data Flow
As data flows within an ISO network, each layer appends appropriate heading
information to frames of information flowing within the network, while
removing the heading information added by a lower layer. In this manner,
layer n interacts with layer n − 1 as data flows through an ISO network.
Figure 2.4 illustrates the appending and removal of frame header information as data flows through a network constructed according to the ISO
Reference Model. Because each higher level removes the header appended by
a lower level, the frame traversing the network arrives in its original form at
its destination.
As you will surmise from the previous illustrations, the ISO Reference
Model is designed to simplify the construction of data networks. This simplification is due to the potential standardization of methods and procedures
networking standards
Level 7
Level 6
Network
Modem
Modem
FEP
Control
program
Access
method
Application
Level 5
Terminal
Level 1
Level 2
Level 3
Level 4
Figure 2.3
Layer
OSI model schematic.
Outgoing
frame
Received
frame
AH Data
7 Application
6 Presentation
PH AH Data
5
Session
SH PH AH Data
4
Transport
TH SH PH AH Data
3
Network
2
Data link
1
Physical
NH TH SH PH AH Data
DH NH TH SH PH AH Data
Data bits
Data AH
Data AH PH
Data AH PH SH
Data AH PH SH TH
Data AH PH SH TH NH
Data AH PH SH TH NH DH
Data bits
Application
Presentation
Session
Transport
Network
Data link
Physical
Legend:
DH, NH, TH, SH, PH and AH are appropriate headers Data Link, Network header,
Transport header, Session header, Presentation header and Application header
added to data as the data flows through an ISO Reference model network
Figure 2.4
Appending and removal of frame header information.
47
48
chapter two
to append appropriate heading information to frames flowing through a
network, permitting data to be routed to its appropriate destination following
a uniform procedure.
2.3 IEEE 802 Standards
The Institute of Electrical and Electronics Engineers (IEEE) Project 802 was
formed at the beginning of the 1980s to develop standards for emerging
technologies. The IEEE fostered the development of local area networking
equipment from different vendors that can work together. In addition, IEEE
LAN standards provided a common design goal for vendors to access a
relatively larger market than if proprietary equipment were developed. This,
in turn, enabled economies of scale to lower the cost of products developed
for larger markets.
The actual committee tasked with the IEEE Project 802 is referred to as the
IEEE Local and Metropolitan Area Network (LAN/WAN) Standards Committee. Its basic charter is to create, maintain, and encourage the use of IEEE/ANSI
and equivalent ISO standards primarily within layers 1 and 2 of the ISO Reference Model. The committee conducts a plenary meeting three times a year
and currently has 13 major working groups, each of which may have several
meetings per year at locations throughout the world.
802 Committees
Table 2.1 lists the IEEE 802 committees involved in local and metropolitan area
networks. In examining the lists of committees in Table 2.1, it is apparent that
the IEEE early on noted that a number of different systems would be required
to satisfy the requirements of a diverse end-user population. Accordingly, the
IEEE adopted the CSMA/CD, Token-Bus, and Token-Ring as standards 802.3,
802.4, and 802.5, respectively.
The IEEE Committee 802 published draft standards for CSMA/CD and
Token-Bus local area networks in 1982. Standard 802.3, which describes a
baseband CSMA/CD network similar to Ethernet, was published in 1983. Since
then, several addenda to the 802.3 standard have been adopted to govern the
operation of CSMA/CD on different types of media. Those addenda include
10BASE-2, which defines a 10-Mbps baseband network operating on thin
coaxial cable; 1BASE-5, which defines a 1-Mbps baseband network operating
on twisted-pair; 10BASE-T, which defines a 10-Mbps baseband network operating on twisted-pair; and 10BROAD-36, which defines a broadband 10-Mbps
network that operates on thick coaxial cable.
networking standards
TABLE 2.1
49
IEEE Series 802 Committees/Standards
802
Overview — Architecture
802.1
Bridging — Management
802.2
Logical Link Control
802.3
CSMA/CD Access Method
802.4
Token-Passing Bus Access Method
802.5
Token-Passing Ring Access Method
802.6
Metropolitan Area Networks (DQDB Access Method)
802.7
Broadband LAN
802.8
Fiber Optic Technical Advisory Group
802.9
Integrated Voice and Data Networks
802.10
Network Security
802.11
Wireless LANs
802.12
Demand Priority Access
The IEEE 802.3 committee includes a large number of projects that resulted
in the refinement and expansion of the CSMA/CD protocol. Some of those
projects were completed several years ago, while others are currently ongoing.
Table 2.2 lists nine examples of IEEE 802.3 CSMA/CD projects. A Fast Ethernet, which is denoted as 802.3µ in Table 2.2, is an addendum to the 802.3
standard, which was finalized in 1995. 802.3z represents the 802 committee
project that was responsible for developing the Gigabit Ethernet standard.
The next major standard published by the IEEE was 802.4, which describes
a token-passing bus–oriented network for both baseband and broadband
transmission. This standard is similar to the Manufacturing Automation
Protocol (MAP) standard developed by General Motors.
The third major LAN standard published by the IEEE was based on IBM’s
specifications for its Token-Ring network. Known as the 802.5 standard,
it defines the operation of token-ring networks on shielded twisted-pair
cable at data rates of 1 and 4 Mbps. That standard was later modified to
acknowledge three IBM enhancements to Token-Ring network operations.
These enhancements include the 16-Mbps operating rate, the ability to release
a token early on a 16-Mbps network, and a bridge routing protocol known as
source routing.
50
chapter two
TABLE 2.2
IEEE 802.3 CSMA/CD Projects
803.2aa
Maintenance Revision #5 (100Base-T)
802.3ab
1000Base-T
802.3ad
Link Aggregation
802.3c
vLAN tag
802.3ae
10 Gbps Ethernet
802.3ag
Maintenance Revision #6
802.3i
Ethernet (10BASE-T)
802.3µ
Fast Ethernet
802.3x
Full Duplex
802.3z
Gigabit Ethernet
Two Ethernet standards that represent initial follow-on to the initial
standard are 802.3µ and 802.12, both of which have their foundation in
IEEE efforts that occurred during 1992. In that year the IEEE requested proposals for ‘‘Fast Ethernet,’’ designed to raise the Ethernet operating rate from
10 Mbps to 100 Mbps. This request resulted in two initial proposals. One
proposal, now referred to as a series of 100BASE proposals, was developed
by a consortium that included Synoptics Communications, Inc., 3Com Corporation, and Ungermann-Bass, Inc. This proposal retained the CSMA/CD
access proposal, which formed the basis for the operation of earlier versions of Ethernet. Now included in 802.3µ are 100BASE-TX, 100BASE-FX,
and 100BASE-T4.
100BASE-TX defines the specifications for 100-Mbps CSMA/CD over two
pairs of category 5 unshielded twisted-pair (UTP) cable. 100BASE-FX specifies
100-Mbps Ethernet over two pairs of optical fiber cable, while 100BASE-T4
defines the operation of 100-Mbps Ethernet over four pairs of category 3, 4,
and 5 UTP or shielded twisted-pair (STP) cable.
The second 100-Mbps proposal, which is now referred to as 100VGAnyLAN, was initially developed by AT&T Microelectronics and HewlettPackard Company. This proposal replaced the CSMA/CD access protocol by
a demand-priority scheme that enables the support of Ethernet, Token-Ring,
FDDI, and other types of local area networks. Since this proposal described
operations on voice grade (VG) twisted pair, it received the mnemonic 100VGAnyLAN. Because the operation of 100VG-AnyLAN is based upon the passing
networking standards
51
of a token that is used to prioritize access to a network, the actual name of the
802.12 committee is Demand Priority Access.
During 1994, the IEEE 802.9 working group completed a document that
creates a 16.384-Mbps physical layer for operation on UTP category 3 or
higher cable. Referred to as isoENET, the document is technically referred to
as 802.9a. While both 100VG-AnyLAN and isoENET received a considerable
level of interest when they were proposed, they never achieved any significant
degree of commercial acceptance. Due to this, our coverage in this book of those
versions of Ethernet will be limited to a brief overview of each technology.
The CSMA/CD protocol requires stations to listen for activity before transmitting data. This means that a four-wire connection with separate pairs
for transmit and receive cannot be operated simultaneously to transmit and
receive data, precluding true full-duplex operations from occurring. However,
when an Ethernet station is connected to a port on a LAN switch, the two wire
pairs between the station enable the switch port and workstation to simultaneously transmit and receive data without the possibility of a collision occurring.
This method of full duplex CSMA/CD transmission was standardized by the
IEEE as the 802.3x standard during 1996.
While the IEEE 802.3z standard for the operation of Gigabit Ethernet transmission was completed during 1998, that standard was limited to defining
transmission at 1 Gbps over different types of optical fiber. It was not until
1999 that the 802.3ab standard was issued, which provided the physical layer
specification for 1 Gbps transmission over metallic twisted-pair standardized
as 1000BASE-T. Although it remained to be finalized, 10 Gbps Ethernet’s
physical layer specification over optical fiber was being worked on by the
IEEE 802.3ae project.
Data Link Subdivision
One of the more interesting facets of IEEE 802 standards was the initial
subdivision of the ISO Open System Interconnection Model’s data link layer
into two sublayers: logical link control (LLC) and medium access control
(MAC). Figure 2.5 illustrates the relationship between IEEE 802 local area
network standards and the first three layers of the OSI Reference Model.
The separation of the data link layer into two entities provides a mechanism
for regulating access to the medium that is independent of the method for
establishing, maintaining, and terminating the logical link between workstations. The method of regulating access to the medium is defined by the MAC
portion of each LAN standard. This enables the LLC standard to be applicable
to each type of network.
52
chapter two
OSI
Reference
model
802.1 High-level interface
(internet working)
Network
802.2 Logical link control
Figure 2.5
Model.
Data
link
802.3
Medium
access
control
802.4
Medium
access
control
802.5
Medium
access
control
802.6
Medium
access
control
802.3
802.4
802.5
802.6
Physical
Physical
Physical
Physical
Physical
Relationship between IEEE standards and the OSI Reference
Medium Access Control
The MAC sublayer is responsible for controlling access to the network. To
accomplish this, it must ensure that two or more stations do not attempt to
transmit data onto the network simultaneously. For Ethernet networks, this is
accomplished through the use of the CSMA/CD access protocol.
In addition to network access control, the MAC sublayer is responsible for
the orderly movement of data onto and off of the network. To accomplish this,
the MAC sublayer is responsible for MAC addressing, frame type recognition,
frame control, frame copying, and similar frame-related functions.
The MAC address represents the physical address of each station connected
to the network. That address can belong to a single station, can represent a
predefined group of stations (group address), or can represent all stations on
the network (broadcast address). Through MAC addresses, the physical source
and destination of frames are identified.
Frame type recognition enables the type and format of a frame to be
recognized. To ensure that frames can be processed accurately, frame control
prefixes each frame with a preamble, which consists of a predefined sequence
networking standards
53
of bits. In addition, a frame check sequence (FCS) is computed by applying an
algorithm to the contents of the frame; the results of the operation are placed
into the frame. This enables a receiving station to perform a similar operation.
Then, if the locally computed FCS matches the FCS carried in the frame, the
frame is considered to have arrived without error.
Once a frame arrives at a station that has the same address as the destination
address in the frame, that station must copy the frame. The copying operation
moves the contents of the frame into a buffer area in an Ethernet adapter card.
The adapter card removes certain fields from the frame, such as the preamble
and start of frame delimiter, and passes the information field into a predefined
memory area in the station into which the adapter card is inserted.
Refer to Chapter 4 for detailed information concerning Ethernet frame formats, as well as information concerning how the MAC layer controls the
transmission and reception of data on an Ethernet local area network.
Logical Link Control
Logical link control frames are used to provide a link between network layer
protocols and media access control. This linkage is accomplished through the
use of service access points (SAPs), which operate in much the same way as
a mailbox. That is, both network layer protocols and logical link control have
access to SAPs and can leave messages for each other in them.
Like a mailbox in a post office, each SAP has a distinct address. For the
logical link control, a SAP represents the location of a network layer process,
such as the location of an application within a workstation as viewed from the
network. From the network layer perspective, a SAP represents the place to
leave messages concerning the network services requested by an application.
LLC frames contain two special address fields, known as the destination
services access point and the source services access point. The destination
services access point (DSAP) is one byte in length and specifies the receiving
network layer process. The source services access point (SSAP) is also one
byte in length. The SSAP specifies the sending network layer process. Both
DSAP and SSAP addresses are assigned by the IEEE. Refer to Chapter 4 for
detailed information concerning LLC frame formats and data flow.
Additional Sublayering
As previously mentioned, the standardization of high-speed Ethernet resulted
in an additional sublayer at the data link layer, and the subdivision of
the physical layer. Figure 2.6 illustrates the relationship between the first two
layers of the ISO Reference Model and the IEEE 802.3µ Fast Ethernet sublayers.
54
chapter two
Data
link
layer
Logical link control
Media access control
Reconciliation sublayer
Medium-independent interface
Physical
layer
Physical coding sublayer
Physical medium attachments
Physical medium dependent
Figure 2.6
IEEE 802.3µ sublayering.
The additional sublayering illustrated in Figure 2.6 became necessary, as it
was desired to support different media with one standard. To accomplish
this required the physical layer to be independent from the data link layer,
because there can be different coding schemes used to support transmission
on different types of media.
To retain the CSMA/CD access protocol while supporting the use of different
media required the use of different connectors, resulting in the introduction of
a physical medium-dependent (PMD) sublayer. Because different data coding
schemes are required to support 100 Mbps on different types of media, a
physical coding sublayer was introduced. This sublayer defines the coding
method used for transmission on different types of media. To map messages
from the physical coding sublayer onto the transmission media resulted in
those functions being performed by the physical medium attachment sublayer.
Thus, the physical layer was subdivided into three sublayers.
Although not shown on Figure 2.6, it should be noted that an AutoNegotiation function resides under the PMD. The Auto-Negotiation function
was added to provide an ease of migration from 10 Mbps Ethernet to 100 Mbps
Ethernet and results in the Media-Independent Interface (MII) supporting both
10 and 100 Mbps data transfer. To accomplish this the MII clock is capable of
operating at 2.5 MHz and 25 MHz.
At the data link layer an additional sublayer, known as the reconciliation
sublayer, was introduced. This sublayer is responsible for reconciling the MII
from the physical layer, with the MAC signal.
networking standards
55
Logical link control
Data
link
layer
Media access control
Reconciliation
Gigabit media-independent
interface
Physical
layer
Physical coding sublayer
Physical medium attachment
Physical medium dependent
Figure 2.7
Subdivision of the physical layer of Gigabit Ethernet.
Recognizing that Gigabit Ethernet would operate on different types of media
also resulted in the subdivision of its physical layer. That subdivision is
illustrated in Figure 2.7.
The reconciliation sublayer represents a transparent interface between the
MAC sublayer and the physical layer which decouples the MAC layer from
the physical layer. The Gigabit Media Independent Interface (GMII) includes
transmit and receive data paths that are 8 bits in width, which, when coupled
with a clock that now operates at 125 MHz, results in a data transfer capability
of 1 Gbps.
From a comparison of Figure 2.6 and Figure 2.7, you will note that the
sublayering of Gigabit Ethernet is similar to that of Fast Ethernet. However,
the sublayers perform different functions. For example, under Fast Ethernet
coding is based on the FDDI specification. In comparison, under Gigabit
Ethernet reliance is shifted to the physical sublayers previously specified for
the Fiber channel, as the latter operates at approximately Gigabit data rates
and was selected for use by the 802.38 project members.
2.4 Internet Standards
The Internet as we know it dates to 1967 when the Advanced Research
Projects Agency (ARPA), operating as a part of the United States Office of the
56
chapter two
Secretary of Defense, issued a Request for Proposal (RFP) for the creation of a
packet switching network. The result of the RFP was a contract issued to Bolt,
Beranek and Newmann (BBN), a then small company based in Cambridge,
MA, whose efforts resulted in a network that enabled scientists and educators
to share information. That network was known as ARPAnet.
RFC Evolution
In an effort to share information about the operation of ARPAnet, Steve Crocker,
a then graduate student at UCLA, published the first Request for Comment
(RFC) in April, 1969, which was titled ‘‘Host Software.’’ The term RFC was
used, as Mr Crocker wanted others to comment on the information he provided.
As ARPAnet expanded and evolved into the Internet, various organizations
came into being. The Internet Activities Board (IAB) became responsible for
Internet design and planning. Two task forces reside under the IAB — the Internet Engineering Task Force (IETF) and the Internet Research Task Force (IRTF).
The responsibility of the IETF involves the coordination of various technical
aspects of the Internet to include the development and modification of protocols that may be necessary to obtain a desired level of functionality.
The IAB is currently responsible for defining protocols and operational
procedures that require both dissemination to the Internet community and
archiving. To accomplish this, the IAB primarily issues documents known
as Requests for Comments (RFCs), the majority of which begin as working
memorandums issued by the IETF.
Types and Submission
There are several types of RFCs. Types of RFCs can include a Draft Standard
RFC, a Proposed Standard RFC, a Full Standard RFC, an Experimental RFC, a
Best Current Practice RFC and a ‘‘For Your Information’’ RFC. While RFCs are
not referred publications, they are technically reviewed by either individual
technical experts, the RFC editor, or members of a task force. Anyone can
submit a document for publication as an RFC. Once submitted to the RFC
editor the document may be edited to comply with certain format rules, which
are currently specified in RFC 2223, issued in October, 1997, which obsoleted
RFC 1543.
An initial RFC submission is usually treated as a Preliminary Draft and
is electronically distributed for comment. On its way to becoming a Full
Standard, and it should be noted that many RFCs are not intended to be
standards, the Preliminary Draft may first become a Proposed Standard.
networking standards
Preliminary
draft
Proposed
standard
6 months
minimum
Draft
standard
57
Standard
4 months
minimum
Time
Figure 2.8
Internet standards time track.
Figure 2.8 illustrates the typical track time for the development of an Internet
standard. Once a Preliminary Draft is submitted it can take approximately six
months for the receipt of comments concerning the draft and to allow the
draft to be moved forward to be published as a Proposed Standard, and either
dropped or promoted to a Draft Standard. After a review period of at least four
months, a Draft Standard can be recommended for adoption as a Standard
by the Internet Engineering Steering Group (IESC). The IESC consist of the
chairperson of the IETF and other members of that group, and it performs
an oversight and coordinating function for the IETF. Although the IESG is
responsible for recommending the adoption of an RFC as a Standard, the IAB
is responsible for the final decision concerning its adoption.
Obtaining RFCs
There are many Web sites where you can locate RFCs. The RFC Editor
maintains the official repository of all RFCs and indexes to them. The RFC
Editor web location is:
http://www.rfc-editor.org
In addition to the RFC Editor web site there are many locations that mirror
RFC information. One popular web site is maintained by the Computer and
Information Science Department of Ohio State University. The address of that
web site is:
http://www.cis.ohio-state.edu/services/rfc
At this site you can review a complete index of all RFCs, view specific RFCs,
and obtain the capability to perform a keyboard search of a comprehensive
database of RFCs.
58
chapter two
2.5 Cabling Standards
Any discussion of Ethernet networks requires knowledge of both existing and
pending cabling standards. In this section we will focus our attention upon
the EIA/TIA-568 standard, first examining existing standards and then turning
our attention to developing cabling standards that may be in place by the time
you read this book.
EIA/TIA-568
The Electronics Industry Association/Telecommunications Industries Association ‘‘Commercial Building Telecommunications Standard,’’ commonly
referred to as EIA/TIA-568, was ratified in 1992. This standard specifies a
variety of building cabling parameters, ranging from backbone cabling used
to connect a building’s telecommunication closets to an equipment room, to
horizontal cabling used to cable individual users to the equipment closet.
The standard defines the performance characteristics of both backbone and
horizontal cables as well as different types of connectors used with different
types of cable.
Backbone Cabling
Four types of media are recognized by the EIA/TIA-568 standard for backbone
cabling. Table 2.3 lists the media options supported by the EIA/TIA-568
standard for backbone cabling.
Horizontal Cabling
As previously indicated, horizontal cabling under the EIA/TIA-568 standard
consists of cable that connects equipment in a telecommunications closet to
a user’s work area. The media options supported for horizontal cabling are
TABLE 2.3
EIA/TIA-568 Backbone Cabling Media Options
Media Type
Maximum Cable Distance
100-ohm UTP
800 meters (2624 feet)
150-ohm STP
700 meters (2296 feet)
50-ohm thick coaxial cable
500 meters (1640 feet)
62.5/125-µ multimode optical fiber
2000 meters (6560 feet)
networking standards
59
the same as specified for backbone cabling, with the exception of coaxial
cable for which 50-ohm thin cable is specified; however, cabling distances are
restricted to 90 meters in length from equipment in the telecommunications
closet to a telecommunications outlet. This permits a patch cord or drop
cable up to 10 meters in length to be used to connect a user workstation
to a telecommunications outlet, resulting in the total length of horizontal
cabling not exceeding the 100-meter restriction associated with many LAN
technologies that use UTP cabling.
UTP Categories
One of the more interesting aspects of the EIA/TIA-568 standard is its recognition that different signaling rates require different cable characteristics. This
resulted in the EIA/TIA-568 standard initially classifying UTP cable into five
categories. Those categories and their suitability for different types of voice
and data applications are indicated in Table 2.4.
In examining the entries in Table 2.4, note that categories 3 through 5
support transmission with respect to indicated signaling rates. This means
that the ability of those categories of UTP to support different types of LAN
transmission will depend upon the signaling method used by different LANs.
For example, consider a LAN encoding technique that results in 6 bits encoded
into 4 signaling elements that have a 100-MHz signaling rate. Through the use
of category 5 cable, a data transmission rate of 150 Mbps ((6/4) × 100) could
be supported.
Category 3 cable is typically used for Ethernet and 4 Mbps Token-Ring
LANs. Category 4 is normally used for 16-Mbps Token-Ring LANs, while category 5 cable supports 100-Mbps Ethernet LANs, such as 100VG-AnyLAN
TABLE 2.4
EIA/TIA-568 UTP Cable Categories
Category 1
Voice or low-speed data up to
56 Kbps; not useful for LANs.
Category 2
Data rates up to 1 Mbps.
Category 3
Supports transmission up to
16 MHz.
Category 4
Supports transmission up to
20 MHz.
Category 5
Supports transmission up to
100 MHz.
60
chapter two
and 100BASE-T, and will support ATM to the desktop at a 155-Mbps
operating rate. Two additional metallic cable categories being considered for
standardization are category 5 extended (cat 5e) and category 6. Category 5e
represents more stringent existing specifications as well as specifications for
existing parameters that we will shortly review. Although cat 5e is only specified for operations up to 100 MHz, it is used to support 1000BASE-T. However,
the proposed cat 6 standard that will support signaling up to 200 MHz should
eventually become the preferred cable for supporting Gigabit Ethernet over
copper media.
Cable Specifications
There are two basic metrics that define the capability of EIA/TIA-568 cable
with respect to the signaling rate they support, which in turn defines the cable
category. Those metrics are attenuation and near-end crosstalk (NEXT).
Attenuation
Attenuation represents the loss of signal power as a signal propagates from
a transmitter at one end of a cable toward a receiving device located at the
distant end of the cable. Attenuation is measured in decibels (dB) as indicated:
Attenuation = 20 log10
(transmit voltage)
receive voltage
For those of us a little rusty with logarithms, let’s examine a few examples
of attenuation computations. First, let’s assume the transmit voltage was 100,
while the receive voltage was 1. Then,
Attenuation = 20 log10
(100)
= 20 log10 100
1
The value of log10 100 can be obtained by determining the power to which
10 should be raised to equal 100. Because the answer is 2(102 = 100), log10 100
has a value of 2, and 20 log10 100 then has a value of 40.
Now let’s assume the transmit voltage was 10 while the receiver voltage was
1. Then,
(10)
= 20 log10 10
Attenuation = 20 log10
1
Because the value of log10 10 is 1(101 = 10), then 20 log10 10 has a value of 20.
From the preceding, note that a lower level of signal power loss results in a
lower level of attenuation.
networking standards
61
NEXT
Crosstalk represents the electromagnetic interference caused by a signal on
one wire pair being emitted onto another wire pair, resulting in the generation
of noise. Because transmit and receive pairs are twisted and the transmit
signal is strongest at its source, the maximum level of interference occurs at
the cable connector and decreases as the transmit signal traverses the cable.
Recognizing this fact of physics, crosstalk is measured at the near end, hence
the term near-end crosstalk (NEXT).
NEXT denotes the induced or coupled signal flowing from the transmit
pair to the receive pair even though the two pairs are not interconnected.
Mathematically, NEXT is defined in decibels (dB) as follows:
NEXT = 20 log10
(transmitted voltage)
coupled voltage
In the preceding equation the transmit voltage represents the power placed on
the transmit pair, while the coupled signal is measured on the receive pair at
the location where the transmit voltage was generated. Note that a larger dB
NEXT measurement is better as it indicates a lower level of crosstalk and is
the opposite of attenuation, because a lower attenuation reading indicates less
signal loss and is better than a higher reading for that parameter. Table 2.5
indicates the EIA/TIA-568 specification limits for categories 3, 4, and 5 UTP
cable. In examining Table 2.5, note that both attenuation and NEXT must
be measured over a range of frequencies. That range is based upon the
cable category. For example, because category 3 cable is designed to support
signaling rates up to 16 MHz, attenuation and NEXT should be measured up
to and including the highest signaling rate supported by that type of cable,
which is 16 MHz.
Other Metrics
When the EIA/TIA considered the development of additional cabling specifications, it recognized the need to include additional parameters in the
specifications it developed. Three of those additional specifications concern
power sum NEXT (PS NEXT), Equal Level Far End Crosstalk (EL FEXT) and
the power sum attenuation to crosstalk ratio (PS ACR).
PS NEXT
Power sum NEXT actually represents a computation and not an individual
measurement. PS NEXT is obtained by computing the algebraic summation of
62
chapter two
TABLE 2.5
Frequency
(MHz)
EIA/TIA-568 Attenuation and NEXT Limits in dB
Category 3
Category 4
Category 5
Attenuation
NEXT
Attenuation
NEXT
Attenuation
NEXT
1.0
4.2
39.1
2.6
53.3
2.5
60.3
4.0
7.3
29.3
4.8
43.3
4.5
50.6
8.0
10.2
24.3
6.7
38.2
6.3
45.6
10.0
11.5
22.7
7.5
36.6
7.0
44.0
16.0
14.9
19.3
9.9
33.1
9.2
40.6
20.0
—
—
11.0
31.4
10.3
39.0
25.0
—
—
—
—
11.4
37.4
31.2
—
—
—
—
12.8
35.7
62.5
—
—
—
—
18.5
30.6
100.0
—
—
—
—
24.0
27.1
the individual NEXT effects on each pair by the other pairs in a cable. Because
1000BASE-T uses four pairs, PS NEXT represents an important measurement
for qualifying cabling required to support Gigabit Ethernet over copper media.
PS NEXT represents a measure of difference in signal strength between
disturbing pairs and a disturbed pair. This means that a larger number, which
represents less crosstalk, is more desirable than a small number that represents
more crosstalk.
EL FEXT
Equal Level Far End Crosstalk (EL FEXT) also represents a calculated specification and not a single measurement. EL FEXT is computed by subtracting
the attenuation of the disturbing pair from the Far End Crosstalk that the pair
introduces in adjacent pairs.
To illustrate the computation of EL FEXT, assume FEXT was measured to
be −47 dB while attenuation was determined to be −12 dB. Then, EL FEXT
becomes −47 − (−12) or −35 dB. Note that EL FEXT provides a normalized computation based upon the length of a cable since attenuation varies
by length.
networking standards
TABLE 2.6
63
Recent and Emerging EIA/TIA Cable Specifications
Category 5
Category 5e
Category 6
(Proposed)
Frequency Range
1–100 MHz
1–100 MHz
1–200 MHz
Attenuation
24 dB
24 dB
21.7 dB
NEXT
27.1 dB
30.1 dB
39.9 dB
Power sum NEXT
N/A
27.1 dB
37.1 dB
ACR
3.1 dB
6.1 dB
18.2 dB
Power sum ACR
N/A
3.1 dB
15.4 dB
EL FEXT
17 dB
17.4 dB
23.2 dB
Power sum EL FEXT
14.4 dB
14.4 dB
20.2 dB
Return Loss
8 dB
10 dB
12.0 dB
Propagation Delay
548 ns
548 ns
548 ns
Delay Skew
50 ns
50 ns
50 ns
Specification
PS ACR
Similar to PS NEXT and EL FEXT, the power sum attenuation to crosstalk
ratio (PS ACR) represents a computation and not an individual measurement.
PS ACR is determined by computing an algebraic summation of individual
ACR effects, with four results at each end of a link being tested.
Because ACR represents a measure of attenuation to crosstalk as a ratio, PS
ACR also represents a summed ratio. Thus, a larger number that represents
more signal and less noise is more desirable than a smaller number, which
represents more noise and less signal.
Cat 5e and Cat 6
When this new edition was prepared, category 5e had been standardized while
category 6 was being proposed as a new specification. To provide a frame of
reference between the newly specified category 5e and proposed category 6
cabling, Table 2.6 provides a comparison of those specifications to category 5
cable. In examining Table 2.6 note that several category 5 specifications are
not actually specified by that cabling specification and are only listed for
comparison purposes.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter three
Ethernet Networks
From the title of this chapter, it is apparent that there is more than one type
of Ethernet network. From a network access perspective, there is actually
only one Ethernet network. However, the CSMA/CD access protocol used
by Ethernet, as well as its general frame format and most of its operating
characteristics, were used by the IEEE to develop a series of Ethernet-type
networks under the IEEE 802.3 umbrella. Thus, this chapter will first focus on
the different types of Ethernet networks by closely examining the components
and operating characteristics of Ethernet and then comparing its major features
with the different networks defined by the IEEE 802.3 standard. Once this
is accomplished, we will focus our attention on the wiring, topology, and
hardware components associated with each type of IEEE 802.3 Ethernet
network. This will enable us to examine the construction of several types of
802.3 networks using a variety of hardware devices and then illustrate how
those networks can be connected to one another — a process referred to as
internetworking.
Although significant advances in Ethernet technology have occurred over
the past decade, many features and constraints associated with newer technology are based upon the original technology. Due to this we will begin at
the beginning in this chapter and examine the characteristics of each Ethernet
Network in the order in which they were developed.
3.1 Ethernet
One of the key concepts behind Ethernet — that of allocating the use of a shared
channel — can be traced to the pioneering efforts of Dr. Norman Abramson
and his colleagues at the University of Hawaii during the early 1970s. Using a
ground-based radio broadcasting system to connect different locations through
the use of a shared channel, Abramson and his colleagues developed the concept of listening to the channel before transmission, transmitting a frame of
65
66
chapter three
information, listening to the channel output to determine whether a collision
occurred, and, if it did, waiting a random period of time before retransmission.
The resulting University of Hawaii ground-based radio broadcasting system,
called ALOHA, formed the basis for the development of numerous channel
contention systems, including Ethernet. In addition, the subdivision of transmission into frames of data was the pioneering work in the development of
packet-switching networks. Thus, Norman Abramson and his colleagues can
be considered the forefathers of two of the most important communications
technologies, contention networks and packet-switching networks.
Evolution
The actual development of Ethernet occurred at the Xerox Palo Alto Research
Center (PARC) in Palo Alto, California. A development team headed by
Dr. Robert Metcalfe had to connect over 100 computers on a 1-km cable. The
resulting system, which operated at 2.94 Mbps using the CSMA/CD access
protocol, was referred to as ‘‘Ethernet’’ in a memorandum authored by Metcalfe. He named it after the luminiferous ether through which electromagnetic
radiation was once thought to propagate.
During its progression from a research-based network into a manufactured
product, Ethernet suffered several identity crises. During the 1970s, it endured
such temporary names as the ‘‘Alto Aloha Network’’ and the ‘‘Xerox Wire.’’
After reverting to the original name, Xerox decided, quite wisely, that the
establishment of Ethernet as an industry standard for local area networks
would be expedited by an alliance with other vendors. A resulting alliance
with Digital Equipment Corporation and Intel Corporation, which was known
as the DIX Consortium, resulted in the development of a 10-Mbps Ethernet network. It also provided Ethernet with a significant advantage over Datapoint’s
ARCNet and Wang Laboratories’ Wangnet, proprietary local area networks
that were the main competitors to Ethernet during the 1970s.
The alliance between Digital Equipment, Intel, and Xerox resulted in the
publication of a ‘‘Blue Book Standard’’ for Ethernet Version 1. An enhancement to that standard occurred in 1982 and is referred to as Ethernet Version 2
or Ethernet II in many technical publications. Although the DIX Consortium
submitted its Ethernet specification to the IEEE in 1980, it wasn’t until 1982
that the IEEE 802.3 CSMA/CD standard was promulgated. Because the IEEE
used Ethernet Version 2 as the basis for the 802.3 CSMA/CD standard, and
Ethernet Version 1 has been obsolete for over approximately two decades, we
will refer to Ethernet Version 2 as Ethernet in the remainder of this book.
ethernet networks
67
Network Components
The 10-Mbps Ethernet network standard originally developed by Xerox,
Digital Equipment Corporation, and Intel was based on the use of five hardware components. Those components include a coaxial cable, a cable tap,
a transceiver, a transceiver cable, and an interface board (also known as
an Ethernet controller). Figure 3.1 illustrates the relationships among Ethernet components.
Coaxial Cable
One of the problems faced by the designers of Ethernet was the selection of an
appropriate medium. Although twisted-pair wire is relatively inexpensive and
easy to use, the short distances between twists serve as an antenna for receiving
electromagnetic and radio frequency interference in the form of noise. Thus,
the use of twisted-pair cable restricts the network to relatively short distances.
Coaxial cable, however, has a dielectric shielding the conductor. As long
as the ends of the cable are terminated, coaxial cable can transmit over
greater distances than twisted-pair cable. Because the original development of
Ethernet was oriented toward interconnecting computers located in different
Transceiver
cable
Interface board
(controller)
in computer
Transceiver
Cable
core
Cable tap
Coaxial cable
Figure 3.1 Ethernet hardware components. When thick coaxial cable is used
for the bus, an Ethernet cable connection is made with a transceiver cable and
a transceiver tapped into the cable.
68
chapter three
buildings, the use of coaxial cable was well suited for this requirement. Thus,
the initial selection for Ethernet transmission medium was coaxial cable.
There are two types of coaxial cable that can be used to form the main
Ethernet bus. The first type of coaxial cable specified for Ethernet was a relatively thick 50-ohm cable, which is normally colored yellow and is commonly
referred to as ‘‘thick’’ Ethernet. This cable has a marking every 2.5 meters to
indicate where a tap should occur, if one is required to connect a station to the
main cable at a particular location. These markings represent the minimum
distance one tap must be separated from another on an Ethernet network.
The outer insulation or jacket of the yellow-colored cable is constructed using
PVC. A second popular type of 50-ohm cable has a Teflon jacket and is colored
orange-brown. The Teflon jacket coax is used for plenum-required installations in air-handling spaces, referred to as plenums, to satisfy fire regulations.
When installing a thick coaxial segment the cable should be rolled from a
common cable spool or cable spools manufactured at the same time, referred
to as a similar cable lot, to minimize irregularities between cables. Under
the Ethernet specifications when the use of cable from different lots cannot
be avoided, cable sections should be used that are either 23.4 m, 70.2 m,
or 117 m in length. Those cable lengths minimize the possibility of excessive signal reflections occurring due to variances in the minor differences
in cable produced by different vendors or from different cable lots from the
same vendor.
A second type of coaxial cable used with Ethernet is smaller and more
flexible; however, it is capable of providing a transmission distance only onethird of that obtainable on thick cable. This lighter and more flexible cable is
referred to as ‘‘thin’’ Ethernet and also has an impedance of 50 ohms. When
the IEEE standardized Ethernet, the thick coaxial cable–based network was
assigned the designation 10BASE-5, while the network that uses the thinner
cable was assigned the designator 10BASE-2. Later in this chapter we will
examine IEEE 802.3 networks under which 10BASE-5, 10BASE-2, and other
Ethernet network designators are defined.
Two of the major advantages of thin Ethernet over thick cable are its cost
and its use of BNC connectors. Thin Ethernet is significantly less expensive
than thick Ethernet. Thick Ethernet requires connections via taps, whereas
the use of thin Ethernet permits connections to the bus via industry standard
BNC connectors that form T-junctions.
Transceiver and Transceiver Cable
Transceiver is a shortened form of transmitter-receiver. This device contains electronics to transmit and receive signals carried by the coaxial cable.
ethernet networks
69
The transceiver contains a tap that, when pushed against the coaxial cable,
penetrates the cable and makes contact with the core of the cable. Ethernet transceivers are used for broadband transmission on a coaxial cable and
usually include a removable tap assembly. The latter enables vendors to
manufacture transceivers that can operate on thick and thin coaxial cable,
enabling network installers to change only the tap instead of the entire device
and eliminating the necessity to purchase multiple types of transceivers to
accommodate different media requirements. In books and technical literature
the transceiver, its tap, and its housing are often referred to as the medium
attachment unit (MAU).
The transceiver is responsible for carrier detection and collision detection.
When a collision is detected during a transmission, the transceiver places
a special signal, known as a jam, on the cable. This signal, described in
Chapter 4, is of sufficient duration to propagate down the network bus and
inform all of the other transceivers attached to the bus node that a collision
has occurred.
The cable that connects the interface board to the transceiver is known
as the transceiver cable. This cable can be up to 50 meters (165 feet) in
length and contains five individually shielded twisted pairs. Two pairs are
used for data in and data out, and two pairs are used for control signals in
and out. The remaining pair, which is not always used, permits the power
from the computer in which the interface board is inserted to power the
transceiver.
Because collision detection is a critical part of the CSMA/CD access protocol,
the original version of Ethernet was modified to inform the interface board that
the transceiver collision circuitry is operational. This modification resulted in
each transceiver’s sending a signal to the attached interface board after every
transmission, informing the board that the transceiver’s collision circuitry
is operational. This signal is sent by the transceiver over the collision pair
of the transceiver cable and must start within 0.6 microseconds after each
frame is transmitted. The duration of the signal can vary between 0.5 and
1.5 microseconds. Known as the signal quality error and also referred to
as the SQE or heartbeat, this signal is supported by Ethernet Version 2.0,
published as a standard in 1982, and by the IEEE 802.3 standard. Although
the heartbeat (SQE) is between the transceiver and the system to which it is
attached, under the IEEE 802.3 standard transceivers attached to a repeater
must have their heartbeat disabled.
The SQE signal is simply a delayed response by a few bit times to the
transmission of each frame, informing the interface card that everything is
working normally. Because the SQE signal only flows from the transceiver
70
chapter three
back to the interface card, it does not delay packet transmission nor does it
flow onto the network. Today most transceivers have a switch or jumper that
enables the SQE signal, commonly labeled SQE Test, to be disabled. Because
repeaters must monitor signals in real time and cannot use the Ethernet time
gap of 9.6 ms between frames (which we will discuss later in this book), this
means that they are not capable of recognizing a heartbeat signal. It should be
noted that a twisted-pair 10BASE-T Ethernet hub is also a repeater. If you fail
to disable the SQE Test signal, the repeater electronics to include hub ports
will misinterpret the signal as a collision. This will result in the transmission
of a jam signal on all hub ports other than the port receiving the SQE Test
signal, significantly degrading network performance.
Interface Board
The interface board, or network interface card (NIC), is inserted into an
expansion slot within a computer and is responsible for transmitting frames
to and receiving frames from the transceiver. This board contains several
special chips, including a controller chip that assembles data into an Ethernet
frame and computes the cyclic redundancy check used for error detection.
Thus, this board is also referred to as an Ethernet controller.
Most Ethernet interface boards contain a DB-15 connector for connecting
the board to the transceiver. Once thin Ethernet cabling became popular,
many manufacturers made their interface boards with both DB-15 and BNC
connectors. The latter was used to permit the interface board to be connected
to a thin Ethernet cable through the use of a T-connector. Figure 3.2 illustrates
the rear panel of a network interface card containing both DB-15 and BNC
connectors. With the development of twisted-pair-based Ethernet, such as
10BASE-T, modern Ethernet interface boards, which are commonly referred
to as network interface cards (NICs), also include an RJ-45 connector to
accommodate a connection to twisted-wire-based networks.
Cabling Restrictions
Under the Ethernet standard developed by Xerox, Digital Equipment Corporation, and Intel Corporation, a thick coaxial cable is permitted a maximum
length of 500 meters (1640 feet). Multiple cable segments can be joined
together through the use of repeaters; however, the maximum cable distance between two transceivers is limited to 2.5 km (8200 feet), and no more
than four repeaters can be traversed on any path between transceivers.
Each thick trunk cable segment must be terminated with what is known as
an N-series connector on each end of the cable. The terminator ‘‘terminates’’
ethernet networks
71
Figure 3.2 Ethernet interface board connectors. The first
generation Ethernet interface boards (network interface
cards) contain both DB-15 and BNC connectors to support
the use of either thick or thin coaxial cable. A second
generation of interface cards included an RJ-45 connector
to Accommodate a connection to twisted-wire-based
networks.
the network and blocks electrical interference from flowing onto what would
otherwise be exposed cable. One N-series connector also serves as a ground,
when used with an attached grounding wire that can be connected to the
middle screw of a dual AC electrical power outlet.
Figure 3.3 illustrates a thick Ethernet cable segment after an installer fastened N-series plugs to each cable end. This is normally accomplished after
the desired length of coaxial cable is routed to form the required network bus.
Next, an N-series terminator connector is fastened onto one N-series plug,
while an N-series terminator with ground wire is fastened onto the N-series
plug at the opposite end of the cable segment.
In addition, as previously mentioned, attachments to the common bus must
be separated by multiples of 2.5 meters. The latter cabling restriction prevents
reflections caused by taps in the main cable from adding up in phase and being
mistaken by one transceiver for another’s transmission. For the total network,
up to 1024 attachments are allowed, including all cable sections connected
through the use of repeaters; however, no more than 100 transceivers can be
on any one cable segment.
Repeaters
A repeater is a device that can be used to connect two network segments
together to form a larger local area network topology. The repeater receives,
72
chapter three
Thick ethernet cable segment
N-series
plugs
N-series
jacks
N-series
terminator
N-series
terminator
with ground wire
Figure 3.3 Each Ethernet thick coaxial cable segment has N-series plugs on
each end. They are terminated through the use of N-series terminators, one of
which contains a ground wire or ground wire connection.
amplifies, and retransmits signals, restoring the symmetry and position of each
signal. Signal amplification results in the restoration of the original amplitude
characteristics of the data signal. The restoration of signal symmetry results
in each output signal pulse matching the shape of the originally transmitted
signal. The last function performed by a repeater is the restoration of the
signal position. More formally referred to as retiming, this repeater function
results in the data signal output in its correct position by time, removing any
prior shift or displacement in the placement of the received signal. That shift
or displacement is known as jitter, while a very small shift or displacement of
a transmitted signal is referred to as a wander.
Because a repeater operates at the physical layer, it is transparent to data
and simply regenerates signals. Figure 3.4 illustrates the use of a repeater to
connect two Ethernet cable segments. As indicated, a transceiver is taped
to each cable segment to be connected, and the repeater is cabled to the
transceiver. When used to connect cable segments, a repeater counts as one
station on each connected segment. Thus, a segment capable of supporting
up to 100 stations can support only 99 additional stations when a repeater is
used to connect cable segments. Although not shown, each cable segment is
terminated at each end with an N-series terminator and grounded at one end
of one segment.
ethernet networks
Cable
segment
73
Transceiver
R
Transceiver
Cable
segment
Figure 3.4 Using a repeater. Cable segments can be joined together by a
repeater to expand the network. The repeater counts as a station on each
cable segment.
In examining Figure 3.4, note that any data traffic carried on the top cable
segment will be repeated onto the lower cable segment. Similarly, any cable
traffic transported on the lower cable segment will be repeated onto the upper
cable segment. Thus, the use of a repeater simply provides a mechanism to
extend transmission between cable segments and should not be confused with
the use of a bridge that normally isolates traffic to different segments unless
data is destined to a station on a different segment.
The 5-4-3 Rule
When Ethernet was developed it was recognized that the use of repeaters to
connect segments to form a larger network would result in pulse regeneration
delays that could adversely affect the probability of collisions. Thus, a limit
was required on the number of repeaters that could be used to connect segments together. This limit in turn limited the number of segments that could
be interconnected. A further limitation involved the number of populated segments that could be joined together, because stations on populated segments
generate traffic that can cause collisions, whereas nonpopulated segments
are more suitable for extending the length of a network of interconnected
segments. A result of the preceding was the ‘‘5-4-3 rule.’’ That rule specifies
that a maximum of five Ethernet segments can be joined through the use
of a maximum of four repeaters. In actuality, this part of the Ethernet rule
74
chapter three
really means that no two communicating Ethernet nodes can be more than
two repeaters away from one another. Finally, the ‘‘three’’ in the rule denotes
the maximum number of Ethernet segments that can be populated. Figure 3.5
illustrates an example of the 5-4-3 rule for the original bus-based Ethernet.
Note that this rule is also applicable to hub-based Ethernet LANs, such as
10BASE-T, which we will examine later in this chapter.
3.2 IEEE 802.3 Networks
The IEEE 802.3 standard is based on Ethernet. However, it has several significant differences, particularly its support of multiple physical layer options,
which include 50- and 75-ohm coaxial cable, unshielded twisted-pair wire,
and the use of optical fiber. Other differences between various types of IEEE
802.3 networks and Ethernet include the data rates supported by some 802.3
networks, their methods of signaling, the maximum cable segment lengths
permitted before the use of repeaters, and their network topologies.
Network Names
The standards that define IEEE 802.3 networks have been given names that
generally follow the form ‘‘s type l.’’ Here, s refers to the speed of the network
in Mbps, type is BASE for baseband and BROAD for broadband, and l refers
to the maximum segment length in 100-meter multiples. Thus, 10BASE-5
refers to an IEEE 802.3 baseband network that operates at 10 Mbps and has a
maximum segment length of 500 meters. One exception to this general form
Terminator
Terminator
R1
Ethernet
segment 1
R2
Ethernet
segment 2
R3
Ethernet
segment 3
R4
Ethernet
segment 4
Ethernet
segment 5
Legend:
O = network node
R = repeater
Figure 3.5 The 5-4-3 rule. Under the 5-4-3 Ethernet rule a maximum of five
segments can be connected through the use of four repeaters, with a maximum
of three segments populated with nodes.
ethernet networks
75
is 10BASE-T, which is the name for an IEEE 802.3 network that operates at
10 Mbps using UTP wire.
Although recently introduced high-speed Ethernet networks do not exactly
follow the preceding form, they continue to resemble prior naming conventions. For example, 100BASE-TX, 100BASE-T4, and 100BASE-FX represent
three physical layer versions of the Fast Ethernet standard. Each version
operates at 100 Mbps and uses baseband signaling; however, the suffix now
represents different types of media instead of a maximum segment length in
100-meter multiples.
Table 3.1 compares the operating characteristics of five IEEE 802.3 networks
with Ethernet. Note that the comparisons indicated in Table 3.1 do not
consider differences in the composition of Ethernet and IEEE 802.3 frames.
Those differences preclude compatibility between Ethernet and IEEE 802.3
networks, and are discussed in detail in Chapter 4.
Also note that Table 3.1 does not include IEEE 802.3µ (Fast Ethernet)
and IEEE 802.12 (100VG-AnyLAN) Ethernet networks nor any of the emerging Gigabit Ethernet networks specified under the IEEE 802.z standard. The
first standard actually represents three standards for 100-Mbps CSMA/CD
operations. The second standard supports 100-Mbps operations using a
demand-priority scheme in place of the CSMA/CD access protocol. Fast
Ethernet, 100VG-AnyLAN, and Gigabit Ethernet networks are covered at the
end of this chapter when we turn our attention to high-speed Ethernet networks. Thus, the networks compared in Table 3.1 are limited to a 10-Mbps
operating rate.
10BASE-5
As indicated in Table 3.1, the IEEE 10BASE-5 standard resembles the original
Ethernet more closely than the other 802.3 standards. In fact, an examination of the operating characteristics of Ethernet and 10BASE-5 indicates
that these networks are exactly the same. Similarities between Ethernet and
10BASE-5 include the use of DB-15 connectors on interface boards and
transceivers (MAUs) and the termination of 10BASE-5 cable segments with
N-series terminators.
However, there are differences in the frame format used by each network,
and these differences preclude compatibility between Ethernet and all IEEE
802.3 standards. In addition, under the IEEE 802.3 specification, several
network components have different names.
Figure 3.6 illustrates the major terminology changes between Ethernet and
the IEEE 802.3 10BASE-5 network. These changes are in the media interface:
10
CSMA/CD
Baseband
Manchester
500
100
50-ohm
coaxial
(thick)
Bus
Access protocol
Type of signaling
Data encoding
Maximum segment
length (meters)
Stations/segment
Medium
Topology
Ethernet
Bus
Star
Star
Unshielded
twisted pair
Unshielded
twisted pair
50-ohm
coaxial
(thin)
50-ohm
coaxial
(thick)
Bus
12/hub
100
Manchester
Baseband
CSMA/CD
10
10BASE-T
12/hub
250
Manchester
Baseband
CSMA/CD
1
1BASE-5
30
185
Manchester
Baseband
CSMA/CD
10
10BASE-2
100
500
Manchester
Baseband
CSMA/CD
10
10BASE-5
Ethernet and IEEE 802.3 1-Mbps and 10-Mbps Network Characteristics
Operating rate
(Mbps)
Operational
Characteristics
TABLE 3.1
Bus
75-ohm
coaxial
100
1,800
Manchester
Broadband
CSMA/CD
10
10BROAD-36
76
chapter three
ethernet networks
Ethernet
IEEE 10BASE-5
Station
Station
Controller
NIC
Transceiver cable
Transceiver
77
AUI
MAU
Legend:
AUI = Attachment unit interface
MAU = Media attachment unit
NIC = Network interface card
Figure 3.6 Ethernet and 10BASE-5 media interface differences. Terminology
changes under the IEEE 10BASE-5 standard resulted in the transceiver being
called the media attachment unit, while the transceiver cable is known as the
attachment unit interface.
the transceiver cable is referred to as the attachment unit interface (AUI), and
the transceiver, including its tap and housing, is referred to as the medium
attachment unit (MAU). The Ethernet controller, also known as an interface
board, is now known as the network interface card (NIC). Other similarities
between Ethernet and 10BASE-5 include the use of DB-15 connectors on
network interface cards and MAUs and the termination of 10BASE-5 cable
segments with the use of N-series 50-ohm terminators at each cable segment
end. In addition, the 802.3 standard requires one end of the thick coax segment
to be grounded for electrical safety. All other metal portions on the coax should
be insulated and fastened in place using plastic or other nonmetallic cable
ties to avoid accidentally connecting to the electrical ground.
The actual connection between the NIC and the MAU is limited to a
maximum of 50 meters (164 feet). The 15-pin AUI connector is a socket
(female) receptacle while the connector on the MAU is a 15-pin AUI connector
with a pin (male) receptacle. The AUI cable consists of four pairs of shielded
twisted wires, three of which transport data signals while one pair carries
12-volt DC power from the Ethernet interface to the MAU. The three data
78
chapter three
signal wire pairs include transmit data that go from the Ethernet interface to
the network, receive data that carry data from the network to the interface,
and a collision indicator signal wire pair that transports collision indications
to the interface.
Although a standard AUI cable is approximately 0.4 inches in diameter and
has a limited degree of flexibility, many cable manufacturers market thinner
and more flexible cable. However, the trade-off when using such cable is a
lower allowable cabling length from the interface to the MAU, because thinner
cable has a higher amount of signal attenuation.
Both the original Ethernet and the IEEE 802.3 10BASE-5 standards support
a data rate of 10 Mbps and a maximum cable segment length of 500 meters.
10BASE-5, like Ethernet, requires a minimum spacing of 2.5 meters between
MAUs and supports a maximum of five segments in any end-to-end path
through the traversal of up to four repeaters in any path. Within any path,
no more than three cable segments can be populated — have stations attached
to the cable — and the maximum number of attachments per segment is
limited to 100. As previously discussed in our coverage of Ethernet, these
restrictions are sometimes referred to as the 5-4-3 rule, referencing a maximum
of five segments linked through four repeaters, with no more than three
segments populated.
Advantages and Disadvantages
Similar to any technology there are various advantages and disadvantages
associated with the use of a 10BASE-5 network. Major advantages associated
with the use of 10BASE-5 include the span distance of 500 m (1650 feet),
which makes it extremely useful for connecting multiple locations within a
building without the use of repeaters, its ability to span up to 2.5 km (8200 feet)
by connecting multiple segments together to form an extended backbone, and
its heavily shielded media, making it suitable for use in electrically noisy
environments.
The thick coax required by 10BASE-5 makes it a very inflexible network
where the movement of a node can be difficult, if not impossible, to accomplish
without a major restructuring of the network. In addition, the failure of any
part of the cable or any node can cause the entire network to fail, resulting
in 10BASE-T being fault-intolerant. This makes troubleshooting a failure,
difficult and time-consuming, because each node and its connection has to be
checked until the point of failure is located. A fourth disadvantage is the fact
that power from different sources results in a difference in voltage between
points on the network. This voltage differential causes current to flow through
the shields of the cable, which can cause noise to be introduced into the
ethernet networks
79
center conductor, a situation referred to as creating a ground loop. 10BASE-5
networks are very susceptible to ground loops, which represent a weakness of
the technology. Although common during the early 1980s, most new network
requirements only consider 10BASE-5 as a backbone to connect multiple
10BASE-T hubs to create one large network as illustrated in Figure 3.7.
10BASE-2
10BASE-2 is a smaller and less expensive version of 10BASE-5. This standard
uses a thinner RG-58 coaxial cable, more commonly referred to as RG-58 A/U
or RG-58 C/U, thus earning the names of cheapnet and thinnet, as well as thin
Ethernet. Although 10BASE-2 cable is both less expensive and easier to use
than 10BASE-5 cable, it cannot carry signals as far as 10BASE-5 cable.
The coaxial cable used by 10BASE-2 is approximately 0.5 cm or 3/16 inch
in diameter. This cable must have a 50-ohm impedance rating and a stranded
center conductor. Although those specifications are met by RG-58 A/U and
RG-58 C/U cable, it is important to note that vendors sometimes use those
nomenclatures for cables with impedance ratings different from 50 ohms.
Thus, it is important to verify both the nomenclature and impedance rating of
the coaxial cable.
Coaxial
T
T
Cable
10BASE-T hub
AUI
cable
10BASE-T hub
UTP
WS
AUI
cable
UTP
WS
WS
WS
Legend:
T
= Transceiver
WS = Workstation
UTP = Unshielded twisted pair
Figure 3.7 Using 10BASE-5 as a backbone. The use of 10BASE-5 as a backbone enables 10BASE-T hubs to be located further from one another than if
they were wired together using unshielded twisted-pair cable.
80
chapter three
Under the 10BASE-2 standard, the maximum cable segment length is
reduced to 185 meters (607 feet), with a maximum of 30 stations per segment. This length limitation is often confused by persons that assume the 2 in
the network nomenclature allows a 200-meter cable length. Perhaps the IEEE
should have used the designator 10BASE-1.85, but it used 10BASE-2 even
though the cable length limit is 185 meters.
Unlike 10BASE-5 that has spacing rules for MAUs, there are none for
10BASE-2. However, because the minimum length of coaxial cable required
for a 10BASE-2 segment cannot be shorter than 0.5 meters (1.64 feet), this indirectly provides a minimum spacing between MAU connections of 0.5 meters.
Another difference between 10BASE-5 and 10BASE-2 concerns the integration of transceiver electronics into the network interface card under the
10BASE-2 standard. This permits the NIC to be directly cabled to the main
trunk cable. In fact, under 10BASE-2 the Thin Ethernet cable is routed directly
to each workstation location and routed through a BNC T-connector, one end
of which is pressed into the BNC connector built into the rear of the network
interface card.
The term BNC is considered by many persons as a mnemonic for ‘‘Bayonet
Nut Connector,’’ as you must push and turn one connector to mate with
another. The mating side of BNC connectors always has the same dimensions;
however, the side that connects to the cable can have different diameters
based upon the type of coaxial cable that the connector is designed to support
and the jacket (PVC, Teflon, etc.) of the cable.
Figure 3.8 illustrates the cabling of a one-segment 10BASE-2 network, which
can support a maximum of 30 nodes or stations. BNC barrel connectors can
be used to join two lengths of thin 10BASE-2 cable to form a cable segment,
as long as the joined cable does not exceed 185 meters in length. A BNC
terminator must be attached to each end of each 10BASE-2 cable segment.
One of the two terminators on each segment contains a ground wire that should
be connected to a ground source, such as the screw on an electrical outlet.
• •
Station 1
Station 2
Station 3
Station 30
Figure 3.8 Cabling a 10BASE-2 network. A 10BASE-2 cable segment cannot
exceed 185 meters and is limited to supporting up to 30 nodes or stations.
ethernet networks
81
The Thinnet Tap
The coaxial cable that forms the network bus must be routed directly into
each T-connector, which in turn is mated to a node or station. This means it is
not possible to locate a node away from the BNC T-connector by connecting
a single coaxial stub or drop cable between the T-connector and the node,
limiting the flexibility of this network unless you employ a thinnet tap.
Otherwise, the use of a single coaxial drop cable can result in the occurrence
of signal reflections that can generate transmission errors.
A thinnet tap was developed by vendors to enable individual nodes to be
located at a distance from the main 10BASE-2 network cable. This tap consists
of a tap assembly that connects to the main coaxial 10BASE-2 cable and a
drop cable with two coaxial cables and the equivalent of a BNC T-adapter.
When the drop cable is connected to a tap assembly it triggers a switch in
the tap, which electrically connects each side of the tap to one of the cables
in the drop cable. This action loops the signal from the top of the node and
back to the other side of the tap, resulting in an unbroken network, which
from an electrical perspective, is the same as connecting the node via a
conventional BNC T-connector. Figure 3.9 illustrates a schematic diagram of
a tap assembly.
Multifunction Transceivers
Because of the similarities between 10BASE-5 and 10BASE-2, several vendors introduced transceivers that support both thick and thin coaxial cable.
Figure 3.10 illustrates the Universal Ethernet Transceiver manufactured by
Tap assembly
Coaxial cable
Coaxial cable
Drop cable
(two cables)
Figure 3.9 A 10BASE-2 tap assembly. Through the use of a 10BASE-2 tap
assembly, it becomes possible to locate nodes at a distance from the backbone network.
82
chapter three
Figure 3.10 Universal Ethernet Transceiver. The Universal Ethernet
Transceiver manufactured by Transition Engineering of Edina, Minnesota,
supports both 10BASE-2 and 10BASE-5 coaxial cable connections.
(Photograph courtesy of Transition Engineering, Inc.)
Transition Engineering of Edina, Minnesota. The transceiver illustrated in the
upper left corner of Figure 3.10 is a 10BASE-2 transceiver. It can be converted
into a 10BASE-5 transceiver with an adapter that converts the BNC on the
thinnet transceiver into a vampire tap–type connector, as illustrated in the
lower right corner of Figure 3.10.
Grounding
Depending upon the type of coaxial-based Ethernet installed, you may require
the cable to be grounded. The 10BASE-5 specification indicates that the
coaxial cable should be grounded at one, and only one, point. In comparison,
the 10BASE-2 specification indicates that thin coaxial cable may be grounded
at one, and only one, point.
Regardless of the type of coaxial cable–based network you are using, it is
normally a good idea to ground the cable at one location. Doing so alleviates
the potential for static electricity to build up. In addition, many local electrical
codes require network cabling to be grounded at one location. When grounding
ethernet networks
83
a coaxial-based network, it is also important to ensure you do so at only one
location. Otherwise, multiple grounds on an Ethernet segment can result in a
risk to equipment, shock to people, and network errors.
You can ground a network segment through the use of a coaxial cable
terminator with ground wire, similar to the terminator illustrated in the lower
right portion of Figure 3.3. If you install a repeater on one end of a segment, it
is important to note that most repeaters can be set up to ground or not ground.
Thus, you may be able to use a repeater to ground a segment. It’s important
to note that a 10BASE-2 network will have a 50 ohm terminator at each end.
Only one of the terminators should be grounded. If you ground both, although
the network will not completely fail, this will result in an increased error rate
as well as slow data transfer on the network.
Expanding a 10BASE-2 Network
You can increase the number of stations that can be attached to a 10BASE-2
network, the total network distance, or both through the use of repeaters. As
with a 10BASE-5 network, between any two nodes there can be up to five
cable segments connected through four repeaters. Like a 10BASE-5 segment,
each end of every 10BASE-2 segment must be terminated with a 50-ohm
terminator, with one end of each segment grounded.
Figure 3.11 illustrates an expanded 10BASE-2 network consisting of three
populated cable segments; two segments are used to extend the network span
by simply connecting two repeaters. The latter segments are also commonly
referred to as interrepeater cable segments (IRCS). Each cable segment shown
in Figure 3.11 requires one MAU connection to a repeater. Because there
are a maximum of 30 stations or MAU connections per segment, this means
that segments connected by a repeater are limited to actually supporting
29 workstations.
Two-Tier Cabling You can further expand a 10BASE-2 network by establishing a two-tier method of cabling. Under this cabling concept, the upper
tier can consist of up to 30 repeaters, connected to one another through the
use of interrepeater cable segments. These form a backbone network without
direct station attachments, so populated segments on the backbone have to
be connected on each segment end to a repeater. The lower tier then consists of repeaters connected to populated cable segments, with the lower-tier
repeater cabled to the upper-tier repeater to obtain a connection to the backbone. Figure 3.12 illustrates the use of a two-tier cabling hierarchy that can
be used to extend a 10BASE-2 network. As in Figure 3.11, between any pair
of workstations there can be a maximum of five cable segments and four
84
chapter three
IRCS
R
R
R
S
IRCS
S
S
S
Legend:
IRSC = Interrepeater cable segment
S
R
= Stations
S
• •
S
Figure 3.11 Expanding a 10BASE-2 network. Up to five cable segments
connected through the use of four repeaters can be used to expand a
10BASE-2 network.
repeaters. Of the five cable segments, up to three can be populated, and each
populated segment is limited to a maximum of 30 connections, of which one
or two will be repeater connections depending upon where the cable segment
resides with respect to the interconnected network. In addition, the maximum
distance between any two stations is limited to 100 meters (328 feet), while
the minimum distance between BNC T-connectors is 0.5 meters (1.6 feet).
Advantages and Disadvantages
Similar to a 10BASE-5 network, the use of coaxial cable for a 10BASE-2
network makes it suitable for use in locations that require a high degree of
immunity from electrical noise. Two additional advantages associated with
10BASE-2 include the fact that it is easy to establish because devices are daisychained together, and it is relatively inexpensive. Disadvantages of 10BASE-2
are also similar to the disadvantages associated with 10BASE-5. That is, a
10BASE-2 network is difficult to reconfigure; a single device failure or cable
ethernet networks
85
D
Tier 1
(backbone)
C
R
B
R
A
R
R
Tier 2
(repeaters connected
to workstations)
R
S
• •
E
S
Legend:
R = Repeater
S = Workstation
S
• •
S
Figure 3.12 Two-tier thin Ethernet 10BASE-2 network. A two-tier network
permits expansion of a 10BASE-2 network by adding another level of repeaters
to the backbone. The segment between repeaters B, C, and D is considered to
be populated, because repeater C is counted as one connection on the segment.
break can bring down the entire network, making it fault-intolerant, and the
isolation of a failure can be very time-consuming due to the need to check
every device and cable section.
Normally 10BASE-2 is well-suited for small networks that are relatively
static with respect to node relocations. Similar to 10BASE-5, the use of a
10BASE-2 network is often used as a backbone technology for interconnecting
10BASE-T hubs that must be located at cable distances beyond the range of that
twisted-pair network. Although a 10BASE-2 backbone length is considerably
less than that of a 10BASE-5 backbone, thinnet cable is cheaper and easier
to install, making it popular for backbones that must span distances up to
185 meters per segment.
Combining 10BASE-5 and 10BASE-2 Networks
There are many situations in which a combination of thick and thin coaxial
cable can better satisfy your cabling requirements than the exclusive use of one
type of cable. For example, you may have several clusters of terminals located
86
chapter three
in different geographically separated areas within a large office complex.
Instead of connecting the two clusters through the use of several thin 10BASE2 cable segments and multiple repeaters, it may be easier to install one or more
thick 10BASE-5 cable segments to serve as a backbone network for attachment
to two 10BASE-2 cable segments. Thus, the ability of thick coaxial cable to
support a longer transmission distance can eliminate or reduce the number of
repeaters required in a combined network. In addition, a network built using
thin 10BASE-2 cable, which was originally designed for a network span under
925 meters, cannot cover greater distances without mixing network media.
Thus, the use of thick coaxial cable may provide the span of coverage not
obtainable from the exclusive use of thin coaxial cable.
When thick and thin coaxial cable are combined to form a single cable
segment, you must consider the length of each type of cable you anticipate
using to develop a combined media cable segment. If the segment is entirely
made up of thin cable, its maximum length is 607 feet, while a cable segment
made up of thick cable has a maximum length of 1640 feet. Thus, a combined
thin and thick cable segment should vary in length between 607 and 1640 feet.
If L represents the length of a combined media cable you wish to construct,
you can use the following equation to determine the maximum length of
thin cable:
1640 − L feet
thin cable =
3.28
For example, suppose you want to construct a cable segment 1400 feet long.
Then, the maximum length of thin 10BASE-2 cable you can use becomes:
1640 − 1400
= 73 feet
3.28
Thus, you could use 73 feet of 10BASE-2 cable and route that cable through
T-connectors, which are fastened to the BNC connectors on the rear of network
interface cards. Then, you could use thick cable for the remaining 1327 feet
of cable required in the cable segment.
Figure 3.13 shows an example of a combined 10BASE-5 and 10BASE-2
cable segment that connects two groups of stations located at opposite ends
of a building without using repeaters. At one end of the building, 10BASE-2
cable is used with BNC T-connectors to connect a large number of stations
clustered together. At the other end of the building, we will assume there are
only a few stations. Thus, we would use 10BASE-5 cable and connect each
station to the thick cable through the use of an MAU (transceiver) and an AUI
(transceiver cable).
ethernet networks
87
10BASE-2 cable
BNC
terminator
BNC
T-connectors
N-series jack
to BNC jack
S
• • •
S
10BASE-5 cable
Ground
wire
N-series
terminator
MAUs
S
AUI
cable
S
Legend:
MAU = Media access unit (transceiver)
AUI = Attachment unit interface (transceiver cable)
S = Station
Figure 3.13 Combined 10BASE-5 and 10BASE-2 cable segment. A combined
thick and thin coaxial segment can be used to economically connect a group
of stations clustered in one area of a building to a few stations located a
relatively long distance away from the cluster.
In examining Figure 3.13, note that the end of the thin cable is terminated
with a BNC terminator. The connection of thick and thin cable is accomplished through the use of an N-series jack to BNC jack, while the thick cable
is terminated with an N-series terminator that has a ground wire.
10BROAD-36
10BROAD-36 is the only broadband network based on the CSMA/CD access
protocol standardized by the IEEE. Unlike a baseband network, in which
Manchester-encoded signals are placed directly onto the cable, the 10BROAD36 standard requires the use of radio frequency (RF) modems. Those modems
modulate nonreturn to zero (NRZ)–encoded signals for transmission on one
channel at a specified frequency, and they demodulate received signals by
listening for tones on another channel at a different frequency.
88
chapter three
Cable
A 10BROAD-36 network is constructed with a 75-ohm coaxial cable, similar
to the cable used in modern cable television (CATV) systems. Under the IEEE
802.3 broadband standard, either single or dual cables can be used to construct
a network. If a single cable is used, the end of the cable (referred to as the
headend) must be terminated with a frequency translator. That translator
converts the signals received on one channel to the frequency assigned to
the other channel, retransmitting the signal at the new frequency. Because
the frequency band for a transmitted signal is below the frequency band
10BROAD-36 receivers scan, we say the frequency translator upconverts a
transmitted signal and retransmits it for reception by other stations on the
network. If two cables are used, the headend simply functions as a relay point,
transferring the signal received on one cable onto the second cable.
Advantages
A broadband transmission system has several advantages over a baseband
system. Two of the primary advantages of broadband are its ability to support
multiple transmissions occurring on independent frequency bands simultaneously, and its ability to support a tree structure topology carrying multiple
simultaneous transmissions. Using independent frequency bands, you can
establish several independent networks. In fact, each network can be used to
carry voice, data, and video over a common cable. As for topology, broadband
permits the use of a tree structure network, such as the structure shown in
Figure 3.14. In this example, the top of the tree would be the headend; it would
contain a frequency translator, which would regenerate signals received at
one frequency back onto the cable at another predefined frequency.
Disadvantages
Like the 10BASE-5 network, the 10BROAD-36 system uses an MAU. However,
this MAU is more intelligent than a 10BASE-5 MAU, as it is responsible
for modifying the 802.3 frame slightly for broadband transmission via a
modulator. These frame modifications include the scrambling of the preamble,
so that the modulation of the preamble does not result in a loss of clocking, and
the addition of a postamble to each frame. The latter assists in the detection
of the end-of-frame. Although a 10BASE-5 AUI cable can be connected to
a 10BROAD-36 MAU, a 10BROAD-36 MAU obviously cannot be used with
a 10BASE-5 network. However, the design specifications of the 10BROAD36 network enable this network to support the connection of CSMA/CD
ethernet networks
89
HE
Legend:
HE = Headend
Figure 3.14 Broadband tree topology support. The headend receives transmissions at one frequency and regenerates the received signals back onto the
network at another predefined frequency.
baseband equipment, and also provide a large degree of compatibility and
interoperability with baseband systems.
The higher noise immunity of 75-ohm coaxial cable permits a 10BROAD-36
network to span 3600 meters, making this medium ideal for linking buildings
on a campus. In addition, the ability of 10BROAD-36 to share channel space
on a 75-ohm coaxial cable permits organizations that have an existing CATV
system, such as one used for video security, to use that cable for part or all
of their broadband CSMA/CD network. Although these advantages can be
significant, the cost associated with more intelligent MAUs and RF modems
has limited the use of broadband primarily to campus environments that
require an expanded network span. In addition, the rapid development and
acceptance of 10BASE-T and later twisted-pair-based Fast Ethernet and the
decline in the cost of fiber cable to extend the span of 10BASE-T networks
have severely limited what many people once anticipated as a promising
future for 10BROAD-36 networks.
1BASE-5
The 1BASE-5 standard was based on AT&T’s low-cost CSMA/CD network,
known as StarLan. Thus, 1BASE-5 is commonly referred to as StarLan,
90
chapter three
although AT&T uses that term to refer to CSMA/CD networks operating
at both 1 and 10 Mbps using unshielded twisted-pair cable. The latter is
considered the predecessor to 10BASE-T.
The 1BASE-5 standard differs significantly from Ethernet and 10BASE-5
standards in its use of media and topology, and in its operating rate. The
1BASE-5 standard operates at 1 Mbps and uses UTP wiring in a star topology;
all stations are wired to a hub, which is known as a multiple-access unit
(MAU). To avoid confusion with the term media access unit, which also has
the same abbreviation, we will refer to this wiring concentrator as a hub.
Topology
Each station in a 1BASE-5 network contains an NIC, cabled via UTP on a
point-to-point basis to a hub port. The hub is responsible for repeating signals
and detecting collisions. As we will shortly note, this cabling configuration
represents a topology supported by two succeeding versions of Ethernet,
10BASE-T and 100BASE-T networks.
The maximum cabling distance from a station to a hub is 250 meters; up
to five hubs can be cascaded together to produce a maximum network span
of 2500 meters. The highest-level hub is known as the header hub, and it
is responsible for broadcasting news of collisions to all other hubs in the
network. These hubs, which are known as intermediate hubs, are responsible
for reporting all collisions to the header hub.
Usage
AT&T’s 1-Mbps StarLan network, along with other 1BASE-5 systems, initially
received a degree of acceptance for use in small organizations. However,
the introduction of 10BASE-T, which provided an operating rate ten times
that obtainable under 1BASE-5, severely limited the further acceptance of
1BASE-5 networks.
The growth in the acceptance of 10BASE-T resulted in economies of scale
in the manufacture of 10BASE-T hubs and network interface cards. This,
in turn, enabled vendors to match or exceed price cuts of other vendors.
One key result of this situation was the ability of end users to obtain a
low-cost, high-performance local area networking capability through the use
of 10BASE-T equipment. Needless to say, 10BASE-T became the preferred
network technology of both small and large organizations during the early
1990s, and it essentially replaced the use of 1BASE-5.
ethernet networks
91
10BASE-T
In the late 1980s, a committee of the IEEE recognized the requirement of
organizations for transmitting Ethernet at a 10-Mbps operating rate over lowcost and readily available unshielded twisted-pair cable. Although several
vendors had already introduced equipment that permitted Ethernet signaling
via UTP cabling, such equipment was based on proprietary designs and was
not interoperable. Thus, a new task of the IEEE was to develop a standard for
802.3 networks operating at 10 Mbps using UTP cable. The resulting standard
was approved by the IEEE as 802.3i in September 1990, and is more commonly
known as 10BASE-T, with the T referencing the use of twisted-pair wire.
UTP Use
The 10BASE-T standard supports an operating rate of 10 Mbps at a distance
of up to 100 meters (328 feet) over 100 ohm Unshielded Twisted-Pair (UTP)
cable without the use of a repeater. The UTP cable requires two pairs of
twisted wire. One pair is used for transmitting, while the other pair is used for
receiving. Each pair of wires is twisted together, and each twist is 180 degrees.
Any electromagnetic interference (EMI) or radio frequency interference (RFI)
is therefore received 180 degrees out of phase; this theoretically cancels out
EMI and RFI noise while leaving the network signal. In reality, the wire
between twists acts as an antenna and receives noise. This noise reception
resulted in a 100-meter cable limit, until repeaters were used to regenerate
the signal.
Because UTP cable previously installed in commercial buildings contains
either three or four wire pairs, the RJ-45 jack used with 10BASE-T has eightpin connectors. However, 10BASE-T uses two pairs of wires, and thus only
four pins are actually used. Table 3.2 compares the 10BASE-T pin numbers
with the RJ-45 jack numbers and indicates the signal names of the pins used
with 10BASE-T UTP cable.
Although 10BASE-T was designed to support transmission distances of up
to 100 meters under the Electronic Industry Association/Telecommunications
Industry Association (EIA/TIA) cabling standard (previously described in
Chapter 2), the cabling distance consists of three segments. The first segment,
which can be up to 90 meters in length, runs from a patch panel in a wiring
closet to a wall plate in an office. The second and third segments, which can
be up to a total of 10 meters in length, allow patch cables at each end of the
link while restricting signal loss on the total of three interconnected segments.
The actual cable standard defined by the EIA/TIA for 10BASE-T is based upon
the signaling rate of the network and not the type of network. Under EIA/TIA
92
chapter three
TABLE 3.2
10BASE-T Wiring Sequence
10BASE-T
Pin #
RJ-45
Jack #
10BASE-T
Signal Name
1
1
Transmit Data +
2
2
Transmit Data −
3
3
Receive Data +
—
4
Not used
—
5
Not used
6
6
Receive Data −
—
7
Not used
—
8
Not used
cable standards, which were described in Chapter 2, category 3 cable, which
provides support for a signaling rate up to 16 MHz, must be used in a 10BASET network. However, it is quite common for most organizations installing
10BASE-T today to use category 5 cabling, which supports transmission up
to 100 MHz. This allows an organization to upgrade to a higher-speed LAN
without having to replace their cabling infrastructure.
Link Integrity Test
Each version of Ethernet continuously monitors the received data path as a
mechanism for determining if the link is operating correctly. To ensure the
other station has a signal to receive even when there is no frame to receive, a
link integrity test signal is transmitted during periods where there is no actual
network traffic. Through the transmission of link test pulses a sufficient level
of activity will occur on the receive data path to provide the receiver with the
ability to check the integrity of the link.
Network Components
A 10BASE-T network can be constructed with network interface cards, UTP
cable, and one or more hubs. Each NIC is installed in the expansion slot of
a computer and wired on a point-to-point basis to a hub port. One exception
to the preceding is built-in Ethernet chipsets now commonly added to laptop
and notebook motherboards. When using such computers you only need to
wire them to a hub port to become a participant on a network. When all of the
ethernet networks
93
ports on a hub are used, one hub can be connected to another to expand the
network, resulting in a physical star, logical bus network structure.
The design of a 10BASE-T hub centric–based network provides a wiring
scheme that can tolerate wiring failures much better than a bus-based coaxial
cable network. For example, if a coaxial cable is broken at any point, the entire
network segment fails. In comparison, the star-wiring topology of a 10BASE-T
hub-based network eliminates the single point of failure of a common cable.
This is because the failure of a cable between a hub port and a workstation
will not affect other workstations. Although the hub is a central point of
failure, it is normally a very reliable device, and you can probably expect
numerous cable failures from everyday office activity to occur before a hub
failure occurring.
Network Interface Cards Most 10BASE-T network interface cards contain
multiple connectors, which enable the card to be used with different types of
802.3 networks. For example, the NIC illustrated in Figure 3.15 includes an
RJ-45 jack as well as BNC and DB-15 connectors. The RJ-45 jack supports the
direct attachment of the NIC to a 10BASE-T network, while the BNC connector
permits the NIC to be mated to a 10BASE-2 T-connector. The DB-15 connector
enables the NIC to be cabled to a transceiver, and is more commonly referred
to as the NIC’s attachment unit interface (AUI) port.
Depending on the type of connectors built into your NIC, you may have
several options to consider for connecting the NIC to a specific type of IEEE
802.3 network. For example, assume that your NIC is limited to supporting
a thick or thin coaxial cable connection. In that case, you can still use the
NIC to connect to a 10BASE-T network, using a transceiver similar to the one
illustrated in Figure 3.16.
Figure 3.16 illustrates a 10BASE-T Ethernet transceiver manufactured by
Transition Engineering, Inc. This transceiver allows you to connect an existing
thick or thin 10BASE-5 or 10BASE-2 NIC’s DB-15 AUI port to a 10BASE-T
network. To do this, you would cable the NIC’s DB-15 AUI port to the DB-15
connector at the left of the transceiver. Then, you would cable the transceiver
to a hub port using UTP.
Hub The wiring hub in a 10BASE-T network functions as a multiport
repeater: it receives, retimes, and regenerates signals received from any
attached station. The hub also functions as a filter: it discards severely
distorted frames.
All hubs that conform to IEEE 10BASE-T specifications perform a core set
of tasks in addition to receiving and regenerating signals. 10BASE-T hubs test
94
chapter three
Figure 3.15 Multiple-media network interface card. Some multiple-media
network interface cards, such as the one illustrated, support the direct attachment to UTP and thin coaxial cable, while the DB-15 connector permits the
card to be cabled to an MAU connected to thick coaxial cable.
Figure 3.16 Transition Engineering 10BASE-T transceiver. With a 10BASE-T
transceiver, you can connect existing thick or thin 10BASE-5 or 10BASE-2
NICs to a 10BASE-T network. (Photograph courtesy of Transition Engineering, Inc.)
ethernet networks
95
each port connection, detect and handle excessive collisions, and ignore data
that exceeds the maximum 802.3 frame size.
A 10BASE-T hub tests the integrity of the link from each hub port to
a connected station by transmitting a special signal to the station. If the
device doesn’t respond, the hub will automatically shut down the port and
may illuminate a status light-emitting diode (LED) to indicate the status of
each port.
Hubs monitor, record, and count consecutive collisions that occur on each
individual station link. Because an excessive number of consecutive collisions
will prevent data transfer on all of the attached links, hubs are required to
cut off or partition any link on which too many collisions have occurred.
This partitioning enables the remainder of the network to operate in situations where a faulty NIC transmits continuously. Although the IEEE 802.3
standard does not specify a maximum number of consecutive collisions,
the standard does specify that partitioning can be initiated only after 30 or
more consecutive collisions occur. Thus, some hub vendors initiate partitioning when 31 consecutive collisions occur, while other manufacturers use a
higher value.
Another operating function of 10BASE-T hubs is to ignore continuous data
transmissions in which the frame length exceeds the maximum length of
1518 bytes. That length does not include the 8 byte preamble, which results
in the maximum length of an Ethernet frame being 1518 bytes when inside
an adapter and 1528 bytes when flowing on the wire. Such excessive length
frames usually result from collisions and are referred to as jabbering. Most
hubs also partition a jabbering port, and some hubs include a jabber indicator
LED in addition to a partition status LED on each port.
Although a wiring hub is commonly referred to as a concentrator, this term
is not technically correct. A 10BASE-T wiring hub is a self-contained unit that
typically includes 8, 10, or 12 RJ-45 ports for direct connection to stations,
and a BNC and/or DB-15 AUI port for expanding the hub to other network
equipment. The BNC and AUI ports enable the 10BASE-T hub to be connected
to 10BASE-2 and 10BASE-5 networks, respectively. For the latter, the AUI
port is cabled to a 10BASE-5 MAU (transceiver), which is tapped into thick
10BASE-5 coaxial cable. One 10BASE-T hub can be connected to another with
a UTP link between RJ-45 ports on each hub.
Figure 3.17 illustrates the connectors on a typical 10BASE-T hub. On
some hubs, one RJ-45 jack is labeled uplink/downlink for use in cascading hubs, while other vendors permit any RJ-45 port to be used for
connecting hubs.
chapter three
96
RJ-45 modular
jacks for terminal
connections
1
2
3
4
5
6
7
8
Thinwire
BNC connector
9
10 11 12
Ethernet
8
15
AC power cord
connector
1
9
Standard thickwire
connector
Figure 3.17 10BASE-T hub connectors. In addition to 8, 10, or 12 RJ-45
modular jacks for terminal connections, most 10BASE-T hubs contain a BNC
and DB-15 port to permit attachment to thin and thick backbone networks.
5
3
7
1
6
2
8
4
Figure 3.18 A 10BASE-T single hub–
based network. A hub functions as a
multiport repeater, regenerating data
received on one port to all other ports.
Figure 3.18 illustrates the use of a 10BASE-T hub to form a single 10BASE-T
network segment. Note that the twisted-pair connections allow workstations
to be connected from different locations with respect to the hub, providing
more cabling flexibility than bus-based Ethernet networks.
In examining Figure 3.18 note that station 1 is shown transmitting data to
the hub. As previously mentioned, the hub functions as a repeater. Thus, it is
shown retransmitting data received on one port back to all stations connected
to other ports on the hub.
Wiring Requirements There are two types of wiring used with 10BASE-T
networks — straight-through and crossover. Each type of wiring is used to
support a different type of equipment connection, thus it is important to
understand where each type of wiring is used.
Straight-through Wiring The wiring used with UTP cable for connecting
a 10BASE-T hub to the network interface card consists of straight-through
ethernet networks
97
wiring. That is, each RJ-45 pin used by 10BASE-T would be routed directly
from the connector at one end of the cable to the RJ-45 connector at the
opposite end of the cable. Figure 3.19 indicates the 10BASE-T hub to NIC
wiring requirements.
Crossover Wiring When one hub is connected to another, a special cable
must be used. In that cable, the transmit and receive wire pairs have to be
crossed over; that is, the receive pair at one end (pins 1 and 2) must be
connected to the transmit pair (pins 3 and 6) at the other end of the cable
and vice versa. Figure 3.20 illustrates the routing of crossover wiring used to
connect two UTP hub ports to one another.
The only exception to using a crossover cable is when a special hub port
containing the crossover function is provided. On some hubs this type of
port is labeled crossover, and it enables a straight-through cable to be used to
cascade one hub to another. On other hubs an X is placed on the crossover,
functioning port as a label to indicate pin reversals.
Unlike a hub, a concentrator consists of a main housing into which modular
cards are inserted. Although some modular cards may appear to represent
hubs, and do indeed function as 10BASE-T hubs, the addition of other
modules permits the network to be easily expanded from one location and
allows additional features to be supported. For example, the insertion of a
fiber-optic interrepeater module permits concentrators to be interconnected
over relatively long distances of approximately 3 km.
HUB UTP
RJ-45 pin
1
2
3
NIC UTP
RJ-45 pin
Transmit data +
Transmit data −
Receive data +
1
2
3
4
Not used
4
5
Not used
5
6
Receive data −
6
7
Not used
7
8
Not used
8
Figure 3.19
10BASE-T hub to NIC wiring.
98
chapter three
UTP
RJ-45 pin
Signal
UTP
RJ-45 pin
Wire routing
Transmit data +
1
1
Transmit data −
2
2
Receive data +
3
3
Not used
4
4
Not used
5
5
Receive data −
6
6
Not used
7
7
Not used
8
8
Figure 3.20 A crossover cable is
required to connect hubs together.
Wiring Considerations
Because 10BASE-T uses RJ-45 jacks and telephone wire is commonly available,
we may be tempted to use a silver satin color patch cable to connect a
workstation to a hub port. Don’t! The reason you should stay away from silver
satin wire used to connect telephones to jacks is because they do not have
twisted pairs inside. The lack of twisted pair means that the cable becomes
a long antenna and can result in excessive crosstalk as transmit and receive
signals interface with one another. In addition, excessive crosstalk signals
can appear as simultaneous traffic, which can result in the false triggering
of collision detection circuitry. The result of this triggering will be ‘‘late
collisions,’’ a term used to indicate collisions that are detected too late in the
transmission of a frame to permit near immediate retransmission of the frame.
Instead, the loss of the frame will be detected by application software and
the resulting retransmission will be considerably slower. You can eliminate a
major cause of the previously described problems by restricting your use of
cable to twisted-pair rated for use for an applicable version of Ethernet. That
is, use cat 3 cable for 10BASE-T and cat 5 for 100BASE-T, while cat 5e or cat 6
should be used for Gigabit Ethernet.
Expanding a 10BASE-T Network
A 10BASE-T network can be expanded with additional hubs once the number
of stations serviced uses up the hub’s available terminal ports. In expanding
a 10BASE-T network, the wiring that joins each hub together is considered to
represent a cable segment, while each hub is considered as a repeater. Thus,
ethernet networks
99
under the 802.3 specification, no two stations can be separated by more than
four hubs connected together by five cable segments. In addition, two of the
five segments cannot be populated, which results in 10BASE-T adhering to
the previously described 5-4-3 rule.
Figure 3.21 illustrates the expansion of a 10BASE-T network through the use
of five hubs. Note that the connection between station A and station B traverses
five segments and four hubs, and does not violate IEEE 802.3 connection
rules. Because the maximum span of a 10BASE-T network is 100 meters per
segment times five segments, or 500 meters, it is quite common for 10BASET networks to use a 10BASE-5 or even a 10BASE-2 cable backbone. As in
Figure 3.12, in which a 10BASE-5 cable provided a backbone for a 10BASE2 network, either 10BASE-2 or 10BASE-5 cable can be used to support an
extension of a 10BASE-T network. In such situations, the AUI port of a
10BASE-T hub is connected to a MAU (transceiver) connected to the 10BASE5 (thick coaxial) cable, or the BNC connector is mated to a thin coaxial
cable’s T-connector. Another reason for using a thin or thick coaxial cable
for a backbone is that you can thereby avoid the four-hub cascading limit
of 10BASE-T. For example, if you cabled ten hubs to a 10BASE-5 coaxial
B
A
Legend:
= Stations
Figure 3.21 Expanding a 10BASE-T network. No two stations can be separated by more than four hubs in a 10BASE-T network.
100
chapter three
cable backbone, each station would traverse at most two hubs, and would
thus comply with the IEEE rules. In comparison, you could not connect more
than five hubs together in a conventional 10BASE-T network because of IEEE
802.3 rules.
Full-Duplex Operations
Because 10BASE-T wiring has separate transmission and receive signal paths
it can instantly detect a collision as data is transmitted. However, if no
collisions are possible, the ability to have simultaneously active signal paths
is wasted. Because the direct connection of stations to a switch port precludes
the possibility of a collision, 10BASE-T full duplex became possible in
a switch environment, a network topology that will be covered later in
this book.
3.3 Use of Fiber-Optic Technology
An addition to Ethernet, which allows transmission via pulses of light
instead of electrical current, is known as FOIRL, an acronym that stands
for fiber-optic repeater link. Although FOIRL is not an Ethernet network,
this specification governs the transmission of Ethernet across dual-fiber
cable — one fiber is used for the transmission of data in the form of light
pulses, while the second fiber is used for the reception of data in the form of
light pulses.
The fiber-optic repeater link (FOIRL) represents one of two fiber-optic specifications developed to support different types of Ethernet networks operating
at 10 Mbps. The second standard, referred to as 10BASE-F, is designed to
interoperate with older FOIRL-based equipment and provides backward compatibility with the older specification. Both standards provide immunity from
electrical hazards to include electromechanically generated noise and lightning, enabling networks to be extended through gasoline refineries and other
areas where metallic media would be banned.
The use of fiber permits you to support multiple Ethernet segments at
distances up to 2000 meters (6600 feet) from one another. At a remote location
connected through the use of fiber-optic technology, you can connect a single
station directly using a fiber transceiver, or you can connect a 10BASE-T or
fiber hub and support multiple stations. You would use the AUI port of the
hub to provide a connection via a standard transceiver to different types of
Ethernet networks, while you would use an optical transceiver to provide a
connection to the dual-fiber cable.
ethernet networks
101
FOIRL
The older IEEE 802.3 FOIRL standard enables a fiber link segment to be up to
1000 meters between two repeaters while the more recent 10BASE-F standard
doubles the distance to 2000 meters. The greater distance is, however, only
obtainable when 10BASE-F repeaters are used at both ends. When 10BASE-F
is mixed with FOIRL, the maximum length of the segment is reduced to
1000 meters.
Other differences between FOIRL and 10BASE-F include the names associated with different fiber-optic components associated with each network
technology and the optical transmit power levels, receive sensitivity, and
power loss associated with each optical technology. In our examination of
the use of 10-Mbps fiber-optic media, we will first turn our attention to the
use of FOIRL components and, when appropriate, discuss its similarities
and differences with respect to the more modern 10BASE-F standard. Once
we complete our examination of FOIRL, we will then turn our attention to
10BASE-F.
Optical Transceiver
The optical transceiver is common to both FOIRL and 10BASE-F, with only
the optical power transmit and receive levels differing between the two
specifications. An optical transceiver consists of a pulse-generating LED,
a photodetector, and associated transmit and receive circuitry. Transmit
circuitry turns the LED on and off to convert electrical voltages representing data into a series of light pulses for transmission over the fiber. The
photodetector recognizes received light pulses, while the receive circuitry
generates electrical voltages and shapes pulses to correspond to the received
light pulses.
Today, you can purchase a fiber network access unit (NAU), an optical
transceiver mounted on an adapter card for installation in the system unit of
a personal computer, for less than $150. A second type of optical transceiver
used on Ethernet networks is built into fiber hubs, the use of which will be
covered next in this section. This type of optical transmitter may be designed
to share the use of a common power source and circuitry, resulting in a
per-port hub cost usually less than the cost of a fiber adapter.
Fiber Hubs
Under the original FOIRL specification the use of a fiber-optic segment was
only applicable between repeaters. This meant that you could not actually
102
chapter three
connect a distant station directly to a port on a hub. While industry waited for
the development of 10BASE-F standards several vendors, including AT&T,
incorporated FOIRL capability into hubs to provide network customers with
additional capabilities while awaiting the ability to manufacture 10BASE-F
products once appropriate standards were promulgated. Thus, when we
discuss the operation of FOIRL ports in hubs, you should note that this actually references proprietary equipment, because that standard only governed
transmission on fiber between repeaters.
A fiber hub is a special type of hub that contains a number of FOIRL ports,
one AUI port, and usually one or more 10BASE-T ports. For example, AT&T’s
original StarLan fiber hub contained six FOIRL ports, one 10BASE-T port, and
one AUI port. In comparison, Transition Engineering’s Model 1050 fiber-optic
hub contains 12 FOIRL ports and one AUI interface port.
In effect, a fiber hub is a grouping of fiber NAUs and one or more 10BASE-T
and/or AUI ports. Thus, you can use a fiber hub to support several extendeddistance Ethernet connections, and then link those connections directly to a
10BASE-T network with a 10BASE-T port built into the fiber hub, or indirectly
to any type of Ethernet network with an AUI port built into the fiber hub.
A more modern 10BASE-F hub consists of a series of fiber link (10BASE-FL)
ports, with the key difference between 10BASE-F and FOIRL hubs being the
optical transmit power and optical receiver sensitivity supported by each port
on each hub.
Fiber Adapter
A third type of hardware product used with FOIRL is a fiber adapter. The
fiber adapter is a media conversion device that converts between twistedpair and fiber-optic cable. This device extends the transmission distance
between a wire hub and an attached workstation — or another 10BASE-T wire
hub — from 100 meters (328 feet) to 1000 meters. If a 10BASE-F fiber adapter
is used with a 10BASE-F-compliant hub port, the transmission distance is
extended to 2000 meters. For both FOIRL and 10BASE-F, unless the fiber
adapter is connected directly to a fiber hub, an adapter is required at each end
of an extended fiber link.
Wire and Fiber Distance Limits
Figure 3.22 illustrates the transmission distance limits associated with the use
of 10BASE-T and FOIRL-compliant fiber hubs and fiber adapters. Note that a
fiber network access unit installed in the system unit of a PC can communicate
ethernet networks
103
Wire closet
Fiber
hub
Fiber 2 km max
Fiber
adapter
Wire
hub
Wire 15 m
max
Wire
100m
max
Wire
hub
Wire 15 m
max
Wire 100 m max
UTP NAU
in PC
Wire
hub
Wire
100 m
max
Wire
hub
Fiber
adapter
Fiber
2 km max
Fiber NAU
in PC
Figure 3.22
10BASE-T and FOIRL transmission distance limits.
up to 1000 meters via fiber-optic cable, either directly into a fiber hub or to a
fiber adapter connected to a wire hub. Also note that in the upper right corner
of Figure 3.22, it was assumed that three 10BASE-T wire hubs were installed
in a wire closet. If that wire closet is located on a low floor in a building and
your organization leases another floor hundreds of feet above the wire closet
floor, the use of fiber-optic cable provides a mechanism to link two network
segments together. Similarly, the wire closet shown in the illustration could
be in one building on a university campus, with fiber used to connect that
building to another building. This use of fiber-optic cable not only extends
the transmission distance of the network, but also eliminates the possibility
of electromagnetic interference (EMI), because fiber cable is immune to EMI.
The distance limitation associated with the use of FOIRL has nothing to
do with the transmission constraints associated with optical signals. Instead,
the FOIRL distance limitation is related to the timing constraints in the IEEE
802.3 standard.
Because the FOIRL standard is limited to point-to-point connections, when
used to interconnect two 10BASE-2 networks you must consider the cable
length of each segment. For example, assume you have two 10BASE-2
networks located in different areas on a large factory floor, as illustrated
104
chapter three
OR
OR
Legend:
= Workstation
OR = Optical repeater
Figure 3.23
peater link.
Connecting 10BASE-2 networks using a fiber-optic interre-
in Figure 3.23. If segment A has a cable length of 300 meters and segment B has a cable length of 500 meters, what is the maximum FOIRL
cable length? Because the maximum cabling distance of a 10BASE-2 network is 2500 meters, subtracting the sum of the cable lengths from that
cable constraint (2500 − (300 + 500)) results in a maximum fiber-optic link of
1700 meters.
Another constraint you must consider is the Ethernet repeater limitation.
Ethernet limits any interconnected network to a maximum of four repeaters
between any two nodes. Thus, when considering the use of fiber-optic
repeaters, you must take into consideration the maximum network length
as well as the number of repeaters that can be placed between nodes. As we
will shortly note, under 10BASE-F a mechanism exists that allows the 5-4-3
rule to be exceeded.
10BASE-F
The development and promulgation of the IEEE 10BASE-F standard officially provided network users with the ability to use fiber-optic hubs. As
previously noted in the section covering the FOIRL specification, vendors
integrated FOIRL capability into hubs; however, the actual standard that
officially provided this capability was the 10BASE-F standard.
Under the 10BASE-F standard, which actually represents a set of fiberoptic media standards, three types of segments for optical transmission were
defined. Those segments include 10BASE-FL, 10BASE-FB, and 10BASE-FP,
each of which we will now examine.
10BASE-FL
The 10BASE-FL standard provides a fiber-optic link segment that can be
up to 2000 meters in length, providing that only 10BASE-FL equipment is
ethernet networks
105
used at both ends of the segment. Otherwise, the mixing of 10BASE-FL and
FOIRL equipment reduces the maximum length of the optical segment to
1000 meters.
The development of the 10BASE-FL standard was intended as a replacement for the older FOIRL specification. Unlike FOIRL that was restricted to
providing an optical connection between repeaters, 10BASE-FL enables an
optical segment to be routed between two workstations, two repeaters, or
between a workstation and a repeater port.
The actual connection of a copper-based network node to a 10BASE-FL segment can be accomplished in one of two ways. First, a stand-alone fiber-optic
MAU (FOMAU) can be connected via the 15-pin AUI connector on a network
adapter to provide an electrical-to-optical conversion capability. A second
method is to use a 10BASE-T/FL converter. The latter is used when only a
10BASE-T port is available on a NIC. Both the 10BASE-FL FOMAU and the
10BASE-T/FL converter include two fiber-optic connectors, one for transmitting and one for receiving data. Some 10BASE-T/FL converters include two
RJ-45 connectors that provide a high degree of cabling flexibility. One RJ-45
connector functions as a crossover cable and allows hub-to-hub communications via optical media, while the second connector is for workstations that
use a straight-through connection.
The top portion of Figure 3.24 illustrates the connection of a workstation to
a 10BASE-FL FOMAU via a 15-pin AUI connector. The lower portion of that
illustration shows the use of a 10BASE-T/FL converter. Both devices provide
you with the ability to transmit up to 2000 meters via a fiber link when a
10BASE-F-compliant optical device is at the other end of the optical link.
In addition to the use of a 10BASE-FL FOMAU and 10BASE-T/FL converter, other types of converters have been developed to extend a fiber-optic
transmission capability to other types of Ethernet networks. For example, a
10BASE-2/FL converter enables you to extend the transmission distance of a
10BASE-2 network via a fiber-optic connection.
The creation of a 10BASE-F optical hub is accomplished by the inclusion of
two or more FOMAUs in the hub. This results in the hub becoming an optical
repeater, which retransmits data received on one port onto all other ports.
Under the 10BASE-F standard you can use multiple 10BASE-FL connections
to connect several individual workstations at distances up to 2000 meters to a
common hub equipped with FOMAU ports.
Network Media
When examining the potential use of a converter or a FOMAU, it is important
to verify the type of optical media supported. Multimode fiber (MMF) is
106
chapter three
Ethernet
NIC
AUI cable
10BASE-FL TX
FOMAU RX
Optical
cable
(a) Using a fiber-optic MAU (FOMAU)
10BASE-T/FL
Ethernet
NIC
UTP cable
H
W
Optical
cable
(b) Using a 10BASE-T/FL converter
Figure 3.24
segment.
Options for connection of a workstation to a 10BASE-FL
commonly used, with the most popular type of fiber having a 62.5-micron (µ)
fiber-optic core and a 125-µ outer cladding. This type of multimode fiber
is referenced by the numerics 62.5/125 µ. The wavelength of light used on
a 62.5/125-µ MMF fiber link is 850 nanometers (nm), and the optical loss
budget, a term used to reference the amount of optical power lost through
attenuation, should not exceed the range 9.7 to 19.0 dB, with the exact amount
dependent upon the type of fiber used. Table 3.3 provides a comparison of the
optical attenuation for 10BASE-FL and FOIRL for six types of multimode fiber.
In examining the entries in Table 3.3, you will note that 10BASE-FL has a
higher loss budget than FOIRL for each type of multimode fiber. This explains
why the transmission distance of 10BASE-FL optical repeaters exceeds the
distance obtainable using FOIRL repeaters.
The connectors used on the previously described multimode fiber are
referred to as ST connectors. The ST connector represents a spring-loaded
bayonet connector whose outer ring locks onto the connection similar to
the manner by which BNC connector’s junction on 10BASE-2 segments. To
facilitate the connection process, an ST connector has a key on the inner
sleeve and an outer bayonet ring. To make a connection you would line up
the key on the inner sleeve of the ST plug with a slot on an ST receptacle,
push the connector inward and then twist the outer bayonet ring to lock the
ethernet networks
TABLE 3.3
and FOIRL
107
Comparing the Loss Budget (Optical Attenuation) of 10BASE-FL
Multimode, graded
index fiber size (µm)
50/125
50/125
50/125
62.5/125
Numerical aperture
.20
.21
.22
.275
83/125
.26
100/140
.30
10BASE-FL
Loss budget (dB)
9.7
9.2
9.6
13.5
15.7
19.0
FOIRL
Loss budget (dB)
7.2
6.7
7.1
11.0
13.2
16.5
connection in place. This action not only results in a tight connection but, in
addition, provides a precise alignment between the two portions of fiber-optic
cable being joined.
It is important to note that the optical loss depends upon several factors.
First, the length of the fiber governs optical loss, with a typical fiber illuminated
at 850 nm having a loss between 4 dB and 5 dB per 1000 meters. Second, these
of more connectors results in a higher optical loss. Third and perhaps most
important to note since this is easy to rectify, if your connectors or fiber splices
are poorly made or if dirt, finger oil or dust resides on connector ends, you
will obtain a higher level of optical loss than necessary.
10BASE-FB
A second 10BASE-F specification is 10BASE-FB, with the B used to denote
a synchronous signaling backbone segment. The 10BASE-FB specification
enables the limit on the number of repeaters previously described in the 5-4-3
rule section to be exceeded. A 10BASE-FB signaling repeater is commonly
used to connect repeater hubs together into a repeated backbone network
infrastructure that can span multiple 2000-m links.
10BASE-FP
A third 10BASE-F specification was developed to support the connection
of multiple stations via a common segment that can be up to 500 meters
in length. Referred to as 10BASE-FP, the P references the fact that the end
segment is a fiber-passive system. Under the 10BASE-FP specification a single
fiber-optic passive-star coupler can be used to connect up to 33 stations. Those
stations can be located up to 500 meters from a hub via the use of a shared
fiber segment.
108
chapter three
Although 10BASE-FP provides a clustering capability that might represent
a practical networking solution to the requirements of some organizations, as
well as reduce the cost associated with the use of multiple individual optical
cables, most organizations prefer to link hubs together. Doing so also allows
the use of a single pair of optical cables; however, the transmission distance
to the cluster is then extended to 2000 meters.
3.4 High-Speed Ethernet
To differentiate Gigabit and 10 Gbps Ethernet from other versions of Ethernet
whose operating rates exceed 10 Mbps but have a maximum operating rate
one-tenth that of Gigabit, those versions of Ethernet that operate at or below
100 Mbps were classified as high-speed Ethernet and are covered in this
section. This enables Gigabit and 10 Gbps Ethernet to be covered as a separate
entity in Section 3.5.
There are three broad categories into which high-speed Ethernet networks
fall. The first two types of high-speed Ethernet networks are represented by de
jure standards for operations at 100 Mbps, while the third standard represents
an extension to 10BASE-T that operates at 16 Mbps. As discussed in Chapter 2,
the IEEE standardized two general types of 100-Mbps Ethernet networks, with
the 802.3µ standard defining three types of 100-Mbps CSMA/CD networks,
while the 802.12 standard defines a demand-priority operation that replaces
the CSMA/CD access protocol. The third high-speed Ethernet network is
considered as a high-speed network only when compared with the operating
rate of Ethernet networks developed before 1992. This type of network, referred
to as isochronous Ethernet (isoENET), operates at 16 Mbps. This section will
focus upon obtaining a overview of the operation and use of each of these
three types of Ethernet networks.
Isochronous Ethernet
Isochronous Ethernet, or isoENET, represents an extension to 10BASE-T
technology. Isochronous Ethernet adds time-sensitive multimedia support
through an addition of 6.144 Mbps of isochronous bandwidth to existing 10Mbps 10BASE-T Ethernet. Here, the term isochronous references a series of
repetitive time slots used for the transmission of constant bit-rate services at
the physical bit-transmission level.
Although isoENET received a considerable degree of publicity during the
early 1990s, it never received any significant degree of commercial acceptance. This was probably due to the development of Fast Ethernet, which
ethernet networks
109
provided over six times the transmission capacity of isoENET. However,
because isoENET provided time slots for the transmission of time-sensitive
information, its design in effect provided a Quality of Service (QoS) capability
that is worth examining.
Isochronous Ethernet dates to 1992, when National Semiconductor and
IBM, with support from Apple Computer, submitted the basics of isoENET
to the IEEE 802.9 Integrated Services Local Area Networks working group.
Better known by its trade names isoEthernet and isoENET, this technique was
standardized by the IEEE as standard 802.9a, with the official designation
Integrated Service Local Area Network (ISLAN16-T). Here, the T in the abbreviation denotes its capability to operate over twisted-pair wiring, while the
16 references its operating rate. In comparison with other Ethernet LANs that
are asynchronous, isoENET was developed to support the 8-KHz sampling
clock used as a worldwide standard for voice transmission. This synchronization capability was layered on top of the 10-Mbps 10BASE-T operating
rate, enabling isoENET to support real-time communications in addition to
conventional 10BASE-T asynchronous LAN transmission.
Isochronous Ethernet represented a hybrid type of Ethernet network, combining standard 10-Mbps 802.3 10BASE-T with a 6.144-Mbps isochronous
networking capability. The 6.144 Mbps of additional bandwidth was designed
to accommodate 96 integrated services digital network (ISDN) B-channels,
either individually or in multiple combinations of N × 64 Kbps. For example,
a videoconference requiring 128 Kbps of bandwidth would be assigned two 64Kbps channels, while digitized voice that requires 64 Kbps when pulse code
modulation (PCM) is used for digitization would be assigned one channel.
Besides being designed to support 96 ISDN B-channels, the isochronous
bandwidth supported one 64-Kbps ISDN D-channel for signaling and one
96-Kbps ISDN M-channel used to support ISDN maintenance functions.
Figure 3.25 illustrates the allocation of isoENET bandwidth.
IsoENET replaced the Manchester encoding used by 10BASE-T with a 4B/5B
encoding scheme, which represents the data encoding method used by the
ANSI X3T9.5 FDDI standard. Under 4B/5B coding, each octet of data is split
into two four-bit nibbles (4B). Each nibble is then coded using five bits (5B),
resulting in an 80-percent level of utilization of the 20-MHz IEEE 802.3 clock
signal. In comparison, Manchester encoding provides a 50-percent utilization
of the 20-MHz clock. The change in the method of data coding provided an
additional 6.144-Mbps bandwidth on existing 10BASE-T wiring, connector,
and hub facilities. However, the use of the additional bandwidth required
the installation of an isoENET hub and isoENET adapter cards for each local
area network node that requires an isochronous communications capability.
110
chapter three
10 Mbps ethernet
16 Mbps
operating
rate
Figure 3.25
6 Mbps (96 ISDN B-channels)
64 Kbps D-channel
Signaling
96 Kbps M-channel
Maintenance
Allocation of isoENET bandwidth.
Users who did not require an isochronous communications capability could
use their existing 10BASE-T adapter cards, and 802.3 traffic would not notice
any change to the operation of a 10BASE-T network. Figure 3.26 illustrates
how an isoENET hub could support conventional 10BASE-T and isoENET
network nodes.
Although at one time about a dozen vendors manufactured isoENET products and its use provides a mechanism to extend multimedia to the desktop,
other Ethernet technologies dulled the demand for its 16-Mbps communications capability. The introduction of 100BASE-T and Gigabit Ethernet appears
to resolve the bandwidth crunch experienced by many networks that added
Internet connections and graphics-intensive applications. Because it appears
that greater bandwidth was more important than obtaining a videoconferencing capability to the desktop, a majority of vendor and customer interest
became focused upon faster Ethernet solutions than that provided by isoENET.
Multimedia
PC
Workstation
isoENET
Ethernet
Video
server
isoENET
Figure 3.26 isoENET supports the addition of 6.155 Mbps to nodes equipped
with isoENET adapter cards.
ethernet networks
111
Fast Ethernet
Fast Ethernet is not actually a local area network, but a term commonly used
to reference a series of three 100-Mbps physical-layer LAN specifications
in the IEEE 802.3µ addendum. Those specifications include 100BASE-TX,
100BASE-FX, and 100BASE-T4. Each specification maintains the use of the
MAC protocol used by earlier Ethernet/IEEE 802.3 standards, CSMA/CD.
100BASE-T specifies 100-Mbps operations using the CSMA/CD protocol
over two pairs of category 5 UTP cable. 100BASE-FX changes the LAN transport media to two pairs of fiber, while 100BASE-T4 supports four pairs of
category 3, 4, and 5 UTP or STP cable. Table 3.4 provides a summary of the
three types of Fast Ethernet with respect to their IEEE media specification
designation, types of media supported, types of connectors supported, and
the coding scheme used.
100BASE-T Overview
At the beginning of this section we noted that the IEEE standardized two types
of 100 Mbps Ethernet networks. The 802.3µ standard defines three types of
CSMA/CD operations and is referred to as Fast Ethernet or 100BASE-T.
A second 100 Mbps Ethernet network standard resulted in the development
of a different access control mechanism, referred to as a demand priority
mechanism. The IEEE standardized the 100 Mbps LAN using a demand
priority mechanism as the 802.12 standard and its support of either Ethernet
or Token-Ring resulted in the name 100VG-AnyLAN being used to reference
the standard. Although both standards received a considerable degree of
interest, actual implementation and commercial success is another story. Of
TABLE 3.4
Fast Ethernet Functionality
IEEE Media
Specifications
Cable
Support
Connector
Support
Coding
Scheme
100BASE-TX
Category 5 UTP (2-pair wire)
RJ-45
4B/5B
100-ohm STP (2-pair wire)
DB-9
100BASE-FX
62.5/125-micron fiber-optic cable
(2 multimode fibers)
SC or ST
4B/5B
100BASE-T4
Category 3, 4, or 5 UTP (4-pair wire)
RJ-45
8B6T
Legend: UTP, unshielded twisted pair; STP, shielded twisted pair.
112
chapter three
the two standards 100BASE-T is by far a commercial success while 100VGAnyLAN is anything but. Thus, the primary focus of our attention in the
remaining of this chapter will be upon 100BASE-T, although we will briefly
conclude this section with an overview of the technology associated with
100VG-AnyLAN.
The standardization of 100BASE-T required an extension of previously
developed IEEE 802.3 standards. In the definition process of standardization development, both the Ethernet media access control (MAC) and
physical layer required adjustments to permit 100-Mbps operational support. For the MAC layer, scaling its speed to 100 Mbps from the 10BASE-T
10-Mbps operational rate required a minimal adjustment, because in theory the 10BASE-T MAC layer was developed independently of the data
rate. For the physical layer, more than a minor adjustment was required,
because Fast Ethernet was designed to support three types of media. Using
work developed in the standardization process of FDDI in defining 125Mbps full-duplex signaling to accommodate optical fiber, UTP, and STP
through physical media-dependent (PMD) sublayers, Fast Ethernet borrowed
this strategy. Because a mechanism was required to map the PMD’s continuous signaling system to the start-stop half-duplex system used at the Ethernet
MAC layer, the physical layer was subdivided. This subdivision is illustrated
in Figure 3.27.
Note that the version of Fast Ethernet that operates using four pairs of
telephone-grade twisted pair wire is known as 100BASE-T4, while 100BASETX operates over two pairs of data-grade twisted-pair. The third version of
Fast Ethernet, which operates over fiber-optic media, is 100BASE-FX. The
Link
layer
100-Mbps
ethernet MAC
Convergence
sublayer (CS)
Physical
layer
Physical media dependent
sublayer (PMD)
Four-pair
unshielded
twisted pair
Figure 3.27
Two-pair
Data-grade
twisted pair
Fiber
optic
Fast Ethernet physical layering subdivision overview.
ethernet networks
113
PMD sublayer supports the appropriate media to be used, while the convergence sublayer (CS), which was later renamed the physical coding sublayer,
performs the mapping between the PMD and the Ethernet MAC layer.
Although Fast Ethernet represents a tenfold increase in the LAN operating
rate from 10BASE-T to ensure proper collision detection, the 100BASE-T
network span was reduced to 200 meters, with a maximum of 100 meters
permitted between a network node and a hub. The smaller network diameter
reduces potential propagation delay. When coupled with a tenfold operating
rate increase and no change in network frame size, the ratio of frame duration
to network propagation delay for a 100BASE-T network is the same as for a
10BASE-T network.
In addition to reducing the 100BASE-T network span to 200 meters, the Fast
Ethernet specification recognized the need to provide an automatic capability
for equipment to support older 10BASE-T equipment. This automatic support
is in the form of an Auto-Negotiation mechanism referred to as Nway, which
will be covered later in this section.
Physical Layer
The physical layer subdivision previously illustrated in Figure 3.27, as indicated in the title of the figure, presents an overview of the true layer
subdivision. In actuality, a number of changes were required at the physical layer to obtain a 10-Mbps operating rate. Those changes include the
use of three wire pairs for data (the fourth is used for collision detection),
8B6T ternary coding (for 100BASE-T4) instead of Manchester coding, and an
increase in the clock signaling speed from 20 MHz to 25 MHz. As indicated
in Table 3.5, in comparison to 10BASE-T the differences at the physical layer
resulted in a tenfold increase in the 100BASE-T operating rate.
When the specifications for Fast Ethernet were being developed, it was recognized that the physical signaling layer would incorporate medium-dependent
functions if support was extended to two-pair cable (100BASE-TX) operations.
TABLE 3.5 100BASE-T System Throughput Compared with 10BASE-T
Transmit on 3 pairs vs. 1 pair
×3.00
8B6T coding instead of Manchester
×2.65
20 to 25 MHz clock increase
×1.25
Total throughput increase
10.00
114
chapter three
To separate medium-dependent interfaces to accommodate multiple physical
layers, a common interface referred to as the medium-independent interface (MII) was inserted between the MAC layer and the physical encoding
sublayer. The MII represents a common point of interoperability between the
medium and the MAC layer. The MII can support two specific data rates,
10 Mbps and 100 Mbps, permitting older 10BASE-T nodes to be supported
at Fast Ethernet hubs. To reconcile the MII signal with the MAC signal, a
reconciliation sublayer was added under the MAC layer, resulting in the
subdivision of the link layer into three parts — a logical link control layer,
a media access control layer, and a reconciliation layer. The top portion of
Figure 3.28 illustrates this subdivision.
That portion of Fast Ethernet below the MII, which is the new physical
layer, is now subdivided into four sublayers. The lower portion of Figure 3.28
illustrates the physical sublayers for 100BASE-T4 and 100BASE-TX.
The physical coding sublayer performs the data encoding, transmit, receive,
and carrier sense functions. Because the data coding method differs between
100BASE-T4 and 100BASE-TX, this difference requires distinct physical coding sublayers for each version of Fast Ethernet.
The physical medium attachment (PMA) sublayer maps messages from the
physical coding sublayer (PCS) onto the twisted-pair transmission media, and
vice versa.
The auto-negotiation block shown in Figure 3.28 is not actually a layer
but a function built into 100BASE-T4 and 100BASE-TX. As noted earlier in
this section, auto-negotiation provides 100BASE-T copper media ports and
adapters with the ability to automatically adjust to 10 or 100 Mbps operations.
The medium-dependent interface (MDI) sublayer specifies the use of a
standard RJ-45 connector. Although the same connector is used for 100BASETX, the use of two pairs of cable instead of four results in different pin
assignments.
100BASE-T4
100BASE-T4 supports a 100-Mbps operating rate over four pairs of category 3, 4, or 5 UTP wiring that supports a segment upto 100 meters in
length. Figure 3.29 illustrates the RJ-45 pin assignments of wire pairs used
by 100BASE-T4. Note that wire pairs D1 and D2 are unidirectional. As indicated in Figure 3.29, three wire pairs are available for data transmission
and reception in each direction, while the fourth pair is used for collision
detection. Each wire pair is polarized, with one wire of the pair transporting
a positive (+) signal while the other transports a negative (−) signal. Thus,
ethernet networks
ISO
model
100BASE-T4 and 100BASE-TX
Data
link
layer
Logical link control (LLC)
Media access control (MAC)
Reconciliation sublayer (RS)
Medium-independent interface
Physical
layer
Figure 3.28
Physical coding
sublayer
8B6T coding
Physical coding
sublayer
4B5B coding
(previously convergence layer)
Physical medium
attachment
(PMA)
Physical medium
attachment
(PMA)
Auto
negotiation
Auto
negotiation
Medium
dependent
interface
Medium
dependent
interface
Medium
(4-pair)
Medium
(2-pair)
100BASE-T4
100BASE-TX
100BASE-T4 versus 100BASE-TX physical and link layers.
Node
1
Node detects
collision on
channel D2
Hub
Transmit D1
1
2
2
3
3
4
Receive D2
7
8
Figure 3.29
4
5
5
6
Hub detects
collision on
channel D1
Bidirectional D3
Bidirectional D4
6
7
8
100BASE-T4 pin assignments.
115
116
chapter three
another common mechanism used to denote the function of the pins in the
100BASE-T4 eight-pin connector is to denote their use (transmit TX, receive
RX, or bi-directional BI), the data pair (D1, D2, D3 or D4) and type of signal
(+ or −) transported. Based upon the preceding, Figure 3.30 illustrates the
signals transported by the 100BASE-T4 eight-in connector.
In examining the signals based upon pin position shown in Figure 3.30,
you can note that the interconnection of devices connected to a 100BASET4 segment would require a crossover of signals. Otherwise, for example,
TX− D1+ would be cabled to TX− D1+ and the transmission at one end
would not be placed onto the receive side at the other end. Thus, a signal crossover cable is normally required to interconnect a station to a hub
port. An exception to this is when the hub has the crossover function built
into the device, which is usually denoted by marketing on the port in
the form of a tilted X. Figure 3.31 illustrates the crossover cable wiring for
100BASE-T4.
In examining the 100BASE-T4 crossover cable note that the TX pair at one
end of a cable is connected or strapped to the RX pair at the other end of
the cable. In addition, the BI− D3 pair at one end is connected to the BI− D4
pair at the other end. Each of the cross-connections occurs in each direction
to provide a crossover cable. Because 100BASE-T4 can operate on category 3,
which is used for 10BASE-T, this feature enables many organizations to
migrate to a 100-Mbps network without changing their wiring infrastructure.
The 100BASE-T4 physical coding sublayer implements 8B6T block coding.
Under this coding technique, each block of eight input bits is transformed into
a unique code group of six ternary symbols. Figure 3.32 provides an overview
of the 8B6T coding process used by 100BASE-T4.
Figure 3.30 Signals transported on the 100BASE-T4
eight-pin connector.
ethernet networks
Pin
number
Signal
Signal
Pin
number
1
2
3
6
4
5
7
8
TX_D1+
TX_D1−
RX_D2+
RX_D2−
BI_D3+
BI_D3−
BI_D4+
BI_D4−
TX_D1+
TX_D1−
RX_D2+
RX_D2−
BI_D3+
BI_D3−
BI_D4+
BI_D4−
1
2
3
6
4
5
7
8
Figure 3.31
Input data stream
Output code groups
117
100BASE-T4 crossover wiring.
12345678
123456
Figure 3.32
12345678
123456
12345678
123456
• • •
8B6T coding process.
The output code groups resulting from 8B6T coding flow out to three parallel
channels that are placed on three twisted pairs. Thus, the effective data rate
on each pair is 100 Mbps/3, or 33.33 Mbps. Because 6 bits are represented
by 8 bit positions, the signaling rate or baud rate on each cable pair becomes
33 Mbps × 6/8, or 25 MHz, which is the clock rate used at the MII sublayer.
100BASE-TX
100BASE-TX represents the use of two pairs of category 5 UTP cabling with
RJ-45 connectors or two pairs of category 5 STP cable terminated with the
common DB-9 communications connector used on the serial port of many
notebook computers.
Because UTP is relatively inexpensive and two pair is less costly than
four pair, 100BASE-TX represents a more popular version of Fast Ethernet
than 100BASE-T4. The UTP wire actually used must meet the EIA/TIA 568
category 5 wire specifications. A 100BASE-TX network requires a hub, and
the maximum cable run is 100 meters from hub port to node, with a maximum
network diameter of 200 meters.
Figure 3.33 illustrates the cabling of two pairs of UTP wires between a hub
and node to support 100BASE-TX transmission. One pair of wires is used for
118
chapter three
Hub port
Network node
Pair 1
Pair 1
Pair 2
Pair 2
Figure 3.33
100BASE-TX cabling.
transmission, while the second pair is used for collision detection and reception of data. The use of a 125-MHz frequency requires the use of a data grade
cable. Thus, 100BASE-TX is based upon the use of category 5 UTP and STP.
Similar to 100BASE-T4, 100BASE-TX uses an eight-pin connector when
UTP is employed. Figure 3.34 illustrates the use of each of the eight pins. Note
that the transmit and receive data signals on each pair transport positive (+)
and negative (−) signals. Also note that 100BASE-TX transmission via UTP
uses the same pin assignments as 10BASE-T hardware, enabling the same
category 5 wiring system to be used without any change to wiring being
required. This provides network administrators with a considerable degree
of flexibility for upgrading from 10BASE-T to 100BASE-TX once category 5
cabling is installed.
Similar to our discussion concerning 100BASE-TX, when two stations are
connected over a segment the transmit data pins on one side must be wired
to the receive data pins on the distant end and vice versa. Thus, unless a hub
has a port with a built-in crossover you need to use a 100BASE-TX crossover
cable to connect a station to a hub. Figure 3.35 illustrates the connection of
pins required for a 100BASE-TX crossover cable.
Similar to 10BASE-T, 100BASE-TX has separate transmit and receive signal
paths. Because these paths can be simultaneously active 100BASE-TX can
support full duplex transmission. If STP cabling is used the connector will be
Figure 3.34
with UTP.
100BASE-TX eight pin connector used
ethernet networks
Pin
number
Signal
Signal
Pin
number
1
2
3
6
TD1+
TD1−
RD2+
RD2−
TD1+
TD1−
RD2+
RD2−
1
2
3
6
Figure 3.35
119
100BASE-TX crossover cable.
Figure 3.36
STP use.
100BASE-TX nine-pin D-connector for
a nine-pin ‘‘D-type’’ connector similar to the ones used on the serial port of
notebook computers. Figure 3.36 illustrates the use of the pins on the 9-pin
D-connector.
Although the 100BASE-TX physical layer structure resembles the 100BASET4 layer, there are significant differences between the two to accommodate
the differences in media used. At the physical coding sublayer, the 100-Mbps
start-stop bit stream from the MII is first converted to a full-duplex 125-Mbps
bit stream. This conversion is accomplished by the use of the FDDI PMD
as the 100BASE-TX PMD. Next, the data stream is encoded using a 4B5B
coding scheme. The 100BASE-TX PMD decodes symbols from the 125-Mbps
continuous bit stream and converts the stream to 100-Mbps start-stop data bits
when the data flow is reversed.
4B5B Coding
The use of a 4B5B coding scheme enables data and control information to
be carried in each symbol represented by a 5-bit code group. In addition,
120
chapter three
an inter-Stream fill code (IDLE) is defined, as well as a symbol used to
force signaling errors. Because 4 data bits are mapped into a 5-bit code, only
16 symbols are required to represent data. The remaining symbols not used
for control or to denote an IDLE condition are not used by 100BASE-TX and
are considered as invalid.
Table 3.6 lists the 4B5B 100BASE-TX code groups. Because an explanation
of the use of the control codes and IDLE code requires an examination of the
MAC frame, we will defer our discussion of those symbols until Chapter 4, as
frame formats are discussed in that chapter.
100BASE-FX
100BASE-FX represents the third 100BASE-T wiring scheme, defining Fast
Ethernet transmission over fiber-optic media. 100BASE-FX requires the use
of two-strand 62.5/125-micron multimode fiber media and supports the 4B5B
coding scheme, identical to the one used by 100BASE-TX.
100BASE-FX provides an extended network diameter of up to 400 meters.
However, in a mixed 100BASE-T4 and 100BASE-FX environment, the
collision domain should not exceed 231 meters, consisting of 100 meters
for 100BASE-T4 and 131 meters for 100BASE-FX. Concerning connectors,
100BASE-FX supports both ST and SC fiber connectors, which were originally
defined for FDDI.
Network Construction and Use
Similar to 10-Mbps Ethernet networks, you can use adapter cards, cable,
and hubs to construct a variety of Fast Ethernet networks to satisfy different
organizational network requirements. However, when doing so, it is important
to understand the role of repeaters in a Fast Ethernet environment, as the
standard defines two types, each having a different effect upon network
construction.
Repeater Rules
When we examined Ethernet we discussed the 5-4-3 rule, which applied to a
common type of repeater. When we discuss Fast Ethernet we must consider
two types of repeaters, referred to as Class I and Class II repeaters.
ethernet networks
TABLE 3.6
4B/5B Code Groups
PCS Code Group
4 3 2 1 0
121
MII (TXD/RXD)
Name
3 2 1 0
Interpretation
1 1 1 1 0
0
0 0 0 0
Data 0
0 1 0 0 1
1
0 0 0 1
Data 1
1 0 1 0 0
2
0 0 1 0
Data 2
1 0 1 0 1
3
0 0 1 1
Data 3
0 1 0 1 0
4
0 1 0 0
Data 4
0 1 0 1 1
5
0 1 0 1
Data 5
0 1 1 1 0
6
0 1 1 0
Data 6
0 1 1 1 1
7
0 1 1 1
Data 7
1 0 0 1 0
8
1 0 0 0
Data 8
1 0 0 1 1
9
1 0 0 1
Data 9
1 0 1 1 0
A
1 0 1 0
Data A
1 0 1 1 1
B
1 0 1 1
Data B
1 1 0 1 0
C
1 1 0 0
Data C
1 1 0 1 1
D
1 1 0 1
Data D
1 1 1 0 0
E
1 1 1 0
Data E
1 1 1 0 1
F
1 1 1 1
Data F
DATA
IDLE
1 1 1 1 1
I
IDLE:
Used as inter-Stream fill code
1 1 0 0 0
J
Start-of-stream delimiter, part 1 of 2;
always used in pairs with K.
1 0 0 0 1
K
Start-of-stream delimiter, part 2 of 2; always
used in pairs with J.
CONTROL
(continued overleaf )
122
chapter three
TABLE 3.6
(Continued)
PCS Code Group
MII (TXD/RXD)
4 3 2 1 0
Name
3 2 1 0
Interpretation
0 1 1 0 1
T
End-of-stream delimiter, part 1 of 2;
always used in pairs with R.
0 0 1 1 1
R
End-of-stream delimiter, part 2 of 2;
always used in pairs with T.
0 0 1 0 0
H
Transmit error; used to force signaling errors.
0 0 0 0 0
V
Invalid code
0 0 0 0 1
V
Invalid code
0 0 0 1 0
V
Invalid code
0 0 0 1 1
V
Invalid code
0 0 1 0 1
V
Invalid code
0 0 1 1 0
V
Invalid code
0 1 0 0 0
V
Invalid code
0 1 1 0 0
V
Invalid code
1 0 0 0 0
V
Invalid code
1 1 0 0 1
V
Invalid code
INVALID
A Class I repeater has a greater budget for timing delay, enabling it to be used
to support dissimilar physical media segments that use dissimilar signaling
such as 100BASE-T4 and 100BASE-TX. This support is possible because the
greater timing budget of the Class I repeater enables it to translate the line
signal received on one port to a different type of signal for transmission
onto other ports. Only one Class I repeater can be installed in a segment. In
comparison, a Class II repeater has a lower budget for timing delay, resulting
in it being faster than a Class I repeater. This enables up to two Class II
repeaters to be used in a segment; however, in doing so the interrepeater cable
length is limited to 5 meters when data terminal equipment are 100 meters
from repeaters. Because a Class II repeater immediately repeats an incoming
signal onto all other ports other than the port data was received on without
a translation process being possible, it can only be used to interconnect
ethernet networks
123
segment types that use the same signaling method, such as 100BASE-TX and
100BASE-FX.
The actual span distance obtainable through the use of repeaters depends
upon the type of repeater used and the media cable. Figure 3.37 illustrates
the cable restrictions associated with Fast Ethernet. In examining the entries
in Figure 3.37, note that the repeater references a Fast Ethernet hub, because
the hub receives information on one port and rebroadcasts the received data
onto all other ports. To exceed the cable limits shown in Figure 3.37, you
must connect the workstation or hub port to a switch, bridge, or router,
which results in a new segment being established. Concerning those cable
limits, although all vendors support the first two cabling distances shown
in Figure 3.37a and 3.37b, there are minor differences in the construction
of Class II–type repeaters between some vendors that may reduce the span
distance from that shown in Figure 3.37c for a mixture of TX and FX or FX
usage. In addition, the use of two Class II repeaters manufactured by different
vendors can also result in a slight reduction from the span distances shown
Direction connection without repeaters
(TX = 100 m, TX & FX = NA, FL1111 = 412m)
(a)
One class I repeater
(TX = 200 m, TX & FX = 260.8m, FX = 272m)
Class I
Repeater
(b)
One class II repeater
(TX = 200 m, TX & FX = 308.8m, FX = 320m)
Class II
Repeater
(c)
Two class II repeaters
(TX = 205 m, TX & FX = 223m, FX = 228m)
Class II
5m
Repeater
Class II
Repeater
(d)
Figure 3.37
strictions.
Fast Ethernet cable re-
124
chapter three
in Figure 3.37d for a mixture of TX and FX and FX repeater use. Thus, it
is highly recommended to check vendor specification sheets concerning the
network guidelines associated with a particular type of repeater.
Because 100BASE-T4 and 100BASE-TX preserve the 10BASE-T MAC layer,
both standards are capable of interoperability with existing 10BASE-T networks as well as with other low-speed Ethernet technology. Through the use
of NWay autosensing logic, Fast Ethernet adapters, hub, and switch ports can
determine if attached equipment can transmit at 10 or 100 Mbps and adjust to
the operating rate of the distant device.
Autonegotiation
Auto-negotiation represents an optical feature which was added to the metallic copper versions of Fast Ethernet and which is now incorporated into most
types of such hardware. The basis for auto-negotiation can be traced to a
technology referred to as NWay. NWay represents a cable and transmission
autosensing scheme proposed by National Semiconductor to the IEEE 802.3
standards group in May 1994. NWay incorporated an autosensing scheme to
permit Ethernet circuits to detect both the cable type and speed of incoming
Ethernet data, as well as enabled Ethernet repeaters to configure themselves
for correct network operations. Because NWay can detect 10-Mbps versus
100-Mbps operations, as well as half- and full-duplex transmission, it formed
the basis for autonegotiation which permits Ethernet circuits to be developed to automatically adjust to the operating rate and cabling scheme used.
This in turn simplifies the efforts of network managers and administrators,
because products incorporating autonegotiation are self-configurable and do
not require the setting of DIP switches or software parameters.
Auto-negotiation relies upon a modification to the link integrity test used
by 10BASE-T, replacing the link test pulse with a burst of pulses referred to as
fast link pulses (FLPs). Each FLP consists of a series of clock and data pulses,
with the data pulses used to form a 16-bit link code work (LCW).
Figure 3.38 illustrates the composition of the LCW. Note that the fivebit selector field makes it possible for auto-negotiation to be supported by
Token-Ring (802.5) networks. In comparison, the eight-bit Technology Ability
field defines the technology supported. The Pause bit in the Technology
Ability Field was added to support full-duplex operations and when set
indicates if the device supports a Pause Mechanism. If so, the other device
can transmit a Pause Frame that will inform the receiving device to delay
its transmission. This form of flow control becomes necessary when a 100Mbps device communicates with a lower-speed device and the latter could be
overwhelmed with data. Thus, the 100-Mbps device will need to inform other
ethernet networks
Selector field
S0
S1
S2
00001 = 802.3
00010 = 802.5
S3
125
Technology ability field
S4
A0
A1
A0
A1
A2
A3
A4
A5
RF
ACK
NP
Figure 3.38
=
=
=
=
=
=
=
=
=
A2
A3
A4
A5
A6
A7
RF ACK NP
10BASE-T
10BASE-T Full Duplex
100BASE-TX
100BASE-TX Full Duplex
100BASE-T4
Pause
Remote Fault
Acknowledge
Next Page
The format of the link code word.
devices to delay their transmission. In Chapter 4 when we examine the format
of different types of Ethernet frames we will also examine the Pause Frame.
Returning our attention to Figure 3.38, the Remote Fault indicates the
presence of a fault detected by the station at the opposite end of the link.
When this bit (RF) is set, it indicates that the distant device is in its link-failed
state, a condition typically resulting from a malfunctioning receiver or broken
cable. The next to last bit in the LCW is the ACK bit. This bit is used to
indicate the successful receipt of the previously transmitted LCW. The last
bit is the next page bit. When set this bit indicates that additional proprietary
information follows.
The link code word is transmitted on power-up or when requested by either
device on a link. The composition of fast link pulses appears similar to a
link integrity test, enabling a 10BASE-T device to respond. However, if the
remote station is a 100 Mbps device, its receive clock will be sufficient to
interpret the individual pulses that form the LCM, enabling the determination
of operations at 10 Mbps or 100 Mbps.
When two devices that support auto-negotiation have multiple capabilities,
they need to determine their highest performance mode of operation. In
doing so, they consult a priority table. Table 3.7 lists the priorities for autonegotiation from highest to lowest. Note that full duplex has a higher priority
than half duplex since the former has a higher throughput capability than
the latter.
Autonegotiation can operate on an individual end basis or with
autonegotiation-compliant devices at both ends of a twisted-pair link. An
example of this autonegotiation capability is shown in Figure 3.39 in which
126
chapter three
TABLE 3.7
Auto-negotiation Priority Table
Priority
Technology
Cable
1
100BASE-TX Full Duplex
2 pair, category 5
2
100BASE-T4
4 pair, category 3
3
100BASE-TX
2 pair, category 5
4
10BASE-T Full Duplex
2 pair, category 3
5
10BASE-T
2 pair, category 3
10BASE-T/100BASE-TX
switching hub with
autonegotiation
Autonegotiation
10BASE-T/
100BASE-TX
Station
A
10BASE-T
Station
B
Figure 3.39 Autonegotiation. Through the use of autonegotiation, the
10BASE-T/100BASE-TX switching hub will operate at 100 Mbps with
station A and at 10 Mbps with station B.
station A has a NIC with autonegotiation capability that is cabled to a 10BASET/100BASE-TX switching hub whose port has an autonegotiation capability,
while station B has a conventional 10BASE-T NIC. In this example, the
hub port connected to station A would operate at 100 Mbps as 100BASE-TX
devices. In comparison, the hub port connection to station B would recognize
the 10BASE-T signals from station B and switch to 10BASE-T operations.
Operation
There are several network configurations you can consider for Fast Ethernet
operations. First, you can construct a 100BASE-TX network similar to a
10BASE-T network, using an appropriate hub and workstations that support
the specific network type. This type of network, commonly referred to as a
shared-media hub-based network, should be considered when most, if not all,
ethernet networks
127
network users access one or a limited number of servers to perform graphicintensive or similar bandwidth-intensive operations, and the sharing of the
100-Mbps transmission capability provides an acceptable average bandwidth
per network node. Figure 3.40a illustrates a Fast Ethernet shared-media hub
network configuration.
In examining Figure 3.40a note that the single CSMA/CD network represents
a collision domain. Here the term collision domain references a network
structure in which a collision will occur if two computers attached to the
network transmit at the same time. If one hub is wired to another or two
segments are joined together by a repeater the resulting network structure
continues to function as a single collision domain.
Another common use for Fast Ethernet is in its incorporation into one
or more ports in an Ethernet switch. An Ethernet switch can be viewed as
Server
All connections operate at 100 Mbps.
(a) Shared-media hub.
Image/
video
server
Application
server
Some connections operate at 100 Mbps and some at 10 Mbps.
(b) 10/100 Mbps ethernet switch.
Legend:
= Workstation
= Fat pipe
Figure 3.40
Fast Ethernet network applications.
128
chapter three
a sophisticated hub that can be programmed to transmit packets arriving
on one input port to a predefined output port. This is accomplished by
the switch reading the destination address in the frame and comparing
that address to a table of preconfigured address-port relationships. Readers
are referred to Chapter 5 for specific information concerning the operation
and use of Ethernet switches. In the interim, in examining the use of Fast
Ethernet switches, note that some older switches contain one or two 100-Mbps
operating ports, while the other ports are conventional 10BASE-T ports. Due
to economies of scale, most Fast Ethernet switches manufactured since the
beginning of the new millennium support 10/100-Mbps port operations on
all switch ports. The only major exception to the support of dual-speed autonegotiation ports is when the switch includes one or more Gigabit Ethernet
ports, with the latter restricted to operating at 1 Gbps. Typically you would
connect servers that have a heavy workload to 100-Mbps Fast Ethernet ports,
such as an image/video server and a database server. Figure 3.40 illustrates the
use of a 10/100-Mbps Ethernet switch. In examining Figure 3.40a and b, note
that the heavily shaded connections to all workstations in Figure 3.40a and to
the servers in Figure 3.40b represent 100-Mbps Fast Ethernet ports requiring
Fast Ethernet adapter cards to be installed in each server or workstation
connected to a 100-Mbps port. A common term used to reference the 100-Mbps
connection is fat pipe.
Because a Fast Ethernet port provides downward compatibility with
10BASE-T, you can interconnect conventional 10BASE-T hubs to a Fast
Ethernet shared-media hub or Fast Ethernet switch. Similar to having the
ability to interconnect two hubs to extend a Fast Ethernet network, you can
interconnect two switches. However, unlike the connection of hubs, which
extends the collision domain, each switch port will not forward collision
signals received, enabling attached segments to operate independently of one
another. This means you can use LAN switches to construct larger networks
by interconnecting shared media segments via one or more switches.
Configuration Guidelines
To facilitate the creation of a Fast Ethernet network you need to consider
the maximum collision domain based upon the type of repeater used and the
media supported. In addition to the previously mentioned Class I and Class II
repeaters, you need to remember that a connection from a station to a hub
port represents a repeater on a single segment. With this in mind, Table 3.8
indicates the maximum collision domain based upon the class of repeater,
and the type of media used.
ethernet networks
TABLE 3.8
129
Fast Ethernet Maximum Collision Domain (Meters)
Type of Media
Type of Repeater
Copper
Fiber
Copper and
Fiber
(T4 and FX)
Copper and
Fiber
(TX and FX)
Single Segment
100
412
N/A
N/A
One Class I
200
272
231∗
260∗
One Class II
200
320
N/A†
308∗
Two Class II
205
228
N/A†
216‡
Notes:
∗ Assumes 100 meters of copper and one fiber link.
† Not applicable as T4 and FX cannot be interconnected using a Class II repeater.
‡ Assumes 105 meters of copper and one fiber link.
In examining the entries in Table 3.8 note that the first entry represents
a station-to-station connection with no intermediate repeaters. As indicated,
the maximum amount of cabling is 100 meters when copper is used and
412 meters when fiber-optic cable is employed. Figure 3.41 illustrates an
example of the structure of a Fast Ethernet network consisting of three shared
media hubs and a switch.
An examination of Figure 3.41 should be performed in conjunction with
the entries in Table 3.8 for the figure to be meaningful. For example, from
Table 3.8, the maximum collision domain for a class II repeater is 205 meters.
If you focus your attention upon collision domain 1 in Figure 3.41, you
will note that the inter-repeater segment is shown as 5 meters to keep the
maximum collision diameter equal or below 205 meters. You can increase the
inter-repeater segment length; however, you would then need to reduce the
other segments to ensure that the maximum collision diameter in the domain
does not exceed 205 meters.
Computer Bus Considerations
In addition to having to consider the type or types of NIC connectors, it
is also extremely important to consider the bus they are designed to work
with along with driver support for different operating systems. Some NICs
are limited to one type of connector, requiring the purchase of a separate
converter if you decide to use a different media other than the media the NIC
was manufactured to directly support.
130
chapter three
Switch
Fiber
link
Class I
repeater hub
Class II
repeater hub
272 meters
100 meters
100 meters
Class II
repeater hub
Collision
domain 2
100 meters
Collision
domain 1
Figure 3.41 Considering the maximum collision domain and cable length
when configuring a Fast Ethernet network.
Concerning computer bus support, in the DOS/Windows world you can
select NICs that work with the original IBM PC 8- and 16-bit industry-standard
architecture (ISA) bus, the 32-bit extended industry-standard architecture
(EISA) bus, IBM’s proprietary and essentially now obsolete microchannel
architecture (MCA) 16- and 32-bit buses, and the 32- and 64-bit peripheralcomponent interconnect (PCI) bus. The key difference between adapters
supporting different computer buses is in their ability to transmit and receive
sustained bursts of traffic such as file transfers. The ISA bus has a throughput
between 4 Mbytes/sec and 5 Mbytes/sec, while the first-generation EISA bus
has a throughput of 16 Mbytes/sec, later modified to support burst speeds
up to 33 Mbytes/sec. IBM’s 16-bit MCA bus can reach 20 Mbytes/sec, with
its 32-bit bus slightly exceeding 36 Mbytes/sec. While all of the preceding
buses can normally support small bursts of traffic onto a network operating at
10 Mbps, when used for sustained transfers onto a 100-Mbps network those
buses cannot keep up with the network, resulting in a suboptimal level of
performance. However, the use of a PCI-based NIC can resolve such problems.
This is because the 32-bit PCI bus can support data bursts up to 132 Mbytes/sec
while a 64-bit PCI bus can reach 264 Mbytes/sec, sufficient for operations at
100 Mbps.
ethernet networks
131
Figure 3.42 The 3Com Corporation Fast Etherlink TX NIC supports the use
of the ISA bus. (Photograph courtesy of 3Com Corporation.)
Figures 3.42 through 3.44 illustrate three 3Com Corporation Fast Ethernet
NICs manufactured to support different computer buses. Figure 3.42 shows
the 3Com 3C515-TX NIC, which supports the ISA bus. Figure 3.43 shows the
3Com Corporation’s Fast Etherlink PT adapter, which supports the EISA bus.
The third 3Com NIC shown in Figure 3.44 is the Fast Etherlink XL, designed
to support the PCI bus.
A third NIC-related item that warrants careful consideration when selecting
an appropriate NIC is the drivers available for use with the adapter. Today,
in addition to different versions of NetWare and Windows, there are other
operating systems ranging from LINUX to different versions of UNIX you may
need to support. Although support for many NICs and protocols is built into
many operating systems, such support is not all-inclusive. This means you
may have to obtain one or more files from the NIC manufacturer to be able
to use the NIC with a particular operating system. Figure 3.45 illustrates how
under Windows NT Version 4.0 you would select the ‘‘Have Disk’’ option if
your NIC wasn’t directly supported, resulting in the display of the ‘‘Insert
Disk’’ dialog box to enable NT to recognize your adapter.
One of the more interesting aspects concerning the use of NIC drivers,
obtained from firsthand experience, is to ensure you have the appropriate
132
chapter three
Figure 3.43 The 3Com Corporation Fast Etherlink PT NIC supports the use
of the EISA bus. (Photograph courtesy of 3Com Corporation.)
Figure 3.44 The 3Com Corporation Fast Etherlink XL NIC supports the use
of the PCI bus. (Photograph courtesy of 3Com Corporation.)
ethernet networks
133
Figure 3.45 When using Windows NT you can specify a ‘‘Have Disk’’ option
to use a NIC adapter not directly supported by that operating system.
driver. On occasion, it was found that the diskette packaged with the NIC
manual and shipped with a newly purchased computer did not contain the
appropriate driver for the computer. Although the operating system would
use the driver, the computer would freeze about once a month, a situation
corrected when it was determined that the computer manufacturer had a Web
site with updated drivers for the NIC shipped with their computer.
100VG-AnyLAN
100VG-AnyLAN was originally proposed by AT&T Microelectronics and
Hewlett-Packard Company as a mechanism to support evolving LAN-based
multimedia applications. To accomplish this, the proposal, which became the
IEEE 802.12 standard, replaced the CSMA/CD access protocol by a demandpriority scheme. This scheme enables the transmission of time-sensitive
video and voice associated with multimedia applications in preference to
conventional data that is relatively time-insensitive. In addition, the use
of a demand-priority scheme extends 100VG-AnyLAN beyond Ethernet,
134
chapter three
enabling this network to support Token-Ring, FDDI, and other types of local
area networks.
Although 100VG-AnyLAN included many interesting technical characteristics, it never achieved widespread adoption. The significant commercial
success of 100BASE-T, as well as the rapid reduction in the cost of
Gigabit Ethernet hardware, more than likely affected its adoption. In late
2000, Hewlett-Packard, which was the major proponent of 100VG-AnyLAN,
announced plans to discontinue the sale and future development of interface
cards, software license and documentation for its e3000 business servers.
Architecture
While recognizing that the market for VG-AnyLAN is minimal, in order to
provide readers with information about different types of Ethernet networks,
in concluding this section we will briefly examine the architecture and basic
operation associated with the technology.
100VG-AnyLAN was designed as a hub-centric network architecture. A
central hub, known as a level-1 or root hub, functions as an inverted tree
base in establishing a 100VG-AnyLAN network. From this hub other hubs
and/or nodes form a star topology, fanning out underneath the root hub, as
illustrated in Figure 3.46. All hubs located in the same network segment must
be configured to support the same frame format — IEEE 802.3 Ethernet or IEEE
802.5 Token-Ring. Through the attachment of a bridge or router to a hub
port you can extend the 100VG-AnyLAN network to interconnect with other
Ethernet or Token-Ring networks, FDDI- and ATM-based networks, or a wide
area network transmission facility.
Each hub in a 100VG-AnyLAN network has one up-link port, labeled
up in Figure 3.46; and n down-link ports, labeled 1 through N. The uplink port on each hub is reserved for connecting lower-level hubs to an
upper-level hub, while the down-link ports are used to connect an upperlevel hub to workstations, bridges, routers, and other network devices to
include lower-level hubs. Up to three levels of cascading can be used on a
100VG-AnyLAN network.
Hub Operation
Each hub port can be configured to operate in one of two modes — normal
or monitor. Ports configured to operate in their normal mode are forwarded
only those packets specifically addressed to the attached node. In comparison,
ports configured to operate in the monitor mode are forwarded every packet
received by the hub.
ethernet networks
135
Level-1 Root hub
Up
1
WS
1-1
2
3
4
WS
1-2
• • •
N
WS
1-4
WS
1-N
Level-2 hub
Up
1
WS
2-1
2
3
• • •
N
WS
2-2
Legend: WS 1-p = workstation hub level-port
WS
2-N
Figure 3.46
topology.
100VG-AnyLAN
Devices connected to nodes gain access to a 100VG-AnyLAN network
through the use of a centrally controlled access method, referred to as demandpriority. Under the demand-priority access method a node issues a request,
referred to as a demand, to the hub it is connected to, to transmit a packet
onto the 100VG-AnyLAN network. Each request includes a priority label
assigned by the upper-layer application. The priority label is either normal,
for normal data packets, or high, for packets carrying time-critical multimedia
information. As you might expect, high-priority requests are granted access to
the network before normal-priority requests are granted.
The level 1 or root hub continuously scans its ports using a round-robin
sequence for requests. Lower-level hubs connected as nodes also perform a
round-robin scanning process and forward node requests to the root hub.
The root hub determines which nodes are requesting permission to transmit
a packet as well as the priority level associated with the packet. Each hub
maintains a separate list for both normal- and high-priority requests.
Normal-priority requests are serviced in their port order until a higherpriority request is received. Upon receipt of a higher-priority request the
hub will complete any packet transmission in progress and then service the
high-priority packet before returning to service the normal-priority list.
To prevent a long sequence of high-priority requests from abnormally
delaying low-priority requests from being serviced, the hub also monitors
136
chapter three
node request-to-send response times. If the delay exceeds a predefined
time, the hub automatically raises the normal-priority level to a highpriority level.
Round-Robin Scanning
Figure 3.47 illustrates an example of the 100VG-AnyLAN hub round-robin
scanning process. Assume all ports initially have normal-priority requests
pending. Then, the round-robin sequence at the root or level 1 hub results
in the packet service order commencing at the level 1 hub’s first port (1-1).
Next, the level 1 hub’s second port is serviced (1-2). When the third port
is examined, it inserts the round-robin sequence generated by the level 2
hub. That is, it inserts the packet order sequence 2-1, 2-2, 2-3, . . ., 2-N. This
sequence is then followed by adding the remaining level 1 hub ports. Thus, the
inserted packet order sequence is followed by the sequence 1-4, . . ., 1-N. Now
let’s assume at time t = 0 nodes 2-1 and 1-4 generate high-priority requests.
Then, the packet service order at the level 1 hub would be revised, becoming
2-1, 1-4, 1-1, 1-2, 2-2, . . ., 2-N, 1-5, . . ., 1-N.
Level-1 Root hub
Up
1
WS
1-1
2
3
•
4
WS
1-2
•
•
N
WS
1-4
WS
1-N
Level-2 hub
Up
1
WS
2-1
2
3
WS
2-2
•
•
•
N
WS
2-N
If all ports have normal-priority requests pending, then:
Level-1 scan 1-1, 1-2, 1-3, 1-4, ..., 1-N
Level-2 scan 2-1, 2-2, 2-3, ..., 2-N
Level-1 resulting packet order sequence
1-1, 1-2, 2-1, 2-2, 2-3, ..., 2-N, 1-4, ..., 1-N
Figure 3.47 100VG-AnyLAN hub
round-robin scanning.
ethernet networks
137
Cabling Requirements
100VG-AnyLAN products provide a 100-Mbps data rate using 4-pair category 3, 4, or 5 unshielded twisted-pair cable. 100VG-AnyLAN also supports
2-pair UTP, 2-pair STP, and fiber-optic cabling. The cabling medium required
for 4-pair 100VG-AnyLAN is 4-pair UTP that meets the specifications of the
EIA/TIA-568 standard for 100-ohm category 3, 4, or 5 cable. Such cable connects to RJ-45 wall jacks and can go through punch-down blocks or patch
panels in a wiring closet.
Figure 3.48 indicates the RJ-45 modular plug pin assignments for 4-UTP
100VG-AnyLAN. In addition, this illustration indicates for comparison purposes the pairs used by 10BASE-T and Token-Ring cable.
Comparing Technologies
Both 100BASE-T and 100VG-AnyLAN provide users with a tenfold increase
in network operating rates with respect to 10BASE-T. However, the 100-Mbps
100VG-AnyLAN (4-UTP)
1
2
3
4
Pair 2
5
6
7
Pair 1
8
Pair 4
Pair 3
Token-ring
1
2
3
4
5
6
7
8
6
7
8
Pair 1
Pair 2
10BASE-T
1
2
3
4
5
Pair 1
Pair 2
Figure 3.48 RJ-45 modular plug
pin assignments.
138
chapter three
operating rate represents one of the few common operating characteristics
between the two technologies.
100BASE-T was designed as a mechanism to provide a growth path from
100BASE-T while enabling organizations to retain the invested base of
10BASE-T adapter cards and cabling for workstations that do not require
a 100-Mbps operational capability. In doing so, no provision was made to
prioritize traffic. Thus, while a lightly or medium loaded 100BASE-T network can transport multimedia data, this may not be true as the network
load increases.
The replacement of the CSMA/CD access protocol by a demand-priority
protocol results in 100VG-AnyLAN being more suitable for multimedia
applications. Unfortunately, there is a cost associated with the additional
technology incorporated into 100VG-AnyLAN, which resulted in network
adapters initially costing 50 percent more than 100BASE-T adapters and
precluded its widespread adoption. Table 3.9 provides a comparison of the
operating characteristics of 100BASE-T and 100VG-AnyLAN.
A comparison of 100VG-AnyLAN to 100BASE-T is probably similar to a
comparison of Beta to VHS. While Beta was a superior technology, the VHS
extended record and playback time better satisfied consumer requirements.
If we turn our focus back to LANs, 100BASE-T better interoperates with
10BASE-T through an auto-negotiation capability. This makes it easier to
construct interoperable networks. Based upon the preceding, 100BASE-T
products dominate the high-speed Ethernet marketplace.
3.5 Gigabit Ethernet
Gigabit Ethernet represents an extension to the 10-Mbps and 100-Mbps
IEEE 802.3 Ethernet standards. Providing a data transmission capability of
1000 Mbps, Gigabit Ethernet supports the CMSA/CD access protocol, which
makes various types of Ethernet networks scalable from 10 Mbps to 1 Gbps,
with the pending standardization of 10-Gigabit Ethernet providing additional
scalability to 10 Gbps.
Components
Similar to 10BASE-T and Fast Ethernet, Gigabit Ethernet can be used as
a shared network through the attachment of network devices to a 1-Gbps
repeater hub providing shared use of the 1-Gbps operating rate or as a switch,
ethernet networks
TABLE 3.9
139
Operating Characteristics Comparison
Data rate
100BASE-T
100VG-AnyLAN
100 Mbps
100 Mbps
Access protocol
CSMA/CD
Demand-priority
Frame support
802.3 Ethernet
802.3 Ethernet
802.5 Token-Ring
Physical topology
Star
Star
Cable support
2-pair Category 5 UTP
4-pair Category 3, 4, or 5 UTP
2-pair STP
Fiber
100BASE-TX
100BASE-T4
100BASE-TX
100BASE-FX
Not planned
Yes
Yes
Yes
Maximum UTP drive distance
100 meters on
Category 5 UTP
100 meters Category
3, 4, UTP
Maximum repeaters allowed
2
4
Maximum UTP network diameter
200 meters
4,000 meters
Other cabling
2-pair Type 1 STP;
optical fiber; 4-pair
Category 3, 4, 5 UTP
2-pair Type 1 STP;
optical fiber; 2-pair
Category 5 UTP
Full duplex support
Yes
Yes
the latter providing 1-Gbps ports to accommodate high-speed access to servers
while lower operating rate ports provide access to 10-Mbps and 100-Mbps
workstations and hubs. Although few organizations can be expected to require
the use of a 1-Gbps shared media network as illustrated in Figure 3.49a,
the use of Gigabit switches can be expected to play an important role in
providing a high-speed backbone linking 100-Mbps network users to large
databases, mainframes, and other types of resources that can tax lower-speed
networks. In addition to hubs and switches, Gigabit Ethernet operations
require workstations, bridges, and routers to use a network interface card to
connect to a 1-Gbps network. Gigabit Ethernet NICs introduced when this book
was written were designed for PCI bus operations and use an SC fiber connector
to support 62.5/125 and 50/125 micron fiber. Such adapters provide 1 Gbps
of bandwidth for shared media operations and 2 Gbps aggregate bandwidth
when used for a full-duplex connection to a switch port. In Chapter 5, when
140
chapter three
In a shared media environment the 1-Gbps bandwidth provided by
Gigabit Ethernet is shared among all users.
(a) Shared media hub use
Server
Server
1 Gbps
port
Switching
hub
100 Mbps
ports
•
•
•
100 Mbps
hub
100 Mbps
hub
In a switch environment each 1-Gbps port can provide a full–duplex
2-Gbps data transfer capability.
Legend:
= workstations
(b) Switching hub use
Figure 3.49
Using Gigabit Ethernet.
we review switching, we will examine in more detail how Gigabit switches
can be used to provide a backbone network capability.
Sublayer Architecture
Similar to Fast Ethernet, it was recognized that Gigabit Ethernet would operate
over different types of media, resulting in the need to develop an architecture
that included sublayers well suited to support the coding necessary for the
different media. Figure 3.50 illustrates the sublayer architecture of Gigabit
Ethernet and the relationship of the sublayers to the lower two layers of the
ISO’s Open System Interconnection (OSI) Reference Model.
ethernet networks
ISO OSI
reference model
(lower 2 layers)
141
Gigabit ethernet sublayers
Logical Link Control (LLC)
Data link
Media Access Control (MAC)
Reconciliation
Physical
Gigabit Media Independent Interface (GMII)
Physical Coding Sublayer (PCS)
Physical Medium Attachment (PMA)
Physical Medium Dependent (PDM)
Figure 3.50
Gigabit Ethernet Sublayer Architecture.
In examining the Gigabit Ethernet sublayer architecture shown in Figure 3.50
you will note its resemblance to Fast Ethernet. The reconciliation sublayer represents a transparent interface between the MAC sublayer and the
physical layer, decoupling the above sublayer from the physical sublayer.
The function of the reconciliation layer is to collect 8-bit bytes from the
MAC layer for transfer onto the Gigabit Media Independent Interface (GMII).
The GMII contains 8-bit transmit and receive data paths and a 125-MHz
clock, resulting in a data transfer rate to and from the MAC layer of
1 Gbps.
Under the GMII the same basic sublayers are used as under Fast Ethernet.
However, instead of using FDDI coding and signaling, Gigabit Ethernet relies
on the physical sublayers of the Fiber Channel standard.
Media Support
There are five types of media supported by Gigabit Ethernet — single-mode
fiber, multimode fiber, short runs of coaxial cable or shielded twisted pair, and
longer runs of unshielded twisted pair. Figure 3.51 illustrates the relationship
of Gigabit Ethernet’s MAC and physical layers to include the drive distances
supported for each type of media. Table 3.10 summarizes the flavors of Gigabit
Ethernet indicating the IEEE designator used to reference Gigabit operations
on a specific type of media and the maximum transmission distance associated
with the use of each type of media.
142
chapter three
MEDIA ACCESS CONTROL
Full and/or half duplex
Logical media-independent interface
MAC layer
8B/10B
coding
Physical layer
*Copper physical
encoder/decoder
850 nm
multimode
fiber
1300 nm
single mode
fiber
1000BASE-CX
up to 25 m
1000BASE-T
up to 100 m
For 50 µm
core max
distance is
550 m for
LX and SX
Up to
5km
For 62.5 µm
core max
distance is
220 m SX
550 m LX
Figure 3.51 Gigabit Ethernet media support. Encoding method for
1000BASE-T to be defined by the IEEE 802.3ab Task Force, probably during
late 1998.
TABLE 3.10 The Flavors of Gigabit Ethernet
Media Designator
Media Type
Transmission Distance
1000BASE-LX
SMF 8/125 µ
5 km
1000BASE-LH
SMF 8/125 µ
70 km
1000BASE-LX
MMF, 50/125 µ
550 m
1000BASE-LX
MMF, 62.5/125 µ
550 m
1000BASE-SX
MMF, 50/125 µ
550 m
1000BASE-SX
MMF, 62.5/125 µ
220 m
1000BASE-CX
Shielded balanced copper(coaxor STP)
25 m
1000BASE-T
UTP,Category 5
100 m
SMF, single-mode fiber; MMF, multimode fiber; UTP, unshielded twisted pair; STP, shielded
twisted pair.
ethernet networks
143
Single-Mode Fiber
The specification that governs the use of single-mode fiber is referred to
as 1000BASE-LX, where L represents recognition of the use of long-wave
light pulses. The maximum distance obtainable for Gigabit Ethernet when
transmission occurs using a 1330-nanometer (nm) frequency on single-mode
fiber is 5 km. The 1000BASE-LX cabling standard is based upon the use of
a 1300 nm laser, which is also used in FDDI. This provides a transmission
distance up to 5 km when the laser is used with single-mode fiber. When
a 1300 nm laser is used with multi-mode fiber the maximum transmission
distance is reduced to 550 m.
1000BASE-LH
Although not part of the original Gigabit Ethernet specification the use of a
1550 nm laser with single-mode fiber has received considerable attention and
is referred to as 1000BASE-LH. When a 1550 nm laser is used with 8/125µ
single-mode fiber, a transmission range up to 70 km becomes possible, which
expands the potential use of Gigabit Ethernet to wide area networking.
In examining Figure 3.51, note that the 8B/10B coding scheme used for
both single-mode and multimode fiber represents the coding scheme used by
the Fibre Channel. Due to the importance of Fibre Channel technology incorporated into Gigabit Ethernet, a short digression to discuss that technology
is warranted.
The Fibre Channel The Fibre Channel actually represents a rather old
technology, dating back to 1988 when the American National Standards
Institute charted a working group to develop a high-speed data transfer
capability for supporting data transfers between computers and peripheral
devices. Initially, the Fibre Channel was developed to support fiber-optic
cabling. When support for copper media was added, an ISO task force renamed
the spelling of fiber to reduce the association of fiber optics while maintaining
the name recognition of the technology. Today both optical and electrical
copper-based media are supported by the Fibre Channel, with operating rates
ranging from 133 Mbps to 1.062 Gbps provided by most equipment. In 1995
ANSI approved 2.134- and 4.25-Gbps speed rates as enhancements to the
previously developed Fibre Channel specifications.
The Fibre Channel supports point-to-point, shared media, and switch network topologies, with Fibre Channel hubs, switches, loop access controllers,
and NICs installed on computers used to provide different network structures.
The Fibre Channel supports a five-layer protocol stack from FC-0 through
144
chapter three
FC-4. The three lower layers form the Fibre Channel physical standard, while
the two upper layers provide the interface to network protocols and applications. The FC-0 layer defines the physical characteristics of the media,
transmitters, receivers, and connectors available for use, transmission rates
supported, electrical and optical characteristics, and other physical layer
standards. The second layer, FC-1, defines the 8B/10B encoding and decoding
method used, a serial physical transport, timing recovery, and serial line
balance. The 8B/10B encoding and decoding scheme was patented by IBM
and is the same technique used in that vendor’s 200-Mbps ESCON channel
technology. Under 8B/10B coding eight data bits are transmitted as a 10-bit
group. The two extra bits are used for error detection and correction.
To enable 1-Gbps operations the transmission rate of the Fibre Channel was
raised to 1.25 Gbps. Then, the use of an 8B/10B coding technique permits
data transfer at 80 percent of the operating rate or 1.256 ∗ 80, resulting in the
modified Fibre Channel technology being capable of supporting the 1-Gbps
data transfer of Gigabit Ethernet.
The third layer, FC-2, functions as the transport mechanism. This layer
defines the framing rules of data transmitted between devices and how data
flow is regulated. Although Ethernet frames are encapsulated within Fibre
Channel frames for transmission between devices, the encapsulation is transparent to the Ethernet frame. Thus, the use of a modified Fibre Channel
technology enabled the IEEE to use a proven technology as a transport mechanism for connecting Gigabit Ethernet devices while enabling the CSMA/CD
protocol and Ethernet frame to be retained.
Multimode Fiber
The support of multimode fiber is applicable to both long-wave (LX) and
short-wave (SX) versions of Gigabit Ethernet. When transmission occurs on
multimode fiber the short-wave version of Gigabit Ethernet, which is referred
to as 1000BASE-SX, uses an 850-nm frequency, providing a 220-meter maximum distance when transmission occurs over a 62.5-micron (µm) core. Note
that in some trade publications and books the maximum transmission distance
may be indicated as either 260 or 300 meters; however, in September 1997,
the IEEE task force working on the development of Gigabit Ethernet standards
lowered the maximum distance to 260 meters from 300 meters and, due to
problems encountered from using an extremely high-frequency light source
over multi-mode fiber referred to as differential mode delay, had again to
lower the maximum transmission distance. That distance is now 220 meters
when 62.5/125µ multi-mode fiber is used. When 50/125µ multi-mode fiber
is used the maximum transmission distance is 550 meters. When a 1300-nm
ethernet networks
145
frequency is used on a 62.5-µm core, which represents 1000BASE-LX, a maximum transmission distance of 550 meters becomes possible. This distance was
formerly 500 meters and was reduced by the IEEE Gigabit Ethernet task force
and then revised back above its original valve. When a 50-micron core fiber is
used, the maximum distance is 550 meters for both LX and SX specifications.
Copper Media
Gigabit Ethernet supports three types of copper media — category 5 STP,
coaxial cable, and UTP. The use of the first two types of copper media
are defined by the 1000BASE-CX standard, which governs the use of patch
cords and jumpers that provide a maximum cabling distance of 25 meters.
This standard also uses the Fibre Channel–based 8B/10B coding method at
a serial line operating rate of 125 Gbps, and runs over a 150-ohm balanced,
shielded, cable assembly, referred to as twinax cable as well as on STP. The
25-meter maximum cabling distance makes 1000BASE-CX suitable for use in
a wiring closet or a computer room as a short jumper interconnection cable. In
comparison, the 1000BASE-T standard defines the use of four-pair HTP cable,
with each wire carrying a 125-MHz signal in each direction.
The transmission of data at 1 Gbps over copper represented a daunting
challenge. Not only was the data rate on a per-wire-pair basis extremely
high at 250 Mbps, but, in addition, developers needed to consider the dualduplex transmission method by which signals could flow in both directions
simultaneously on all four wire pairs. This challenge resulted in the need
to charter a separate working group referred to as the IEEE 802.3ab task
force to develop specifications for the transmission of Gigabit over VTP. The
task force examined three encoding schemes referred to as QAM (quadrature
amplitude modulation), PAM (pulse amplitude modulation), and CAP (carrierless amplitude phase modulation). Each coding scheme uses a different
procedure and operates between 75 and 125 MHz in frequency. Both the
selection of an encoding scheme and the method for manufacturers of NICs to
block frequencies above 30 MHz under FCC regulations were required to be
addressed.
As a result of the efforts of the IEEE 802.3ab task force a five-level version
of PAM was selected as the coding method, with each level representing a
symbol that transports two bits of information. Under five-level PAM each
transmitted symbol represents one of five different levels (−2, −1, 0, +1, +2).
For example, a signal level of −2 could be used to represent the dibit pair
00 while the signal level −1 could represent the dibit pair 01. To support bidirectional transmission over each wire pair digital signal processors (DSPs)
146
chapter three
are used. The DSPs operate similar to DSPs used in modem transmission that
support full-duplex operations over a wire pair. That is, the DSPs support
echo cancellation in which the transported signal is then suppressed.
Gigabit Repeater
To extend the transmission distance of Gigabit Ethernet the IEEE defined
two types of repeaters. Those repeaters are known as Transmission System
Model 1 and Transmission System Model 2. The use of either type of repeater
is governed by the round-trip propagation delay characteristics associated
with different network devices to include data terminal equipment (DTE)
adapter card delays and repeater delays.
Under the Transmission System Model 1 rules are defined for DTE to DTE
direct connections as well as for the use of a single repeater used to interconnect media using different signaling methods. The use of a Transmission
System Model 1 repeater connects segments into a single collision domain.
Table 3.11 lists the maximum transmission distances associated with the use
of a transmission system 1 repeater.
The second type of repeater specified for Gigabit Ethernet is referred to
as a Transmission System Model 2 repeater. Similar to a model 1 system, a
model 2 system is used to interconnect DTEs. The constraints concerning the
use of a Model 2 repeater include use of only one such repeater and the round
trip propagation delay between two interconnected DTEs should not exceed
4096 bit times. While the use of a Model 2 repeater provides more flexibility
TABLE 3.11 Maximum Transmission Distance Using a
Transmission System Model 1 Repeater
Media (DTE to DTE)
Category 5 UTP
Twin-axial
Maximum Distance (Meters)
200
50
1000BASE-LX or 1000BASE-SX
220
Category 5 UTP and
1000BASE-LX or 1000BASE-SX
210∗
Twin-axial and 1000BASE-LX or
1000BASE-SX
220†
Notes:
∗ Assumes 1000 meters of category 5 UTP.
† Assumes 25 meters of twin-axial.
ethernet networks
147
than the use of a Model 1 repeater, the user becomes responsible for ensuring
that the round trip propagation delay constraint is not violated.
When computing the round-trip propagation delay you need to consider the
media delays, DTE delays, repeater delay and a safety margin, with the last
specified as 32 bit times. Table 3.12 indicates the maximum round-trip delays
associated with different Gigabit Ethernet network components.
To illustrate the use of the delay times listed in Table 3.12, let’s assume
we want to interconnect two DTEs through the use of 100 meters of UTP and
110 meters of 1000BASE-LX. When using a Transmission System Model 2
repeater our computations would be as in Table 3.13. Note that the total of
4095 bit times is less than the 4096 maximum permitted bit time, which
means that the use of the repeater is just within tolerance.
TABLE 3.12
Time
Gigabit Ethernet Component Delay
Component
Round Trip Delay (Bit Times)
Safety Margin
32
Each DTE
432
Repeater
976
Category 5 UTP
11.12 bit times/meter
Twin-axial
10.10 bit times/meter
1000BAE-LX or SX
10.10 bit times/meter
TABLE 3.13 Computations for a Transmission Model 2 Repeater
Component
DTE1
100 meters UTP
Repeater
100 meters 1000BASE-LX
DTE2
Bit Time
432
1112
976
1111
432
Safety Margin
32
Total bit time
4095
148
chapter three
Duplex Capability
Gigabit Ethernet supports a full-duplex operating mode for switch-to-switch
and switch-to-end station connections, while a half-duplex operating mode
is used for shared connections using repeaters. Although not supported
in the 802.3z standard, several vendors announced full-duplex, multiport,
hublike devices to be used to interconnect two or more 802.3 links operating
at 1 Gbps. Referred to as a buffered distributor, this device functions as a
repeater, forwarding all incoming data received on one port onto all ports
other than the original port. However, unlike conventional repeaters, the
buffered distributor has the ability to buffer one or more inbound frames on
each port before forwarding them.
In addition to having memory, the buffered distributor supports the IEEE
802.3z flow control standard. These two features enable each port on a
buffered distributor to provide transmission support at a rate that matches the
maximum rate of the 1-Gbps shared bus used to broadcast frames received on
one port to all other ports. Because the buffered distributor supports 802.3z
full-duplex flow control, the situation where an offered load exceeds the
bandwidth of the shared bus is easily handled. This is accomplished by a port
transmitting a flow control frame to the transmitting system, indicating that the
port cannot accommodate additional data. The 802.3z-compliant transmitting
device will then cease transmission until the port forwards a frame indicating
it can accept additional data.
To illustrate the operation of a buffered distributor, consider Figure 3.52,
which illustrates the use of a four-port buffered distributor. Let’s assume that
at a certain point in time devices connected to ports 1 and 3 are transmitting
1
700
Mbps
Legend :
2
3
4
Enable
flow
control
Buffer
area
400
Mbps
Disable
flow
control
= Enable/Disable flow control frames
Figure 3.52 A buffered distributor uses IEEE 802.3z flow control whenever
the offered traffic exceeds the bandwidth of its shared bus.
ethernet networks
149
data to the buffered distributor at 700 Mbps and 400 Mbps, respectively. It
should be noted that those data rates represent traffic carried and not the
operating rate of those devices which would be 1 Gbps. Because the aggregate
rate exceeds 1 Gbps, a normal repeater would not be able to support this
traffic load. However, a buffered repeater includes a memory buffer area into
which the excessive 100-Mbps transmission (1100 − 1000) rapidly fills. To
insure that its buffers do not overflow, the buffered repeater will issue a flow
control signal telling the transmitters to reduce their transmission rate. This
flow control signal will occur once the occupancy of the buffer reaches a
predefined level of occupancy, as indicated in the right portion of Figure 3.52.
Once data from the buffer is serviced to the point where occupancy is at a
predefined low level, the buffered distributor will use flow control to enable
transmission to the distributor by disabling flow control.
To provide an equitable method for sharing the 1-Gbps bus, the buffered
distributor uses a round-robin algorithm to determine which port can transmit
onto the bus. This round-robin method occurs one frame at a time, with
port 1 serviced, followed by port 2, and so on. Through the use of a buffered
distributor, it becomes possible to support applications that require a many-toone transmission capability, such as a LAN with a local server used for graphic
database queries, e-mail, and similar applications. Instead of migrating the
LAN to a switch environment, a buffered distributor may represent another
possible network solution to network bottlenecks for power users.
3.6 10 Gigabit Ethernet
The focus of this section is upon the next generation Ethernet standard, which
increases the CSMA/CD operating rate to 10 Gbps. In this section we will
examine the rationale for 10 Gbps Ethernet, its architecture and expected
operating rates.
Rationale
For many organizations it is hard to imagine the need for transporting data
at 1 Gbps. For other organizations Gigabit Ethernet represents a needed but
insufficient transport capability to satisfy existing or evolving communications
requirements. As we noted in the prior section in this chapter, the use of
1000BASE-LH provides a transmission distance of 70 km and extends Gigabit
Ethernet into the wide area network. This makes Gigabit Ethernet a viable
transport for use by communications carriers. However, because such carriers
transport information from numerous organizations over the WAN, Gigabit
150
chapter three
Ethernet may not provide sufficient capacity. Similarly, when used by an
Internet Service Provider (ISP) in a LAN environment, the growth in the use
of the Internet has placed a considerable load on Gigabit networks. As a result
of the preceding as well as other networking examples, the IEEE formed a High
Speed Study Group (HSSG) during 1999 to develop a standard for 10 Gbps
Ethernet. The primary objectives of the study group included maintaining the
Ethernet frame format while increasing the operating rate of Gigabit Ethernet
by an order of magnitude at a cost between two to three times the cost of
Gigabit Ethernet.
At the time this book revision occurred the evolving 10 Gbps Ethernet
standard was being developed for use over multi-mode and single-mode
optical fiber since the data rate was too high to be transported over copper
media. Initially developed specifications are for full-duplex operation using
an architecture similar to that developed for Gigabit Ethernet.
Architecture
The architecture of 10 Gigabit Ethernet is similar to that used for Gigabit
Ethernet. Figure 3.53 illustrates the sublayers of 10 Gigabit Ethernet compared
to the lower two layers of the ISO OSI Reference Model. In comparing Gigabit
and 10 Gigabit Ethernet you will note the primary difference is the use of
Gigabit ethernet sublayers
ISO OSI
Reference model
(lower 2 layers)
Logical Link Control (LLC)
Data link
Media Access Control (MAC)
Reconciliation
10 Gigabit Media Independent
Interface (10GMII)
Physical
Physical Coding Sublayer (PCS)
Physical Medium Attachment (PMA)
Physical Medium Dependent (PMD)
Medium Dependent Interface (MDI)
Figure 3.53
10 Gigabit Ethernet architecture.
ethernet networks
151
a 10 Gigabit Ethernet Media Independent Interface (10 GMII) for 10 Gigabit
Ethernet. Similar to the GMII used by Gigabit Ethernet, the 10 GMII provides
an interface between the MAC and the physical layer, isolating each from
one another.
A key difference between the 10 GMII and the GMII is the width of the
data path between the MAC sublayer and the physical layer. Under Gigabit
Ethernet the GMII data bus is 8 bits in width. Under 10 GMII a 32-bit data path
is used. Figure 3.54 illustrates the structure of the 10 GMII used for 10 Gigabit
Ethernet. Note that the actual signal interface is 72 bits in width. That signal
interface includes 32-bit data paths for transmit and receive and four control
bits to govern transmission and reception. In addition to the signal interface
the 10 GMII includes two clocking lines and a TX− word− hold lines.
The TX− word− hold line is used to support word-oriented pacing. The 32-bit
data paths provide transmit and receive functions that employ four control
bits, with one control bit used for each 8-bit data byte. A control bit is set
to 1 for denoting delimiters and special characters and to a binary 0 for data.
Examples of delimiter and special characters include:
♦
♦
♦
♦
IDLE used during inter-packet gap and when there is no data to transmit.
SOP used to indicate the start of each packet.
EOP used to indicate the end of each packet.
ERROR used to indicate the detection of an error in the received signal
or when an error condition needs to be transmitted.
Returning our attention to the architecture of 10 Gigabit Ethernet, the physical coding sublayer (PCS) is responsible for coding and decoding data streams
flowing to and from the MAC layer. The PCS layer uses 8B/10B coding
TX_word_hold
TX_clock
4 control bits
MAC
layer
32 data bits
Physical
layer
32 data bits
4 control bits
RX_clock
Figure 3.54
The 10 GMII.
152
chapter three
employed by Gigabit Ethernet, which was originally developed by IBM. In
comparison, the Physical Medium Attachment (PMA) sublayer is responsible
for taking parallel data and serializing code groups into a bit stream suitable for serial bit-oriented communications as well as performing the reverse
operation. In doing so the PMA is also responsible for providing applicable
synchronization. Below the PMA is the Physical Medium Dependent (PMD)
sublayer. The PMD is responsible for signal transmission to include amplification, modulation and wave shaping. Optical transceivers can be considered
to represent devices that operate at the PMD sublayer. Different PMD sublayers can be expected to be employed to support different types of media.
In fact, at the time this book revision was prepared the IEEE 802.3ae Task
Force had developed a draft standard that requires the support of four distinct link distances over different types of optical fiber. To meet the Task
Force distance objectives required four PMDs, as indicated in Table 3.14. In
examining the entries in Table 3.14 you will note the Task Force selected two
versions of wide wavelength division multiplexing. A 1310 nanometer (nm)
version was selected for use over single-mode fiber to provide a transmission
distance of 10 km, while a 1310 nm PMD was selected to provide a drive distance of 300 meters over multimode fiber. The latter PMD recognizes the fact
that many organizations have existing multimode fiber and its replacement
would represent an expenditure of time and material, so the ability to run
10 Gigabit Ethernet over existing multimode fiber became an objective of the
IEEE Task Force.
The use of 850 nm laser over multimode fiber also provides organizations
with an upgrade path without having to change their cabling infrastructure.
The selection of a 1550 nm laser to provide a link transmission of 40 km
provides a mechanism to use 10 Gigabit Ethernet in metropolitan as well as
private long-distance applications.
TABLE 3.14
10 Gigabit PMD Support
Optical
Transceiver
Type of Fiber
Supported
Target Transmission
Distance
850 nm serial
Multi-mode
65 m
1310 nm WWDM
Multi-mode
300 m
1310 nm WWDM
Single mode
10 km
1310 nm serial
Single mode
10 km
1550 nm serial
Single mode
40 km
ethernet networks
153
Returning our attention to the 10 Gigabit Ethernet architecture shown in
Figure 3.53, the Media Dependent Interface (MDI) provides the connector to a
particular media.
Operating Rates
While it would appear that the specifications of a data rate for 10 Gbps Ethernet
would be simple task, in actuality it wasn’t. While proponents of the use of
10 Gbps Ethernet in a LAN environment want the data rate to be ten times
the 1 Gbps Ethernet rate, proponents of the use of 10 Gbps Ethernet in the
WAN prefer the optical carrier (OC)-192 data payload rate of approximately
9.29 Gbps. Due to the diligent requirements of LAN and WAN proponents of
10 Gigabit Ethernet, it appears that both operating rates will be supported by
the evolving standard.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter four
Frame Operations
In this chapter, we will first focus our attention on the composition of
different types of Ethernet frames. In reality, there is only one physical
Ethernet frame. However, the composition of the frame was altered by the
IEEE when the CSMA/CD original Ethernet frame format was standardized
by that organization as the 802.3 frame. In addition, the logical composition
of the data field within the 802.3 frame can vary based upon the protocol
transported.
Once we obtain an understanding of the composition of Ethernet and IEEE
802.3 frames, we will examine the function of fields within each frame, as well
as the manner in which the placement of frames on the media is controlled — a
process known as media access control. After this is accomplished, we will
turn our attention to the manner by which protocol information is carried
within an IEEE 802.3 frame to include logical link control (LLC), NetWare,
TCP/IP, and other protocols.
Although the composition of the IEEE 802.3 frame remains the same from
10-Mbps through 10-Gbps operations, there are certain restrictions on the
framing of frames and the minimum size of the information field at different
network operating rates. Thus, in the last two sections in this chapter, we will
examine certain modifications and restrictions on frames transported on Fast
Ethernet and Gigabit and 10 Gigabit Ethernet networks.
4.1 Frame Composition
Figure 4.1 illustrates the general frame composition of Ethernet and IEEE
802.3 frames. You will note that they differ slightly. An Ethernet frame contains an eight-byte preamble, while the IEEE 802.3 frame contains a seven-byte
preamble followed by a one-byte start-of-frame delimiter field. A second difference between the composition of Ethernet and IEEE 802.3 frames concerns
155
156
chapter four
Ethernet
Preamble
8 bytes
Destination
address
6 bytes
Source
address
6 bytes
Type
2 bytes
Data
Frame check
sequence
46–1500 bytes 4 bytes
IEEE 802.3
Preamble
7 bytes
Start of
frame
delimiter
1 byte
Figure 4.1
Destination
address
2/6 bytes
Source
Length
address
2/6 bytes
2 bytes
Data
Frame check
sequence
46–1500 bytes 4 bytes
Ethernet and IEEE 802.3 frame formats.
the two-byte Ethernet type field. That field is used by Ethernet to specify
the protocol carried in the frame, enabling several protocols to be carried
independently of one another. Under the IEEE 802.3 frame format, the type
field was replaced by a two-byte length field, which specifies the number of
bytes that follow that field as data.
The differences between Ethernet and IEEE 802.3 frames, while minor, make
the two incompatible with one another. This means that your network must
contain either all Ethernet-compatible NICs or all IEEE 802.3–compatible
NICs. Fortunately, the fact that the IEEE 802.3 frame format represents a
standard means that almost all vendors now market 802.3-compliant hardware
and software. Although a few vendors continue to manufacture Ethernet or
dual functioning Ethernet/IEEE 802.3 hardware, such products are primarily
used to provide organizations with the ability to expand previously developed
networks without requiring the wholesale replacement of NICs. Although the
IEEE 802.3 frame does not directly support a type field within the frame, as we
will note in Section 4 in this chapter, the IEEE defined a special type of frame
to obtain compatibility with Ethernet LANs. That frame is referred to as an
Ethernet Subnetwork Access Protocol (Ethernet-SNAP) frame, which enables
a type subfield to be included in the data field. While the IEEE 802.3 standard
has essentially replaced Ethernet, because of their similarities and the fact
that 802.3 was based upon Ethernet, we will consider both to be Ethernet.
Now that we have an overview of the structure of Ethernet and 802.3 frames,
let’s probe more deeply and examine the composition of each frame field. We
will take advantage of the similarity between Ethernet and IEEE 802.3 frames
to examine the fields of each frame on a composite basis, noting the differences
between the two when appropriate.
frame operations
157
Preamble Field
The preamble field consists of eight (Ethernet) or seven (IEEE 802.3) bytes of
alternating 1 and 0 bits. The purpose of this field is to announce the frame
and to enable all receivers on the network to synchronize themselves to the
incoming frame.
Start-of-Frame Delimiter Field
This field is applicable only to the IEEE 802.3 standard and can be viewed as
a continuation of the preamble. In fact, the composition of this field continues
in the same manner as the format of the preamble, with alternating 1 and
0 bits used for the first six bit positions of this one-byte field. The last two
bit positions of this field are 11 — this breaks the synchronization pattern and
alerts the receiver that frame data follows.
Both the preamble field and the start-of-frame delimiter field are removed
by the controller when it places a received frame in its buffer. Similarly, when
a controller transmits a frame, it prefixes the frame with those two fields (if it
is transmitting an IEEE 802.3 frame) or a preamble field (if it is transmitting a
true Ethernet frame).
Destination Address Field
The destination address identifies the recipient of the frame. Although this
may appear to be a simple field, in reality its length can vary between IEEE
802.3 and Ethernet frames. In addition, each field can consist of two or
more subfields, whose settings govern such network operations as the type of
addressing used on the LAN, and whether the frame is addressed to a specific
station or more than one station. To obtain an appreciation for the use of this
field, let’s examine how this field is used under the IEEE 802.3 standard as
one of the two field formats applicable to Ethernet.
Figure 4.2 illustrates the composition of the source and destination address
fields. As indicated, the two-byte source and destination address fields are
applicable only to IEEE 802.3 networks, while the six-byte source and destination address fields are applicable to both Ethernet and IEEE 802.3 networks.
A user can select either a two- or six-byte destination address field; however, with IEEE 802.3 equipment, all stations on the LAN must use the same
addressing structure. Today, almost all 802.3 networks use six-byte addressing, because the inclusion of a two-byte field option was designed primarily
to accommodate early LANs that use 16-bit address fields.
158
chapter four
I/G
*
15 address bits
16 bit address field
(a) 2 byte field (IEEE 802.3)
I/G
*
U/L
46 address bits
48 bit address field
(b) 6 byte field (ethernet and IEEE 802.3)
I/G bit subfield '0' = individual address '1' = group address
U/L bit subfield '0' = universally administrated addressing
'1' = locally administrated addressing
* Set to '0' in source address field
Figure 4.2
Source and destination address field formats.
Both destination and source addresses are normally displayed by network
monitors in hexadecimal, with the first three bytes separated from the last
three by a colon (:) when six-byte addressing is used. For example, the source
address 02608C876543 would be displayed as 02608C:876543. As we will
shortly note, the first three bytes identify the manufacturer of the adapter
card, while the following three bytes identify a specific adapter manufactured
by the vendor identified by the first three bytes or six hex digits.
I/G Subfield
The one-bit I/G subfield is set to a 0 to indicate that the frame is destined to
an individual station, or 1 to indicate that the frame is addressed to more than
one station — a group address. One special example of a group address is the
assignment of all 1s to the address field. Hex ‘‘FFFFFFFFFFFF’’ is recognized
as a broadcast address, and each station on the network will receive and
accept frames with that destination address.
An example of the use of a broadcast destination address is the service
advertising packet (SAP) transmitted every 60 seconds by NetWare servers.
The SAP is used to inform other servers and workstations on the network of
the presence of that server. Because the SAP uses a destination address of
FF-FF-FF-FF-FF-FF, it is recognized by every node on the network.
frame operations
159
When a destination address specifies a single station, the address is referred
to as a unicast address. A group address that defines multiple stations is
known as a multicast address, while a group address that specifies all stations
on the network is, as previously mentioned, referred to as a broadcast address.
U/L Subfield
The U/L subfield is applicable only to the six-byte destination address field.
The setting of this field’s bit position indicates whether the destination address
is an address that was assigned by the IEEE (universally administered) or
assigned by the organization via software (locally administered).
Universal versus Locally Administered Addressing
Each Ethernet NIC contains a unique address burned into its read-only memory
(ROM) at the time of manufacture. To ensure that this universally administered
address is not duplicated, the IEEE assigns blocks of addresses to each
manufacturer. These addresses normally include a three-byte prefix, which
identifies the manufacturer and is assigned by the IEEE, and a three-byte
suffix, which is assigned by the adapter manufacturer to its NIC. For example,
the prefix 02608C identifies an NIC manufactured by 3Com, while a prefix of
hex 08002 identifies an NIC manufactured by Digital Equipment Company,
which was acquired by compaq computer.
Although the use of universally administered addressing eliminates the
potential for duplicate network addresses, it does not provide the flexibility
obtainable from locally administered addressing. For example, under locally
administered addressing, you can configure mainframe software to work with
a predefined group of addresses via a gateway PC. Then, as you add new
stations to your LAN, you simply use your installation program to assign a
locally administered address to the NIC instead of using its universally administered address. As long as your mainframe computer has a pool of locally
administered addresses that includes your recent assignment, you do not have
to modify your mainframe communications software configuration. Because
the modification of mainframe communications software typically requires
recompiling and reloading, the attached network must become inoperative for
a short period of time. Because a large mainframe may service hundreds to
thousands of users, such changes are normally performed late in the evening or
on a weekend. Thus, the changes required for locally administered addressing
are more responsive to users accessing certain types of mainframe computers
than those required for universally administered addressing.
160
chapter four
Source Address Field
The source address field identifies the station that transmitted the frame. Like
the destination address field, the source address can be either two or six bytes
in length.
The two-byte source address is supported only under the IEEE 802.3 standard and requires the use of a two-byte destination address; all stations on
the network must use two-byte addressing fields. The six-byte source address
field is supported by both Ethernet and the IEEE 802.3 standard. When a
six-byte address is used, the first three bytes represent the address assigned
by the IEEE to the manufacturer for incorporation into each NIC’s ROM. The
vendor then normally assigns the last three bytes for each of its NICs.
Table 4.1 lists the NIC identifiers for 85 Ethernet card manufacturers.
Note that many organizations including Cisco Systems, 3Com, IBM, MIPS,
Ungermann-Bass, and Data General were assigned two or more blocks of
addresses by the IEEE. Also note that organizations listed in Table 4.1 range
in scope from well-known communications and computer manufacturers to
universities and even a commercial firm probably best known for its watch
commercials. The entries in Table 4.1 represent a portion of three-byte identifiers assigned by the IEEE over the past decade and do not include identifiers
currently assigned to all vendors. For a comprehensive list of currently
assigned three-byte identifiers, readers should contact the IEEE. You can
contact the IEEE at:
IEEE Standards Department
445 Hoes Lane
P.O. Box 1331
Piscataway, NJ 08855
Telephone: +1 (732) 562-3813
Fax: +1 (732) 562-1571
Many software- and hardware-based network analyzers include the capability to identify each station on a LAN, count the number of frames transmitted
by the station and destined to the station, as well as identify the manufacturer of the NIC used in the station. Concerning the latter capability, this is
accomplished by the network analyzer containing a table of three-byte identifiers assigned by the IEEE to each NIC manufacturer, along with the name of
the manufacturer. Then the analyzer compares the three-byte identifier read
from frames flowing on the network and compares each identifier with the
frame operations
TABLE 4.1
Representative Ethernet NIC Manufacturer IDs
NIC Manufacturer
Three-Byte Identifier
Cisco
00-00-0C
Fujitsu
00-00-0E
Cabletron
00-00-1D
NeXT
00-00-0F
TRW
00-00-2A
Hughes LAN Systems
(formerly Sytek)
00-00-10
Tektronix
00-00-11
Datapoint Corporation
00-00-15
Olicom
00-00-24
AT&T
00-00-3D
NEC
00-00-4C
Network General
00-00-65
MIPS
00-00-6B
Madge Networks
00-00-6F
MIPS
00-00-77
Proteon
00-00-93
Cross Com Communications
(now part of Olicom)
00-00-98
Wellfleet (now Bay Networks)
00-00-A2
Xerox
00-00-AA
RND (RAD Network Devices)
00-00-B0
Western Digital
00-00-C0
Emulex
00-00-C9
Develcon Electronics, Ltd.
00-00-D0
Adaptec, Inc.
00-00-D1
Gandalf Data Ltd.
00-00-F3
Allied Telesis, Inc.
00-00-F4
161
162
chapter four
TABLE 4.1
(Continued)
NIC Manufacturer
Three-Byte Identifier
Racal Datacom
00-07-01
XYlan
00-20-DA
Crescendo (now owned by Cisco)
00-40-OB
Ascom
00-40-15
AST Research
00-40-1C
Netcomm
00-40-28
Nokia Data Communications
00-40-43
Cable and Wireless
00-40-74
AMP Incorporated
00-40-76
DigiBoard
00-40-9D
Cray Research, Inc.
00-40-A6
Mocom Communications Corp.
00-40-C5
PlainTree Systems, Inc.
00-40-EA
3Com
00-60-08
Cisco Systems, Inc.
00-60-09
Cisco Systems, Inc.
00-60-2F
Cisco Systems, Inc.
00-60-3E
Cisco Systems, Inc.
00-60-5C
Cisco Systems, Inc.
00-60-70
Cisco Systems, Inc.
00-60-83
3Com
00-60-8C
3Com
00-60-97
Hewlett-Packard
00-60-BO
Thomas Conrad Corporation
00-80-13
Bell Atlantic
00-80-1A
Newbridge Networks Corporation
00-80-21
Kalpana (acquired by Cisco)
00-80-24
frame operations
TABLE 4.1
(Continued)
NIC Manufacturer
Three-Byte Identifier
University of Toronto
00-80-46
Compaq Computer Corporation
00-80-5F
Nippon Steel Corporation
00-80-6E
Xircom, Inc.
00-80-C7
Shiva
00-80-D3
Zenith Communications Products
00-80-F7
Azure Technologies, Inc.
00-80-FE
Bay Networks
00-A0-00
National Semiconductor
00-A0-D1
Allied Telesyn
00-A0-D2
Intel
00-AA-00
Ungermann-Bass
00-DD-00
Ungermann-Bass
00-DD-01
Racal Interlan
02-07-01
3Com
02-60-8C
BBN
08-00-08
Hewlett Packard
08-00-09
Unisys
08-00-0B
Tektronix
08-00-11
Data General
08-00-A
Data General
08-00-1B
Sun
08-00-20
DEC
08-00-2B
Bull
08-00-38
Sony
08-00-46
Sequent
08-00-47
Stanford University
08-00-56
163
164
chapter four
TABLE 4.1
(Continued)
NIC Manufacturer
Three-Byte Identifier
IBM
08-00-5A
Silicon Graphics
08-00-69
Silicon Graphics
08-00-79
Seiko
08-00-83
Excelan
08-00-6E
Danish Data Elektronix
08-00-75
AT&T
80-00-10
identifiers stored in its identifier table. By providing information concerning
network statistics, network errors, and the vendor identifier for the NIC in
each station, you may be able to isolate problems faster or better consider
future decisions concerning the acquisition of additional NICs.
An example of the use of NIC manufacturer IDs can be obtained by examining two monitoring screen displays of the Triticom EtherVision network
monitoring and analysis program. Figure 4.3 illustrates the monitoring screen
during the program’s autodiscovery process. During this process the program
reads the source address of each frame transmitted on the segment that the
computer executing the program is connected to. Although obscured by the
highlighted bar, the first three bytes of the adapter address first discovered
is 00-60-8C, which represents a block of addresses assigned by the IEEE to 3
Com Corporation. If you glance at the first column in Figure 4.3, you will note
that the second row, fourth row, ninth row, and a few additional rows also
have NIC addresses that commence with hex 00-60-8C. By pressing the F2 key
the program will display the manufacturer of each NIC encountered and for
which statistics are being accumulated. This is indicated in Figure 4.4, which
shows the first three bytes of each address replaced by the vendor assigned
to the appropriate manufacturer ID. Thus, rows 1, 4, 9, and a few other rows
commence with ‘‘3Com’’ to indicate the manufacturer of the NIC.
Organizations can request the assignment of a vendor code by contacting
the IEEE Registration Authority at the previously listed address for the IEEE
provided in this section. A full list of assigned vendor codes is obtainable by
FTP at ftp.ieee.org as the file ieee/info/info.stds.oui. Readers should note that
the list is limited to those companies that agreed to make their vendor code
assignment(s) public.
frame operations
165
Figure 4.3 The Triticom EtherVision source address monitoring feature discovers the hardware address of each NIC. At the time this screen was captured
16 stations were identified.
Type Field
The two-byte type field is applicable only to the Ethernet frame. This field
identifies the higher-level protocol contained in the data field. Thus, this field
tells the receiving device how to interpret the data field.
Under Ethernet, multiple protocols can exist on the LAN at the same time.
Xerox served as the custodian of Ethernet address ranges licensed to NIC
manufacturers and defined the protocols supported by the assignment of type
field values.
Table 4.2 lists 31 of the more common Ethernet type field assignments.
To illustrate the ability of Ethernet to transport multiple protocols, assume
a common LAN was used to connect stations to both UNIX and NetWare
servers. Frames with the hex value 0800 in the type field would identify the
IP protocol, while frames with the hex value 8137 in the type field would
166
chapter four
Figure 4.4 By pressing the F2 key, EtherVision will convert the three-byte
hex NIC manufacturer ID to the vendor name or an appropriate mnemonic.
identify the transport of IPX and SPX protocols. Thus, the placement of an
appropriate hex value in the Ethernet type field provides a mechanism to
support the transport of multiple protocols on the local area network.
Under the IEEE 802.3 standard, the type field was replaced by a length field,
which precludes compatibility between pure Ethernet and 802.3 frames.
Length Field
The two-byte length field, applicable to the IEEE 802.3 standard, defines the
number of bytes contained in the data field. Under both Ethernet and IEEE
802.3 standards, the minimum size frame must be 64 bytes in length from
preamble through FCS fields. This minimum size frame ensures that there
is sufficient transmission time to enable Ethernet NICs to detect collisions
accurately, based on the maximum Ethernet cable length specified for a
network and the time required for a frame to propagate the length of the cable.
frame operations
TABLE 4.2
Representative Ethernet Type Field Assignments
Protocol
Hex Value Assigned
Experimental
0101-DIFF
Xerox XNS
0600
IP
0800
X.75 Internet
0801
NBS Internet
0802
ECMA Internet
0803
CHAOSmet
0804
X.25 Level 3
0805
Address Resolution Protocol
0806
XNS Compatibility
0807
Banyan Systems
0BAD
BBN Simnet
5208
DEC MOP Dump/Load
6001
DEC MOP Remote Console
6002
DEC DECNET Phase IV Route
6003
DEC LAT
6004
DEC Diagnostic Protocol
6005
3Com Corporation
6010–6014
Proteon
7030
AT&T
8008
Excelan
8010
Tymshare
802E
DEC LANBridge
8038
DEC Ethernet Encryption
803D
AT&T
8046–8047
AppleTalk
809B
IBM SNA Service on Ethernet
80D5
AppleTalk ARP
80F3
Wellfleet
80FF–8103
NetWare IPX/SPX
8137–8138
SNMP
814C
167
168
chapter four
Based on the minimum frame length of 64 bytes and the possibility of using
two-byte addressing fields, this means that each data field must be a minimum
of 46 bytes in length. The only exception to the preceding involves Gigabit
Ethernet. At a 1000-Mbps operating rate the original 802.3 standard would
not provide a frame duration long enough to permit a 100-meter cable run
over copper media. This is because at a 1000-Mbps data rate there is a high
probability that a station could be in the middle of transmitting a frame before
it becomes aware of any collision that might have occurred at the other end
of the segment. Recognizing this problem resulted in the development of a
carrier extension, which extends the minimum Ethernet frame to 512 bytes.
The carrier extension is discussed in detail in Section 4.6 when we turn our
attention to the Gigabit Ethernet carrier extension.
For all versions of Ethernet except Gigabit Ethernet, if data being transported
is less than 46 bytes, the data field is padded to obtain 46 bytes. However, the
number of PAD characters is not included in the length field value. NICs that
support both Ethernet and IEEE 802.3 frame formats use the value in this field
to distinguish between the two frames. That is, because the maximum length
of the data field is 1,500 bytes, a value that exceeds hex 05DC indicates that
instead of a length field (IEEE 802.3), the field is a type field (Ethernet).
Data Field
As previously discussed, the data field must be a minimum of 46 bytes in
length to ensure that the frame is at least 64 bytes in length. This means that
the transmission of 1 byte of information must be carried within a 46-byte
data field; if the information to be placed in the field is less than 46 bytes, the
remainder of the field must be padded. Although some publications subdivide
the data field to include a PAD subfield, the latter actually represents optional
fill characters that are added to the information in the data field to ensure a
length of 46 bytes. The maximum length of the data field is 1500 bytes.
Frame Check Sequence Field
The frame check sequence field, applicable to both Ethernet and the IEEE
802.3 standard, provides a mechanism for error detection. Each transmitter
computes a cyclic redundancy check (CRC) that covers both address fields, the
type/length field, and the data field. The transmitter then places the computed
CRC in the four-byte FCS field.
The CRC treats the previously mentioned fields as one long binary number.
The n bits to be covered by the CRC are considered to represent the coefficients
frame operations
169
of a polynomial M(X) of degree n − 1. Here, the first bit in the destination
address field corresponds to the Xn−1 term, while the last bit in the data field
corresponds to the X0 term. Next, M(X) is multiplied by X32 , and the result of
that multiplication process is divided by the following polynomial:
G(X) = X32 +X26 +X23 +X22 +X16 +X12 +X11 +X10 +X8 +X7 +X5 +X4 +X2 +X+1
Note that the term Xn represents the setting of a bit to a 1 in position n. Thus,
part of the generating polynomial X5 + X4 + X2 + X1 represents the binary
value 11011.
This division produces a quotient and remainder. The quotient is discarded,
and the remainder becomes the CRC value placed in the four-byte FCS field.
This 32-bit CRC reduces the probability of an undetected error to 1 bit in every
4.3 billion, or approximately 1 bit in 232 − 1 bits.
Once a frame reaches its destination, the receiver uses the same polynomial
to perform the same operation upon the received data. If the CRC computed
by the receiver matches the CRC in the FCS field, the frame is accepted.
Otherwise, the receiver discards the received frame, as it is considered to have
one or more bits in error. The receiver will also consider a received frame to
be invalid and discard it under two additional conditions. Those conditions
occur when the frame does not contain an integral number of bytes, or when
the length of the data field does not match the value contained in the length
field. The latter condition obviously is only applicable to the 802.3 standard,
because an Ethernet frame uses a type field instead of a length field.
Interframe Gap
Under the 10-Mbps versions of the CSMA/CD protocol a 9.6 microsecond
(µs) quiet time occurs between transmitted frames. This quiet time, which
is referred to as an interframe gap, permits clocking circuitry used within
repeaters and workstations and hub ports to be resynchronized to the known
local clock. Under Fast Ethernet the interframe gap is 0.96 ms, while under
Gigabit Ethernet the gap is reduced to 0.096 ms.
4.2 Media Access Control
In the first section in this chapter, we examined the frame format by which
data is transported on an Ethernet network. Under the IEEE 802 series of
10-Mbps operating standards, the data link layer of the OSI Reference Model
170
chapter four
is subdivided into two sublayers — logical link control (LLC) and medium
access control (MAC). The frame formats examined in Section 4.1 represent
the manner in which LLC information is transported. Directly under the LLC
sublayer is the MAC sublayer. The MAC sublayer, which is the focus of this
section, is responsible for checking the channel and transmitting data if the
channel is idle, checking for the occurrence of a collision, and taking a series
of predefined steps if a collision is detected. Thus, this layer provides the
required logic to control the network.
Figure 4.5 illustrates the relationship between the physical and LLC layers
with respect to the MAC layer. The MAC layer is an interface between user
data and the physical placement and retrieval of data on the network. To better
understand the functions performed by the MAC layer, let us examine the
four major functions performed by that layer — transmitting data operations,
transmitting medium access management, receiving data operations, and
receiving medium access management. Each of those four functions can be
viewed as a functional area, because a group of activities is associated with
Transmit
Medium
access control
Transmit
data
operations
Medium
access control
Transmit
medium access
management
Data
encoding
Receive
LLC data
Physical
layer
Receive
data
operations
Receive
medium access
management
Data
decoding
Channel
Figure 4.5 Medium access control. The medium access control (MAC) layer
can be considered an interface between user data and the physical placement
and retrieval of data on the network.
frame operations
TABLE 4.3
Transmit data
operations
171
MAC Functional Areas
♦ Accept data from the LLC sublayer and construct a frame by
appending preamble and start-of-frame delimiter; insert
destination and source address, length count; if frame is less
than 64 bytes, insert sufficient PAD characters in the data
field.
♦ Calculate the CRC and place in the FCS field.
Transmit
media access
management
♦ Defer transmission if the medium is busy.
♦ Delay transmission for a specified interframe gap period.
♦ Present a serial bit stream to the physical layer for
transmission.
♦ Halt transmission when a collision is detected.
♦ Transmit a jam signal to ensure that news of a collision
propagates throughout the network.
♦ Reschedule retransmissions after a collision until
successful, or until a specified retry limit is reached.
Receive data
operations
♦ Discard all frames not addressed to the receiving station.
♦ Recognize all broadcast frames and frames specifically
addressed to station.
♦ Perform a CRC check.
♦ Remove preamble, start-of-frame delimiter, destination and
source addresses, length count, and FCS; if necessary,
remove PAD fill characters.
♦ Pass data to LLC sublayer.
Receive
media access
management
♦ Receive a serial bit stream from the physical layer.
♦ Verify byte boundary and length of frame.
♦ Discard frames not an even eight bits in length or less than
the minimum frame length.
each area. Table 4.3 lists the four MAC functional areas and the activities
associated with each area. Although the transmission and reception of data
operations activities are self-explanatory, the transmission and reception of
media access management require some elaboration. Therefore, let’s focus our
attention on the activities associated with each of those functional areas.
172
chapter four
Transmit Media Access Management
CSMA/CD can be described as a listen-before-acting access method. Thus,
the first function associated with transmit media access management is to
find out whether any data is already being transmitted on the network and, if
so, to defer transmission. During the listening process, each station attempts
to sense the carrier signal of another station, hence the prefix carrier sense
(CS) for this access method. Although broadband networks use RF modems
that generate a carrier signal, a baseband network has no carrier signal in
the conventional sense of a carrier as a periodic waveform altered to convey
information. Thus, a logical question you may have is how the MAC sublayer
on a baseband network can sense a carrier signal if there is no carrier. The
answer to this question lies in the use of a digital signaling method, known as
Manchester encoding on 10-Mbps Ethernet LANs, that a station can monitor
to note whether another station is transmitting. Although NRZI encoding is
used on broadband networks, the actual data is modulated after it is encoded.
Thus, the presence or absence of a carrier is directly indicated by the presence
or absence of a carrier signal on a broadband network.
Collision Detection
As discussed in Chapter 3, under Manchester encoding, a transition occurs
at the middle of each bit period. This transition serves as both a clocking
mechanism, enabling a receiver to clock itself to incoming data, and as
a mechanism to represent data. Under Manchester coding, a binary 1 is
represented by a high-to-low transition, while a binary 0 is represented by
a low-to-high voltage transition. Thus, an examination of the voltage on the
medium of a baseband network enables a station to determine whether a
carrier signal is present.
If a carrier signal is found, the station with data to transmit will continue
to monitor the channel. When the current transmission ends, the station will
then transmit its data, while checking the channel for collisions. Because
Ethernet and IEEE 802.3 Manchester-encoded signals have a 1-volt average
DC voltage level, a collision results at an average DC level of 2 volts. Thus, a
transceiver or network interface card can detect collisions by monitoring the
voltage level of the Manchester line signal.
Jam Pattern
If a collision is detected during transmission, the transmitting station will
cease transmission of data and initiate transmission of a jam pattern. The jam
frame operations
173
pattern consists of 32 to 48 bits. These bits can have any value other than
the CRC value that corresponds to the partial frame transmitted before the
jam. The transmission of the jam pattern ensures that the collision lasts long
enough to be detected by all stations on the network.
When a repeater is used to connect multiple segments, it must recognize a
collision occurring on one port and place a jam signal on all other ports. Doing
so results in the occurrence of a collision with signals from stations that may
have been in the process of beginning to transmit on one segment when the
collision occurred on the other segment. In addition, the jam signal serves as
a mechanism to cause nontransmitting stations to wait until the jam signal
ends before attempting to transmit, alleviating additional potential collisions
from occurring.
Wait Time
Once a collision is detected, the transmitting station waits a random number of
slot times before attempting to retransmit. The term slot represents 512 bits on
a 10-Mbps network, or a minimum frame length of 64 bytes. The actual number
of slot times the station waits is selected by a randomization process, formally
known as a truncated binary exponential backoff. Under this randomization
process, a randomly selected integer r defines the number of slot times the
station waits before listening to determine whether the channel is clear. If it is,
the station begins to retransmit the frame, while listening for another collision.
If the station transmits the complete frame successfully and has additional
data to transmit, it will again listen to the channel as it prepares another frame
for transmission. If a collision occurs on a retransmission attempt, a slightly
different procedure is followed. After a jam signal is transmitted, the station
simply doubles the previously generated random number and then waits the
prescribed number of slot intervals before attempting a retransmission. Up to
16 retransmission attempts can occur before the station aborts the transmission
and declares the occurrence of a multiple collision error condition.
Figure 4.6 illustrates the collision detection process by which a station can
determine that a frame was not successfully transmitted. At time t0 both
stations A and B are listening and fail to detect the occurrence of a collision,
and at time t1 station A commences the transmission of a frame. As station A’s
frame begins to propagate down the bus in both directions, station B begins
the transmission of a frame, since at time t2 it appears to station B that there
is no activity on the network.
Shortly after time t2 the frames transmitted by stations A and B collide,
resulting in a doubling of the Manchester encoded signal level for a very short
174
chapter four
t0
A
Stations A & B listening
B
Station A begins transmission
A
B
t1
Station B begins transmission
t2
A
B
Station B detects collision and transmits pattern jam
A
B
t3
Station A detects collision before ending transmission
A
B
t4
Figure 4.6
Collision detection.
period of time. This doubling of the Manchester encoded signal’s voltage level
is detected by station B at time t3 , since station B is closer to the collision than
station A. Station B then generates a jam pattern that is detected by station A.
Late Collisions
A late collision is a term used to reference the detection of a collision only
after a station places a complete frame on the network. A late collision is
normally caused by an excessive network segment cable length, resulting in
the time for a signal to propagate from one end of a segment to another part of
the segment being longer than the time required to place a full frame on the
network. This results in two devices communicating at the same time never
seeing the other’s transmission until their signals collide.
A late collision is detected by a transmitter after the first slot time of
64 bytes and is applicable only for frames whose lengths exceed 65 bytes. The
detection of a late collision occurs in exactly the same manner as a normal
collision; however, it happens later than normal. Although the primary cause
of late collisions is excessive segment cable lengths, an excessive number of
repeaters, faulty connectors, and defective Ethernet transceivers or controllers
frame operations
175
can also result in late collisions. Many network analyzers provide information
on late collisions, which can be used as a guide to check the previously
mentioned items when late collisions occur.
Service Primitives
As previously mentioned, the MAC sublayer isolates the physical layer from
the LLC sublayer. Thus, one of the functions of the MAC sublayer is to provide
services to the LLC. To accomplish this task, a series of service primitives was
defined to govern the exchange of LLC data between a local MAC sublayer
and its peer LLC sublayer.
The basic MAC service primitives used in all IEEE MAC standards include
the medium access data request (MA− DATA.request), medium access data confirm (MA− DATA.confirm), medium access data indicate (MA− DATA.indicate),
and medium access data response (MA− DATA.response).
MA− DATA.request
The medium access data request is generated whenever the LLC sublayer
has data to be transmitted. This primitive is passed from layer n to layer
n − 1 to request the initiation of service, and results in the MAC sublayer
formatting the request in a MAC frame and passing it to the physical layer for
transmission.
MA− DATA.confirm
The medium access data confirm primitive is generated by the MAC sublayer
in response to an MA− DATA.request generated by the local LLC sublayer.
The confirm primitive is passed from layer n − 1 to layer n, and includes a
status parameter that indicates the outcome of the request primitive.
MA− DATA.indicate
The medium access data indicate primitive is passed from layer n − 1 to
layer n to indicate that a valid frame has arrived at the local MAC sublayer.
Thus, this service primitive denotes that the frame was received without CRC,
length, or frame-alignment error.
MA− DATA.response
The medium access data response primitive is passed from layer n to layer
n − 1. This primitive acknowledges the MA− DATA.indicate service primitive.
176
chapter four
Primitive Operations
To illustrate the use of MAC service primitives, let us assume that station A on a network wants to communicate with station B. As illustrated in
Figure 4.7, the LLC sublayer of station A requests transmission of a frame to
the MAC sublayer service interface via the issuance of an MA− DATA.request
service primitive. In response to the MA− DATA.request, a frame is transmitted to station B. Upon receipt of that frame, the MAC sublayer at that
station generates an MA− DATA.indicate to inform the LLC sublayer of the
arrival of the frame. The LLC sublayer accepts the frame and generates an
MA− DATA.response to inform the MAC sublayer that it has the frame. That
response flows across the network to station A, where the MAC sublayer generates an MA− DATA.confirm to inform the LLC sublayer that the frame was
received without error.
Half- versus Full-duplex Operation
Ethernet was originally designed as a half-duplex LAN transmission method.
The CSMA/CD algorithm required the receive pair in the two pair wiring used
for 10BASE-T to be used both to receive data and to detect collisions. In fact, if
the transmit and receive wire pairs became simultaneously activated, the MAC
Station
A
Station
B
MAC
service
interface
MAC
service
interface
MA_DATA.request
N
E
T
W
O
R
K
MA_DATA.indicate
MA_DATA.response
MA_DATA.confirm
Figure 4.7
Relationship of medium access control service primitives.
frame operations
177
layer would cause the ongoing transmission to terminate and would initiate
the previously described truncated binary exponential backoff algorithm.
With the development of Ethernet switches during the mid-1980s, it became
possible to cable a station to a switch port directly. When this operation
occurred it eliminated the possibility of a collision. Recognizing the fact that
the CSMA/CD algorithm was not efficient for use in a switch environment,
the IEEE assigned a task force to examine modifying the MAC layer for switch
operations. In 1987 the IEEE 802.3x standard was approved; this introduced
a modified MAC layer that added support for full duplex operations in
a switch environment. A related standard, referred to as the IEEE 802.3y
specification, defines a flow control mechanism, which is important when
devices with dissimilar operating rates communicate with one another through
a switch, such as a server operating at 100 Mbps communicating with a
workstation operating at 10 Mbps. Although the Ethernet switch will include
buffer memory, to preclude such memory from being filled and subsequent
data transmitted by the server being lost, the switch will initiate flow control to
the server to temporarily stop its transmission. Once sufficient buffer memory
is available in the switch, flow control will be disabled. Later in this book we
will examine flow control in more detail.
4.3 Logical Link Control
As discussed in Chapter 2, the LLC sublayer was defined under the IEEE 802.2
standard to make the method of link control independent of a specific access
method. Thus, the 802.2 method of link control spans Ethernet (IEEE 802.3),
Token Bus (IEEE 802.4), and Token-Ring (IEEE 802.5) local area networks.
Functions performed by the LLC include generating and interpreting commands to control the flow of data, including recovery operations for when a
transmission error is detected.
Link control information is carried within the data field of an IEEE 802.3
frame as an LLC protocol data unit (PDU). Figure 4.8 illustrates the relationship
between the IEEE 802.3 frame and the LLC PDU.
As discussed in Chapter 2, service access points (SAPs) function much like
a mailbox. Because the LLC layer is bounded below the MAC sublayer and
bounded above by the network layer, SAPs provide a mechanism for exchanging information between the LLC layer and the MAC and network layers. For
example, from the network layer perspective, a SAP represents the place to
leave messages about the services requested by an application. There are two
broad categories of SAPs, IEEE-administered and manufacturer-implemented.
178
chapter four
Preamble
Start of
frame
delimiter
Destination
address
DSAP
Source Length
address
SSAP Control
Data
Frame check
sequence
Information
Figure 4.8 Formation of LLC protocol data unit. Control information is
carried within a MAC frame.
TABLE 4.4
Representative Examples of SAP Addresses
Address (Hex)
Assignment
IEEE-administered
00
Null SAP
02
Individual LLC sublayer management functions
06
ARPANET Internet Protocol (IP)
42
IEEE 802.1 Bridge-Spanning Tree Protocol
AA
Sub-Network Access Protocol (SNAP)
FE
ISO Network Layer Protocol
Manufacturer-implemented
80
Xerox Network Systems
BC
Banyan VINES
EO
Novell NetWare
FO
IBM NetBIOS
F8
IBM Remote Program Load (RPL)
FA
Ungermann-Bass
Table 4.4 provides six examples of each type of SAP. In examining the
entries in Table 4.4, the hex value AA represents one of the more commonly
used SAPs today. When that value is encoded in both DSAP and SSAP
fields, it indicates a special type of Ethernet frame referred to as an Ethernet
frame operations
179
SNAP frame. The SNAP frame, as we will shortly note when we cover
it in Section 4.4, unlike the Ethernet 802.3 frame, enables several different
protocols to be transported.
The destination services access point (DSAP) is one byte in length and is
used to specify the receiving network layer process. Because an IEEE 802.3
frame does not include a type field, the DSAP field is used to denote the
destination upper-layer protocol carried within the frame. For example, the
DSAP hex value E0 indicates that the data field contains NetWare data.
The source service access point (SSAP) is also one byte in length. The
SSAP specifies the sending network layer process. Because the destination
and source protocols must be the same, the value of the SSAP field will
always match the value of the DSAP field. Both DSAP and SSAP addresses
are assigned by the IEEE. For example, hex address ‘‘FF’’ represents a DSAP
broadcast address.
The control field contains information concerning the type and class of
service being used for transporting LLC data. For example, a hex value of
03 when NetWare is being transported indicates that the frame is using an
unnumbered format for connectionless services.
Types and Classes of Service
Under the 802.2 standard, there are three types of service available for sending and receiving LLC data. These types are discussed in the next three
paragraphs. Figure 4.9 provides a visual summary of the operation of each
LLC service type.
Type 1
Type 1 is an unacknowledged connectionless service. The term connectionless
refers to the fact that transmission does not occur between two devices as
if a logical connection were established. Instead, transmission flows on the
channel to all stations; however, only the destination address acts upon
the data. As the name of this service implies, there is no provision for the
acknowledgment of frames. Neither are there provisions for flow control or
for error recovery. Therefore, this is an unreliable service.
Despite those shortcomings, Type 1 is the most commonly used service,
because most protocol suites use a reliable transport mechanism at the transport layer, thus eliminating the need for reliability at the link layer. In addition,
by eliminating the time needed to establish a virtual link and the overhead
of acknowledgments, a Type 1 service can provide a greater throughput than
other LLC types of services.
180
chapter four
Type 1 unacknowledged connectionless service
PDU
A
B
Type 2 connection-oriented service
A
PDU
PDU
ACK
B
Type 3 acknowledged connectionless source
A
PDU
ACK
B
Legend:
PDU = Protocol data unit
ACK = Acknowledgment
A,B = Stations on the network
Figure 4.9
Local link control service types.
Type 2
The Type 2 connection-oriented service requires that a logical link be established between the sender and the receiver before information transfer. Once
the logical connection is established, data will flow between the sender and
receiver until either party terminates the connection. During data transfer, a
Type 2 LLC service provides all of the functions lacking in a Type 1 service,
using a sliding window for flow control. When IBM’s SNA data is transported
on a LAN, it uses connection-oriented services. Type 2 LLC is also commonly
referred to as LLC 2.
Type 3
The Type 3 acknowledged connectionless service contains provision for the
setup and disconnection of transmission; it acknowledges individual frames
using the stop-and-wait flow control method. Type 3 service is primarily
used in an automated factory process-control environment, where one central
computer communicates with many remote devices that typically have a
limited storage capacity.
Classes of Service
All logical link control stations support Type 1 operations. This level of
support is known as Class I service. The classes of service supported by
frame operations
181
LLC indicate the combinations of the three LLC service types supported by
a station. Class I supports Type 1 service, Class II supports both Type 1 and
Type 2, Class III supports Type 1 and Type 3 service, and Class IV supports
all three service types. Because service Type 1 is supported by all classes,
it can be considered a least common denominator, enabling all stations to
communicate using a common form of service.
Service Primitives
The LLC sublayer uses service primitives similar to those that govern the
exchange of data between the MAC sublayer and its peer LLC sublayer. In doing
so, the LLC sublayer supports the Request, Confirm, Indicate, and Response
primitives described in Section 4.2 of this chapter. The major difference
between the LLC and MAC service primitives is that the LLC sublayer supports
three types of services. As previously discussed, the available LLC services
are unacknowledged connectionless, connection-oriented, and acknowledged
connectionless. Thus, the use of LLC service primitives varies in conjunction
with the type of LLC service initiated. For example, a connection-oriented
service uses service primitives in the same manner as that illustrated in
Figure 4.7. If the service is unacknowledged connectionless, the only service
primitives used are the Request and Indicate, because there is no Response
nor Confirmation.
4.4 Other Ethernet Frame Types
Three additional frame types that warrant discussion are Ethernet-802.3,
Ethernet-SNAP, and the IEEE 802.1Q tagged frame. In actuality, the first two
types of frames represent a logical variation of the IEEE 802.3 frame, in which
the composition of the data field varies from the composition of the LLC
protocol data unit previously illustrated in Figure 4.8. The third type of frame
provides the ability to form virtual LANs (vLANs) as well as to assign a
priority level to a frame.
Ethernet-802.3
The Ethernet-802.3 frame represents a proprietary subdivision of the
IEEE 802.3 data field to transport NetWare. Ethernet-802.3 is one of several
types of frames that can be used to transport NetWare. The actual frame type
used is defined at system setup by binding NetWare to a specific type of frame.
182
chapter four
Preamble
Start of
frame
delimiter
Destination
address
Source
address
IPX header
Length
Data
Frame
check
sequence
Information
Figure 4.10 Novell’s NetWare Ethernet-802.3 frame. An Ethernet-802.3 frame
subdivides the data field into an IPX header field and an information field.
Figure 4.10 illustrates the format of the Ethernet-802.3 frame. Due to the
absence of LLC fields, this frame is often referred to as raw 802.3.
For those using or thinking of using NetWare, a word of caution is in order
concerning frame types. Novell uses the term Ethernet-802.2 to refer to the
IEEE 802.3 frame. Thus, if you set up NetWare for Ethernet-802.2 frames, in
effect, your network is IEEE 802.3–compliant.
Ethernet-SNAP
The Ethernet-SNAP frame, unlike the Ethernet-802.3 frame, can be used to
transport several protocols. AppleTalk Phase II, NetWare, and TCP/IP protocols can be transported due to the inclusion of an Ethernet type field in the
Ethernet-SNAP frame. Thus, SNAP can be considered as an extension that
permits vendors to create their own Ethernet protocol transports. EthernetSNAP was defined by the IEEE 802.1 committee to facilitate interoperability
between IEEE 802.3 LANs and Ethernet LANs. This was accomplished,
as we will soon note, by the inclusion of a type field in the EthernetSNAP frame.
Figure 4.11 illustrates the format of an Ethernet-SNAP frame. Although the
format of this frame is based upon the IEEE 802.3 frame format, it does not
use DSAP and SSAP mailbox facilities and the control field. Instead, it places
specific values in those fields to indicate that the frame is a SNAP frame.
The value hex AA is placed into the DSAP and SSAP fields, while hex 03 is
placed into the control field to indicate that a SNAP frame is being transported.
frame operations
Start of
frame
delimiter
Preamble
Destination
address
DSAP SSAP Control
Bytes
(1)
(1)
(1)
Figure 4.11
Source
Length
address
Data
183
Frame
check
sequence
Organization Ethernet
Information
type
code
(3)
(2)
Ethernet-SNAP frame format.
The hex 03 value in the control field defines the use of an unnumbered format,
which is the only format supported by a SNAP frame.
The three-byte organization code field references the organizational body
that assigned the value placed in the following field, the Ethernet type field.
A hex value of 00-00-00 in the organization code field indicates that Xerox
assigned the value in the Ethernet type field. In comparison, a hex value
of 08-00-07 would indicate Apple computer as the organizational body that
assigned the valve in the following field. Concerning that following field,
the SNAP frame uses two bytes to identify the protocol being transported,
which significantly extends the number of protocols that can be transported.
Although shown as the Ethernet Type field in Figure 4.11, the formal name for
this field is Protocol Identifier (PID). Through the use of the Ethernet-SNAP
frame, you obtain the ability to transport multiple protocols in a manner
similar to the original Ethernet frame that used the type field for this purpose.
Here the hex value of 00-00-00 in the organization code field enables the
values previously listed in Table 4.2 to represent different protocols carried
by the SNAP frame.
IEEE 802.1Q Frame
With the development of LAN switches it became possible to group workstations together based upon such criteria as their MAC address, their switch
port connection or even the higher layer network address assigned to the
workstation. The grouping of workstations resulted in the formation of a
virtual LAN.
Recognizing the importance of a standardized method for informing devices
of the association of frames with a particular vLAN, the IEEE formed a task
184
chapter four
force to work on the standardization effort, resulting in the development of a
‘‘tag’’ for identifying vLAN frames. At the same time as this effort commenced,
a separate task force was working on the prioritization of frames. The work
of the vLAN task force resulted in the specifications for what is referred to as
the 802.1Q frame header. That header incorporates a three-bit priority field
that is used to convey priorities specified by the task force that standardized
frame priority, which is the 802.1p standard. Thus, the 802.1Q frame header
can transport 802.1p priority information. Now that we know the IEEE Ps and
Qs, let’s examine the format associated with the 802.1Q frame header.
Figure 4.12 illustrates the format of an IEEE 802.1Q tagged frame. Note that
the tag is inserted after the source address field and consists of four fields.
The first field is the tag protocol identifier (TPI) and is two bytes in length.
The valve of the TPI field is set to 8100 to identify the frame as an 802.1Q
tagged frame. The second field is a three-bit priority field that can be used to
specify one of eight levels of priority (binary 000 to 111). The Priority field
is followed by a one-bit canonical format identifier (CFI). When set, this field
indicates that a Token-Ring frame is encapsulated within the tagged Ethernet
frame. The fourth field is the vLAN identification field (VID), which is 12 bits
in length. The value in this field uniquely identifies the vLAN to which the
frame belongs. Later in this book, when we examine LAN switches, we will
also examine the operation and utilization of vLANs.
Frame Determination
Through software, a receiving station can determine the type of frame and
correctly interpret the data carried in the frame. To accomplish this, the value
Tagged
ethernet
frame
(bytes)
Preamble
(8)
Tag
format
(bits)
Destination
address
(6)
Source
address
Tag protocol
identifier
(16)
Figure 4.12
802.1Q
Tag
(4)
Priority
(3)
Data
(46 −1500)
CFI
(1)
FCS
(2)
vLAN ID
(12)
The format of the IEEE 802.1Q tagged frame.
frame operations
185
of the two bytes that follow the source address is first examined. If the value
is greater than 1500, this indicates the occurrence of an Ethernet frame. As
previously noted, if the value is 8100, then the frame is an IEEE 802.1Q tagged
frame and software would look further into the tag to determine the vLAN
identification and other information. If the value is less than or equal to 1500,
the frame can be either a pure IEEE 802.3 frame or a variation of that frame.
Thus, more bytes must be examined.
If the next two bytes have the hex value FF:FF, the frame is a NetWare
Ethernet-802.3 frame. This is because the IPX header has the value hex FF:FF
in the checksum field contained in the first two bytes in the IPX header. If the
two bytes contain the hex value AA:AA, this indicates that it is an EthernetSNAP frame. Any other value determined to reside in those two bytes then
indicates that the frame must be an Ethernet-802.3 frame.
4.5 Fast Ethernet
The frame composition associated with each of the three Fast Ethernet standards is illustrated in Figure 4.13. In comparing the composition of the Fast
Ethernet frame with Ethernet and IEEE 802.3 frame formats previously illustrated in Figure 4.1, you will note that other than the addition of starting
and ending stream delimiters, the Fast Ethernet frame duplicates the older
frames. A third difference between the two is not shown, as it is not actually
observable from a comparison of frames, because this difference is associated
with the time between frames. Ethernet and IEEE 802.3 frames are Manchester encoded and have an interframe gap of 9.6 µsec between frames. In
SSD Preamble SFD
1 byte 7 bytes 1 byte
Destination
address
6 bytes
Source
L/T
address
6 bytes 2 bytes
Data
46 to 1500
bytes
FCS
1 byte
ESD
1 byte
Legend:
SSD = Start of stream delimiter
SFD = Start of frame delimiter
L/ T = Length (IEEE 802.3) /type (ethernet)
ESD = End of stream delimiter
Figure 4.13 Fast Ethernet frame. The 100BASE-TX frame differs from the
IEEE 802.3 MAC frame through the addition of a byte at each end to mark the
beginning and end of the stream delimiter.
186
chapter four
comparison, the Fast Ethernet 100BASE-TX frame is transmitted using 4B5B
encoding, and IDLE codes (refer to Table 3.6) representing sequences of I
(binary 11111) symbols are used to mark a 0.96-µs interpacket gap. Now that
we have an overview of the differences between Ethernet/IEEE 802.3 and
Fast Ethernet frames, let’s focus upon the new fields associated with the Fast
Ethernet frame format.
Start-of-Stream Delimiter
The start-of-stream delimiter (SSD) is used to align a received frame for
subsequent decoding. The SSD field consists of a sequence of J and K symbols,
which defines the unique code 11000 10001. This field replaces the first
octet of the preamble in Ethernet and IEEE 802.3 frames whose composition
is 10101010.
End-of-Stream Delimiter
The end-of-stream delimiter (ESD) is used as an indicator that data transmission terminated normally, and a properly formed stream was transmitted. This
one-byte field is created by the use of T and R codes (see Table 3.6) whose bit
composition is 01101 00111. The ESD field lies outside of the Ethernet/IEEE
802.3 frame and for comparison purposes can be considered to fall within the
interframe gap of those frames.
4.6 Gigabit Ethernet
Earlier in this chapter it was briefly mentioned that the Ethernet frame was
extended for operations at 1 Gbps. In actuality the Gigabit Ethernet standard resulted in two modifications to conventional CSMA/CD operations.
The first modification, which is referred to as carrier extension, is only
applicable for half-duplex links and was required to maintain an approximate 200-meter topology at Gigabit speeds. Instead of actually extending
the fame, as we will shortly note, the time the frame is on the wire
is extended. A second modification, referred to as packet burst, enables
Gigabit-compatible network devices to transmit bursts of relatively short
packets without having to relinquish control of the network. Both carrier
extension and packet bursting represent modifications to the CSMA/CD
protocol to extend the collision domain and enhance the efficiency of coppermedia Gigabit Ethernet, respectively. Both topics are covered in detail in
this section.
frame operations
187
Carrier Extension
In an Ethernet network, the attachment of workstations to a hub creates a
segment. That segment or multiple segments interconnected via the use of one
or more repeaters forms a collision domain. The latter term is formally defined
as a single CSMA/CD network in which a collision will occur if two devices
attached to the network transmit at or approximately the same time. The
reason we can say approximately the same time is due to the fact that there
is a propagation delay time associated with the transmission of signals on a
conductor. Thus, if one station is relatively close to another the propagation
delay time is relatively short, requiring both stations to transmit data at nearly
the same time for a collision to occur. If two stations are at opposite ends of
the network the propagation delay for a signal placed on the network by one
station to reach the other station is much greater. This means that one station
could initiate transmission and actually transmit a portion of a frame while
the second station might listen to the network, hear no activity, and begin to
transmit, resulting in a collision.
Figure 4.14 illustrates the relationship between a single collision domain
and two collision windows. Note that as stations are closer to one another the
collision window, which represents the propagation delay time during which
one station could transmit and another would assume there is no network
activity decreases.
Ethernet requires that a station should be able to hear any resulting collision
for the frame it is transmitting before it completes the transmission of the
entire frame. This means that the transmission of the next-to-last bit of a frame
Single collision domain
Collision
window
stations 1−4
1
Collision window 1−8
Legend:
N
Figure 4.14
4
8
= Station n
Relationship between a collision domain and collision windows.
chapter four
188
that results in a collision should allow the transmitting station to hear the
collision voltage increase before it transmits the last bit. Thus, the maximum
allowable cabling distance is limited by the bit duration associated with the
network operating rate and the speed of electrons on the wire.
When Ethernet operates at 1 Gbps, the allowable cabling distance would be
reduced to approximately 10 meters or 33 feet. Clearly, this would be a major
restriction on the ability of Gigabit Ethernet to be effectively used in a shared
media half-duplex environment. To overcome this transmission distance
limitation, Sun Microsystems, Inc., suggested the carrier extension scheme,
which became part of the Gigabit Ethernet standard for half-duplex operations.
Under the carrier extension scheme, the original Ethernet frame is extended
by increasing the time the frame is on the wire. The timing extension occurs
after the end of the standard CSMA/CD frame as illustrated in Figure 4.15.
The carrier extension extends the frame timing to guarantee at least a 512-byte
slot time for half-duplex Ethernet. Note that Ethernet’s slot time is considered
as the time from the first bit of the destination address field reaching the wire
through the last bit of the frame check sequence field. The increase in the
minimum length frame does not change the frame size and only alters the time
the frame is on the wire. Due to this compatibility it is maintained between
the original Ethernet frame and the Gigabit Ethernet frame.
Although the carrier extension scheme enables the cable length of a halfduplex Gigabit network to be extended to a 200-meter diameter, that extension
is not without a price. That price is one of overhead, because extension
symbols attached to a short frame waste bandwidth. For example, a frame
with a 64-byte data field would have 448 bytes of wasted carrier extension
Original
ethernet
frame
Original ethernet slot-time
Idle
Start
of
frame
Preamble
Start
delimiter
Destination
address
Source
address
Length /
type
Data
Frame
check
sequence
End of
frame
Idle
Idle
Start
of
frame
Preamble
Start
delimiter
Destination
address
Source
address
Length /
type
Data
Frame
check
sequence
End of
frame
Extension
Gigabit
ethernet
frame
Idle
512-byte slot-time
Figure 4.15 Half-duplex Gigabit Ethernet uses a carrier extension scheme to
extend timing so that the slot time consists of at least 512 bytes.
frame operations
189
symbols attached to it. To further complicate bandwidth utilization, when
the data field is less than 46 bytes in length, nulls are added to produce a
64-byte minimum-length data field. Thus, a simple query to be transported
by Ethernet, such as ‘‘Enter your age’’ consisting of 44 data characters, would
be padded with 32 null characters when transported by Ethernet to ensure a
minimum 72-byte length frame. Under Gigabit Ethernet, the minimum 512byte time slot would require the use of 448 carrier extension symbols to ensure
that the time slot from destination address through any required extension is
at least 512 bytes in length.
In examining Figure 4.15, it is important to note that the carrier extension
scheme does not extend the Ethernet frame beyond a 512-byte time slot.
Thus, Ethernet frames with a time slot equal to or exceeding 512 bytes have
no carrier extension. Another important item to note concerning the carrier
extension scheme is that it has no relationship to a Jumbo Frames feature
that is proprietary to a specific vendor. That feature is supported by a switch
manufactured by Alteon Networks and is used to enhance data transfers
between servers, permitting a maximum frame size of up to 9 Kbytes to
be supported. Because Jumbo Frames are not part of the Gigabit Ethernet
standard, you must disable that feature to obtain interoperability between that
vendor’s 1-Gbps switch and other vendors’ Gigabit Ethernet products.
Frame Bursting
Frame bursting represents a scheme added to Gigabit Ethernet to counteract the
overhead associated with transmitting relatively short frames. This scheme
was proposed by NBase Communications and is included in the Gigabit
Ethernet standard as an addition to carrier extension.
Under frame bursting, each time the first frame in a sequence of short frames
successfully passes the 512-byte collision window using the carrier extension
scheme previously described, subsequent frames are transmitted without
including the carrier extension. The effect of frame bursting is to average the
wasted time represented by the use of carrier extension symbols over a series
of short frames. The limit on the number of frames that can be bursted is a total
of 1500 bytes for the series of frames, which represents the longest data field
supported by Ethernet. To inhibit other stations from initiating transmission
during a burst carrier extension, signals are inserted between frames in the
burst. Figure 4.16 illustrates an example of Gigabit Ethernet frame bursting.
Note that the interframe gaps are filled with extension bits.
In addition to enhancing network use and minimizing bandwidth overhead,
frame bursting also reduces the probability of collisions occurring. This is
190
chapter four
Extension
bits
Frame 1
Carrier
extensions
Frame 2
Frame N
512 bytes
Frame burst
Figure 4.16
Frame bursting.
because the burst of frames are only susceptible to a collision during the first
frame in the sequence. Thereafter, carrier extension symbols between frames
followed by additional short frames are recognized by all other stations on the
segment, and inhibit those stations from initiating a transmission that would
result in the occurrence of a collision.
4.7 10 Gigabit Ethernet
As noted earlier in this book, 10 Gigabit Ethernet is restricted to operating
over optical fiber. In being restricted to operating over optical fiber, 10 Gigabit
Ethernet represents a full-duplex technology. This means that it does not need
the CSMA/CD protocol that is employed by slower, half-duplex versions of
Ethernet. This also means that in an effort to retain scalability to 10-Gbps
operations the frame formats used by other versions of Ethernet are continued
to be supported. Thus, you can encounter NetWare, true 802.3 or 802.3Q
tagged frames in a 10 Gigabit Ethernet environment.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter five
Networking Hardware and
Software
Until this chapter, our primary hardware focus was on generic products
designed to construct Ethernet-type networks at a single location. Although
we discussed the use of electrical fiber-optic repeaters, that discussion was
limited to a few network expansion examples. In addition, until now we have
essentially avoided discussing the use of bridges, routers, and gateways to
extend Ethernet connectivity beyond a single network, and the role of network
software and how it relates to your personal computer’s operating system. In
this chapter, we will focus our attention on those two areas.
First, we will focus our attention on the basic operation of several hardware
components that are the building blocks essential to extending the connectivity capability of an Ethernet local area network: repeaters, bridges, routers,
brouters, gateways, servers, and wiring hubs. Because Ethernet networks are
no longer restricted to a wired environment, in this chapter we will also
describe and discuss the role of wireless Ethernet LAN adapters, access points
and routers that include a built-in access point. Next, we will discuss the role
and operation of three major types of software required for local area network
operations: computer operating systems, LAN operating systems, and application programs. This will be followed by a discussion of the software used
to route data from one LAN to another using different internet hardware and
software products. By examining hardware and software, we will obtain an
appreciation for the methods used to link LANs both directly and through
wide area networks.
In this chapter and throughout this book, we will use the terms local
area network and network synonymously. We will use the term internetwork
or just internet to refer to the joining of two or more local area networks.
Note that we will use the latter term to refer to the combining of networks
and not to the specific network called the Internet, whose first letter is
191
192
chapter five
capitalized. In Chapter 7 we will examine the connection of Ethernet-based
networks to that network, focusing our attention upon the configuration
of workstations, servers, and routers to enable an Ethernet network to be
connected to the Internet.
5.1 Wired Network Hardware Components
In this section we will examine hardware products essential to the construction and interconnection of local area networks that transmit information
over wired media, such as coaxial cable, twisted pair of optical fiber. These
products provide us with the ability to extend the distance of local area
network coverage, connect local networks to distant ones, and obtain access
to centralized computational facilities.
Repeaters
A repeater is the simplest type of hardware component in terms of design,
operation, and functionality. This device operates at the physical layer of the
ISO Open Systems Interconnection Reference Model, regenerating a signal
received on one cable segment and then retransmitting the signal onto another
cable segment. Figure 5.1 illustrates the operation of a repeater with respect
to the ISO OSI Reference Model.
Types
There are two basic types of repeaters. An electrical repeater simply receives
an electrical signal and then regenerates the signal. During the signal regeneration process, a new signal is formed that matches the original characteristics
of the received signal. This process is illustrated in the lower portion of
Figure 5.1. By transmitting a new signal, the repeater removes any previous
distortion and attenuation, enabling an extension in the permissible transmission distance. Although several network segments can be connected by the
use of repeaters to extend the coverage of a network, there are constraints that
govern the maximum permissible length of a LAN. For example, a 50-ohm
coaxial bus-based Ethernet supports a maximum cabling distance of 2.3 km,
and that distance cannot be extended through the use of repeaters.
The second type of repeater commonly used is an electrical-optical device,
such as the FOIRL repeater, the use and general operation of which were discussed in Chapter 3. This type of repeater converts an electrical signal into an
optical signal for transmission and performs a reverse function when receiving
networking hardware and software
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Network
Network
Data link
Data link
Physical
Physical
Physical
193
Physical
(a) OSI operation
Repeater
(b) Signal regeneration process
Figure 5.1 Repeater operation. A repeater connects two local area networks
or network segments at the OSI physical layer (cable) by regenerating the
signal received on one LAN or LAN segment onto the other network or
network segment.
a light signal. Similar to an electrical repeater, the electrical-optical repeater
extends the distance that a signal can be carried on a local area network.
Because a repeater is restricted to operating at the OSI physical layer, it is
transparent to data flow. This restricts the use of a repeater to linking identical
networks or network segments. For example, you could use repeaters to
connect two Ethernet or two Token-Ring network segments, but not to connect
an Ethernet network to a Token-Ring network.
Utilization
Figure 5.2 illustrates the use of a repeater to connect two Ethernet bus-based
LANs, one serving the accounting department, and the other network serving
the data processing department. In this situation, all messages on one local
area network are passed to the other, regardless of their intended recipient.
The use of repeaters in this manner increases the sum of the traffic on each
LAN. If this system is implemented without knowledge of the traffic flow and
utilization levels on each network, performance problems may result when
separate networks are interconnected through the use of repeaters.
194
chapter five
Connecting ethernet bus-based local area networks
Accounting LAN
R
Data processing LAN
Legend:
= Workstations
R
= Repeaters
Figure 5.2 Using a repeater to link separate network segments. When a
repeater is used to connect separate network segments, traffic on each segment
is regenerated onto the other segment.
One special type of electrical repeater is a twisted-pair-based hub, such
as a 10BASE-T hub. As previously discussed in this book, a hub receives
data on one port and regenerates such data bit by bit onto all other ports.
Another special type of repeater is a buffered distributor. A buffered distributor
represents a relatively new type of IEEE 802.3 hub, which is designed to
connect two or more 802.3 links, each operating at 1 Gbps. Similar to other
802.3 repeaters, the buffer distributor is a nonaddressable device, forwarding
all inbound frames received on one port onto all other ports. Unlike other
repeaters, the buffered distributor, as its name implies, contains an internal
memory area that allows one or more incoming frames to be buffered before
forwarding those frames. Through the buffering of frames and the ability
to use the IEEE 802.3x flow control scheme to manage its internal buffer
level of occupancy, the buffered distributor is not subject to the Ethernet
collision-domain constraint nor other topology limitations.
Constraints
There are several constraints associated with the use of repeaters. Those
constraints include the previously discussed 5-4-3 rule, disabling of the
networking hardware and software
195
SQE test signal to repeaters to include hub ports, and topology restrictions
associated with the use of different types of repeaters. Concerning topology,
when a metallic repeater receives a signal, it uses the received signal to
recover clocking. If you recall our prior discussion of the composition of
Ethernet frames, we noted that a 64-bit preamble prefixes frame data for
synchronization. As a repeater samples the preamble, it may lose a few bits
as it recovers clocking. This bit loss shrinks the gap between frames that
should normally be 9.6 µs at 10 Mbps. In fact, according to the IEEE 802.3
specifications, the interframe gap (IFG) can be as low as 6.8 µs. However,
due to the gap shrinking the Ethernet specification does not permit more
than two repeaters to be located between any two communicating devices
located in the same collision domain. Note that this specification treats both
a shared media hub and a stand-alone repeater as repeaters. Thus, while you
could use one stand-alone repeater to extend the transmission distance from
a station to a hub, you could not perform a similar extension for a second
workstation connected to the hub. This is because the hub functions as a
repeater, which would result in the flow of data through three repeaters,
violating the two-repeater limit within a collision domain.
You can use the information presented in this book as well as vendor
specification sheets to determine the specific constraints associated with
different types of repeaters. Concerning vendor specification sheets, their use
is highly recommended, as the constraints associated with the use of different
types of repeaters can vary from one manufacturer to another.
Bridges
Unlike repeaters, which lack intelligence and are restricted to linking similar
LANs and LAN segments, bridges are intelligent devices that can connect
similar and dissimilar local area networks. To obtain an appreciation for
the functions performed by bridges, let us examine the use of this type of
networking product.
Operation
Figure 5.3 illustrates the operation of a bridge with respect to the OSI Reference
Model, as well as its use to connect two separate Ethernet local area networks.
Although the use of the bridge in the lower portion of Figure 5.3 looks similar
to the use of a repeater previously shown in Figure 5.2, the operation of each
device includes a number of key differences.
When a bridge begins to operate, it examines each frame transmitted on
connected local area networks at the data link layer — a process beyond
196
chapter five
OSI operation
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Network
Network
Data link
Data link
Data link
Data link
Physical
Physical
Physical
Physical
Application example
Network A
Bridge
Network B
Legend:
= Workstations
Figure 5.3 Bridge operation. A bridge connects two local area networks or
network segments at the data link layer.
the capability of a repeater, which operates transparent to data. By reading
the source address included in each frame, the bridge assembles a table of
local addresses for each network. In addition to reading each source address,
the bridge also reads the destination address contained in the frame. If the
destination address is not contained in the local address table that the bridge
constructs, this fact indicates that the frame’s destination is not on the current
network or network segment. In this situation, the bridge transmits the frame
onto the other network or network segment. If the destination address is
networking hardware and software
197
contained in the local address table, this indicates that the frame should
remain on the local network. In this situation, the bridge simply repeats the
frame without altering its routing.
We can summarize the operation of the bridge illustrated in the lower
portion of Figure 5.3 as follows:
♦ Bridge reads all frames transmitted on network A.
♦ Frames with destination address on network A are discarded by
the bridge.
♦ Frames with destination address on network B are retransmitted onto
network B.
♦ The above process is reversed for traffic on network B.
To illustrate the method by which bridges operate, let’s assume a three-port
bridge is used to connect three Ethernet segments together, as illustrated in
the top portion of Figure 5.4. In this example, for simplicity, each port address
is denoted by a letter of the alphabet while each station address is denoted
by a numeric. It should be noted that each bridge port has a MAC address as
it is a participant on a LAN. Although the MAC address of bridge ports and
stations have the same format and are six hex digits in length, we will use one
alphabetic character for each bridge port and one numeric character for each
station for simplicity of illustration.
The table in the lower portion of Figure 5.4 provides examples of the
three operations a bridge performs — flooding frames, forwarding frames, and
discarding frames. In the first entry, station 1 is presumed to transmit data
to station 2. Since we will assume the bridge was just powered on and has
no entries in its address-port table, it does not know which port station 2 is
connected to. Thus, it transmits the frame onto all other ports other than the
port it was received on, a technique referred to as flooding. Because station 2
is on the same segment as station 1, the transmission of frames onto the
other segments only interferes with traffic on those segments but will not
be recognized by a station, because no station connected to port B or C has
address 2.
In addition to flooding packets when the location of the destination address
is not known, there are two other situations where bridge flooding can occur.
Those situations include the receipt of a multicast or broadcast frame at one
port on a bridge, which results in such frames being transmitted onto all ports
other than the port they are received on. Returning to the example shown in
Figure 5.4, when the frame from station 1 is flooded by the bridge, it enters
the source address and port that the frame was received on into its addressport table. Hence the entry ‘‘1, A.’’ When the frame reaches station 2, we will
198
chapter five
1
2
A
B
Bridge
C
5
6
3
4
7
Bridge Operation
Address-Port Table
1 Transmits to 2
Frame flooded
1,A
2 Transmits to 1
Frame discarded
1,A
2,A
5 Transmits to 1
Frame forwarded
1,A
2,A
5,C
6 Transmits to 3
Frame flooded
1,A
2,A
5,C
6,C
3 Transmits to 6
Frame forwarded
1,A
2,A
3,B
5,C
6,C
Data Transfer Requirement
Figure 5.4 Bridges are self-learning networking devices that automatically
construct address-port table entries used to route data between network
segments.
assume 2 transmits a response to 1. As the frame propagates down the segment
and is read by bridge port A, it notes that address 1 is on port A. Thus, there is
no need for the bridge to forward the frame and it simply discards it. However,
the bridge notes address 2 is on port A and updates its address-port table.
For the third example, it was assumed that station 5 transmits to station 1.
Because the address-port table holds the relationship of address 1 with port A,
the bridge forwards the frame onto port A and updates its address-port table.
Next, when station 6 transmits to station 3 and the bridge checks its addressport table, it finds no entry for address 3, resulting in the bridge flooding
the frame onto ports B and C. Then the bridge notes that station 6 is not in
networking hardware and software
199
its address-port table and updates that table. In the last example, station 3
transmits a response to station 6. Because an entry for station 6 was just made
in the address-port table, the bridge forwards the frame onto port C.
In addition to associating addresses with ports, a bridge also time stamps
each entry and updates the time stamp when an address is referenced. As
storage is used up and the bridge needs to purge old entries to make room
for new entries, it will purge the oldest entries based upon the use of the
time stamp.
The previously described method of bridging operation is referred to as
transparent bridging. Refer to Chapter 6 for detailed information concerning
different methods of bridge operations.
Filtering and Forwarding The process of examining each frame is known
as filtering. The filtering rate of a bridge is directly related to its level of
performance. That is, the higher the filtering rate of a bridge, the lower the
probability it will become a bottleneck to network performance. A second
performance measurement associated with bridges is their forwarding rate.
The forwarding rate is expressed in frames per second and denotes the
maximum capability of a bridge to transmit traffic from one network to another.
In Chapter 6, we will expand our examination of bridges to determine their
maximum filtering and forwarding rates when used on Ethernet networks.
Types
There are two general types of bridges: transparent and translating. Each type
of bridge can be obtained as a local or remote device; a remote device includes
a wide area network interface and the ability to convert frames into a WAN
transmission protocol.
Transparent Bridge A transparent bridge provides a connection between
two local area networks that employ the same data link protocol. Thus, the
bridge shown in the lower portion of Figure 5.3 and the bridge shown in
the upper portion of Figure 5.4 are both examples of transparent bridges. At
the physical layer, some transparent bridges have multiple ports that support
different media. Thus, a transparent bridge does not have to be transparent at
the physical level, although the majority of such bridges are.
Although a transparent bridge provides a high level of performance for a
small number of network interconnections, its level of performance decreases
as the number of interconnected networks increases. The rationale for this
loss in performance is based on the method used by transparent bridges to
200
chapter five
develop a route between LANs. Refer to Chapter 6 for specific information
concerning bridge routing and performance issues.
Translating Bridge A translating bridge provides a connection capability
between two local area networks that employ different protocols at the data
link layer. Because networks using different data link layer protocols normally
use different media, a translating bridge also provides support for different
physical layer connections.
Figure 5.5 illustrates the use of a translating bridge to interconnect a TokenRing and an Ethernet local area network. In this example, the bridge functions
as an Ethernet node on the Ethernet and as a Token-Ring node on the
Token-Ring. When a frame from one network has a destination on the other
network, the bridge will perform a series of operations, including frame
and transmission rate conversion. For example, consider an Ethernet frame
destined to the Token-Ring network. The bridge will strip the frame’s preamble
and FCS, then it will convert the frame into a Token-Ring frame format. Once
the bridge receives a free token, the new frame will be transmitted onto the
Token-Ring; however, the transmission rate will be at the Token-Ring network
rate and not at the Ethernet rate. For frames going from the Token-Ring to the
Ethernet, the process is reversed.
Ethernet LAN
Bridge
Token-ring
Legend:
= Workstations
Figure 5.5 Translating bridge operation. A translating bridge connects
local area networks that employ different protocols at the data link layer. In
this example, the translating bridge is
used to connect an Ethernet local area
network to a Token-Ring network.
networking hardware and software
201
One of the problems associated with the use of a translating bridge is the
conversion of frames from their format on one network to the format required
for use on another network. As indicated in Chapter 2, the information field
of an Ethernet frame can vary from 64 to 1500 bytes. In comparison, a TokenRing can have a maximum information field size of 4500 bytes when the ring
operates at 4 Mbps, or 18,000 bytes when the ring operates at 16 Mbps. If a
station on a Token-Ring network has a frame with an information field that
exceeds 1500 bytes in length, the bridging of that frame onto an Ethernet
network cannot occur. This is because there is no provision within either
layer 2 protocol to inform a station that a frame flowing from one network to
another was fragmented and requires reassembly. Using a bridge effectively
in this situation requires that software on each workstation on each network
be configured to use the smallest maximum frame size of any interconnected
network. In this example, Token-Ring workstations would not be allowed to
transmit information fields greater than 1500 bytes.
Features
The functionality of a bridge is based on the features incorporated into this
device. Following is a list of the 11 major bridge features that define both the
functionality and the performance level of a bridge.
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
Filtering and forwarding rate
Selective forwarding capability
Multiple port support
Wide area network interface support
Local area network interface support
Transparent operation at the data link layer
Translating operation to link dissimilar networks
Encapsulation operation to support wide area network use
Stand-alone and adapter-based fabrication
Self-learning (transparent) routing
Source routing
Filtering and Forwarding The filtering and forwarding rates indicate the
ability of the bridge to accept, examine, and regenerate frames on the same
network (filtering) and transfer frames onto a different network (forwarding).
Higher filtering and forwarding rates indicate better bridge performance.
Selective Forwarding Some bridges have a selective forwarding capability.
Bridges with this feature can be configured to forward frames selectively, based
202
chapter five
on predefined source and destination addresses. With a selective forwarding
capability, you can develop predefined routes for frames to take when flowing
between networks, and enable or inhibit the transfer of information between
predefined workstations.
Figure 5.6 illustrates the use of the selective forwarding capability of two
bridges to provide two independent routes for data transfer between an
Ethernet and a Token-Ring network. In this example, you might enable all
workstations with source address 1 or 2 to have data destined to the TokenRing flow over bridge 1, while workstations with a source address of 3 or 4
are configured to use bridge 2. Although the bridges illustrated in Figure 5.5
appear to form a closed loop, in effect the previously described method of
selective forwarding results in a single logical path for each station on the
Ethernet. This ensures there are no closed loops — a condition that would
violate the spanning tree bridging algorithm supported by Ethernet bridges.
The operation of that algorithm is described in Chapter 6.
Multiple Port Support The multiple port support capability of a bridge is
related to its local and wide area network media interface support. Some
1
3
Ethernet
4
2
B1
B1
Token-ring
Legend:
= Workstations
Figure 5.6 Using bridge selective forwarding capability. Using the selective
forwarding capability of bridges enables the data flow between networks to be
distributed based upon source or destination addresses.
networking hardware and software
203
bridges support additional ports beyond the two that make up a basic bridge.
Doing so enables a bridge to provide connectivity between three or more local
area networks.
Figure 5.7 illustrates one potential use of a multiple port bridge to link an
Ethernet network to two Token-Ring networks.
Local and Wide Area Interface Support Local area media interfaces supported by bridges can include thin and thick Ethernet coaxial cable, shielded
and unshielded twisted-pair metallic, cable, and different types of fiber optic
cable. Wide area network interfaces are incorporated into remote bridges that
are designed to provide an internetworking capability between two or more
geographically dispersed LANs linked by a WAN. Common WAN media interfaces can include RS-232 for data rates at or below 19.2 Kbps, ITU X.21 for
packet network access at data rates up to 128 Kbps, ITU V.35 for data rates up
to approximately 6 Mbps, and a T1/E1 interface for operations at 1.544 Mbps
and 2.048 Mbps, respectively.
Transparent Operation Although bridges are thought of as transparent to
data, this is not always true. For interconnecting different networks located in
Multiport
bridge
Token-ring
Legend:
Token-ring
= Workstations
Figure 5.7 Using a multiport bridge. With a multiport bridge, you can interconnect three or more local area networks.
204
chapter five
the same geographical area, bridges are normally transparent to data. However,
some remote bridges use data compression algorithms to reduce the quantity
of data transmitted between bridges connected via a wide area network. Such
compression performing bridges are not transparent to data, although they
restore data to its original form.
Frame Translation For interconnecting different types of local area networks, bridges must perform a translation of frames. For example, an Ethernet
frame must be changed into a Token-Ring frame when the frame is routed
from an Ethernet to a Token-Ring network. As previously mentioned, because
frames cannot be fragmented at the data link layer, you must set the workstations on the Token-Ring network to the smallest maximum information field
of an Ethernet frame, 1500 bytes.
When data is transferred between colocated local area networks, the frame
format on one network is suitable for transfer onto the other network; it is
modified for transfer when the media access control layers differ. When a
bridge is used to connect two local area networks via a wide area network
facility, a WAN protocol is employed to control data transfer. The wide
area network protocol is better suited for transmission over the WAN, as
it is standardized to incorporate error detection and correction, enable a
large number of unacknowledged WAN frames to exist to speed information
transfer, and support full-duplex data transfers. Examples of such wide area
network protocols include IBM’s SDLC, and the ITU’s HDLC and X.25.
Frame Encapsulation Figure 5.8 illustrates the operation of a pair of remote
bridges connecting two local area networks via a wide area network. For
transmission from network A to network B, user data from a network A station
is first converted into logical link control and media access control frames. The
bridge then encapsulates one or more LAN frames into the bridging protocol
frame used for communications over the wide area network. Because the local
area network frame is wrapped in another protocol, we say the LAN frame
is tunneled within the WAN protocol. At the opposite end of the wide area
network, the distant remote bridge performs a reverse operation, removing the
WAN protocol header and trailer from each frame.
Fabrication Some bridges are manufactured as stand-alone products. Such
devices are called plug and play, because you simply connect the bridge ports
to the media and turn them on. Other bridges are manufactured as adapter
cards for insertion into the system unit of a personal computer, workstation,
or reduced instruction set computer (RISC). Through the use of software
networking hardware and software
Station
User
1
LLC
2
MAC
3
Physical
Bridge
4
LAN A
Bridge
MAC
Link
Physical Physical
WAN
Facility
MAC
Link
6
Physical Physical
5
Location
User data
2, 8
LLC-H User data
5
Station
9 User
LLC
8
7 MAC
Physical
Frame composition
1, 9
3, 4, 6, 7
LAN B
205
MAC-H LLC-H User data LLC-T
Link-H MAC-H LLC-H User data LLC-T Link-T
Figure 5.8 Remote bridge operation. A remote bridge wraps the logical link
control (LLC) and media access control (MAC) frames in another protocol for
transmission over a wide area network.
developed in conjunction with hardware, you may obtain more flexibility
with this type of bridge than with a stand-alone device whose software is
fixed in ROM.
Routing Method The routing capability of a bridge governs its capability to
interconnect local area networks and its level of performance. As previously
examined through the use of Figure 5.4, a transparent bridge automatically
develops routing tables. Thus, this device is known as a self-learning bridge
and represents the most elementary type of bridge routing. In the IBM TokenRing frame, there is an optional routing field that can be used to develop
routing information for frames routed through a series of bridges — a technique
referred to as source routing. Refer to Chapter 6 for an in-depth discussion of
bridge routing methods, as well as translating operations that enable a bridge
to connect IEEE 802.3 and 802.5 networks.
Routers
A router is a device that operates at the network layer of the ISO OSI
Reference Model, as illustrated in Figure 5.9. What this means is that a router
examines network addresses and makes decisions about whether data on a
local area network should remain on the network or should be transmitted
206
chapter five
Application
Application
Presentation
Presentation
Session
Session
Transport
Transport
Network
Network
Network
Network
Data link
Data link
Data link
Data link
Physical
Physical
Physical
Physical
Figure 5.9 Router operation. A router operates at the network layer of the
ISO OSI Reference Model. This enables a router to provide network layer
services, such as flow control and frame fragmentation.
to a different network. Although this level of operation may appear to be
insignificant compared with that of a bridge operating at the data link layer,
there is actually a considerable difference in the routing capability of bridges
and routers.
Operation
A bridge uses 48-bit MAC addresses it associates with ports as a mechanism
to determine whether a frame received on one port is ignored, forwarded onto
another port, or flooded onto all ports other than the port it was received on.
At the data link layer there is no mechanism to distinguish one network from
another. Thus, the delivery of frames between networks requires bridges to
learn MAC addresses and base their forwarding decisions on such addresses.
Although this is an acceptable technique for linking a few networks together,
as the number of networks increase and the number of workstations on interconnected networks increase, bridges would spend a considerable amount of
time purging old entries and flooding frames, causing unacceptable performance problems. This resulted in the development of routers that prevent the
flooding of frames between networks and operate by using network addresses.
To understand the concept behind routing, consider Figure 5.10, which
illustrates two networks connected by a pair of routers. In this example,
assume addresses A, B, C, and D represent MAC addresses of workstations,
and E and F represent the MAC address of each router connected to each
networking hardware and software
A
B
Network 10
C
207
D
Network 10
E
F
Router
1
Router
2
Figure 5.10 Routing between two networks. When a workstation on one
network transmits data to a device on another network, an application program
provides the destination network address, which is used by routers as a basis
for their routing decisions.
network. Assuming workstation A has data to transmit to workstation C, an
application program executing on workstation A assigns network 20 as the
destination network address in a packet it prepares at the network layer. That
packet is transported via one or more MAC frames to MAC address E, which
is the data link address of the network adapter card installed in router 1.
In actuality, each device on a network can have a unique network address,
with a portion of the address denoting the network and the remainder of the
address denoting each host or router interface. This enables the application
program on workstation A to transmit a packet destined to workstation C to
router 1, with the data link layer passing the packet to MAC address E as a
frame or sequence of frames. The router receives the frame or sequence of
frames explicitly addressed to it and notes that the packet has the destination
address for network 20. The router then searches its routing tables and
determines that it should relay the packet to router 2. When the packet is
received at router 2, it recognizes the fact that network 20 is connected to the
router via its connection to the LAN by an adapter card using MAC address F.
Thus, the router will transport the packet to its destination using one or more
frames at the data link layer with a MAC source address of F.
The preceding description represents a simplified description of the routing
process. A more detailed description will be presented later in this book after
we investigate the TCP/IP protocol suite. We will examine the TCP/IP protocol
suite in the second section in this chapter when we turn our attention to networking software and focus our attention upon router operations in Chapter 7.
The ability to operate at the network layer enables a router to extend
networking across multiple data links in an orderly and predefined manner.
This means that a router can determine the best path through a series of data
links from a source network to a destination network. To accomplish this,
208
chapter five
the router operates under a network protocol, such as the Internet Protocol
(IP), Digital Equipment Corporation’s DECnet Phase V, or Novell’s IPX. This
networking protocol must operate on both the source and destination network
when protocol-dependent routers are used. If you use protocol-independent
routers, you can interconnect networks using different protocols. The protocolindependent router can be considered a sophisticated transparent bridge. Its
operation and utilization are described in detail in Chapter 7.
In comparison, because a bridge operates at the data link layer, it can
always be used to transfer information between networks operating under
different network protocols. This makes a bridge more efficient for linking
networks that only have one or a few paths, while a router is more efficient
for interconnecting multiple network links via multiple paths.
Network Address Utilization Unlike a bridge, which must monitor all
frames at the media access control layer, a router is specifically addressed
at the network layer. This means that a router has to examine only frames
explicitly addressed to that device. In communications terminology, the
monitoring of all frames is referred to as a promiscuous mode of operation,
while the selective examination of frames is referred to as a nonpromiscuous
mode of operation.
Another difference between the operation of bridges and routers is the
structure of the addresses on which they operate. Bridges operate at the data
link layer, which means that they typically examine physical addresses that
are contained in ROM on adapter cards and used in the generation of frames.
In comparison, routers operate at the network layer, where addresses are
normally assigned by a network administrator to a group of stations having a
common characteristic, such as being connected on an Ethernet in one area
of a building. This type of address is known as a logical address, and can be
assigned and modified by the network administrator.
Table Operation Like bridges, routers make forwarding decisions using
tables. However, unlike a bridge, which may employ a simple table look-up
procedure to determine if a destination address is on a particular network,
a router may employ much more sophisticated forwarding decision criteria.
For example, a router may be configured to analyze several paths using
an algorithm and dynamically select a path based upon the results of the
algorithm. Routing algorithms and protocols are discussed in Chapter 7.
Advantages of Use The use of routers provides a number of significant
advantages over the use of bridges. To illustrate these advantages, we will
networking hardware and software
Los Angeles
209
New York
Network B
Network C
Network A
R2
R1
Network D
Washington, DC
Network E
R3
Network F
Miami
R4
Legend:
= Workstation
R = Router
Network G
Figure 5.11 Building an internet using routers. Routers can be used to establish complex networks in which traffic is varied over network facilities.
This variation is based on the operational status and utilization of different
network paths.
examine the use of routers in Figure 5.11, in which four corporate offices
containing seven local area networks are interconnected through the use of
four routers. In this example, networks A and B are located in a building in
Los Angeles, networks C and D are located in New York, network E is located
in Washington, DC, and networks F and G are located in Miami.
Multiple Path Transmission and Routing Control Suppose a station on
network A in Los Angeles requires transmission to a station on network G
in Miami. Initially, router R1 might use the path R1-R4 to transmit data
between networks. If the path should fail, or if an excessive amount of traffic
210
chapter five
flows between Los Angeles and Miami using that path, router R1 can seek to
establish other paths, such as R1-R3-R4 or even R1-R2-R3-R4. In fact, many
routers will consider each packet as a separate entity, routing the packet to its
destination over the best available path at the time of transmission. Although
this could conceivably result in packets arriving at R4 out of sequence, routers
have the capability to resequence packets into their original order before
passing data onto the destination network.
Flow Control As data flows through multiple paths toward its destination, a
link can become congested. For example, data from stations on network C and
network E routed to network G might build up to the point where the path
R3-R4 becomes congested. To eliminate the possibility of packet loss, routers
will use flow control. That is, they will inhibit transmission onto a link and
notify other routers to inhibit data flow until there is an available level of
bandwidth for traffic.
Frame Fragmentation As previously mentioned, bridges cannot break a
frame into a series of frames when transmission occurs between networks
with different frame sizes. This situation requires workstations to be configured to use the smallest maximum frame size of any of the networks
to be connected together. In comparison, most network protocols supported by routers include a provision for fragmentation and reassembly
of packets.
The higher level of functionality of routers over bridges is not without
a price. That price is in terms of packet processing, software complexity,
and cost. Since routers provide a more complex series of functions than
bridges, their ability to process packets is typically one-half to two-thirds
of the processing capability of bridges. In addition, the development time
required to program a more complex series of functions adds to the cost of
routers. Thus, routers are generally more expensive than bridges. Table 5.1
summarizes the major differences between bridges and routers in terms of
their operation, functionality, complexity, and cost.
Brouters
A brouter is a hybrid device, representing a combination of bridging and
routing capabilities. Although brouters are no longer marketed using this
term, their functionality lives on in modern routers that include a bridging
capability.
networking hardware and software
TABLE 5.1
211
Bridge/Router Comparison
Characteristic
Bridge
Router
Routing based upon an algorithm
or protocol
Normally no
Yes
Protocol transparency
Yes
Only protocol-independent
router
Uses network addresses
No
Yes
Promiscuous mode of operation
Yes
No
Forwarding decision
Elementary
Can be complex
Multiple path transmission
Limited
High
Routing control
Limited
High
Flow control
No
Yes
Frame fragmentation
No
Yes
Packet processing rate
High
Moderate
Cost
Less expensive
More expensive
Operation
When a brouter receives a frame, it examines that frame to determine whether
it is destined for another local area network. If it is, the brouter then checks
the protocol of the frame to determine whether it is supported at the network
layer supported by the router function. If the frame’s protocol is supported,
the brouter will route the frame as if it were a router. However, if the brouter
does not support the protocol, it will instead bridge the frame using layer 2
information.
In comparison to routers, brouters provide an additional level of connectivity between networks, although that connectivity takes place at a lower level
in the OSI Reference Model hierarchy. This is because a router would simply
ignore a frame for which it does not support the network protocol, while a
brouter would bridge the frame.
Utilization
The key advantage of using a brouter is the ability of this device to both
bridge and route data. Its ability to perform this dual function enables a
brouter to replace the use of separate bridges and routers in some networking
212
chapter five
applications. For example, consider the use of a separate bridge and router
in the top portion of Figure 5.12. In this example, the bridge provides an
interconnection capability between two neighboring networks, while the
router provides an interconnection capability to distant networks. By replacing
the separate bridge and router with a brouter, you can obtain the same level
of functionality, as illustrated in the lower portion of Figure 5.12. Of course,
you want to ensure that the filtering and forwarding rates of the brouter are
sufficient to be used in the manner illustrated. Otherwise, the replacement
of separate bridges and routers by brouters may introduce delays that affect
network performance. As previously noted, many modern routers include
a built-in bridging capability and, as such, can be considered to represent
a brouter. Refer to Chapters 6 and 7 for information concerning the processing
requirements of bridges and routers.
Using a separate bridge and router
B
To other networks
R
Using a brouter
BR
Figure 5.12
To other networks
Replacing a separate bridge and router with a brouter.
networking hardware and software
213
Gateway
The well-known phrase ‘‘one person’s passion is another’s poison’’ can, in
many ways, apply to gateways. The term gateway was originally coined to
refer to a device providing a communications path between two local area
networks, or between a LAN and a mainframe computer, from the physical
layer through the application layer. When applied to a device interconnecting
networks, the term gateway was originally used and is still commonly used to
reference a router. In fact, many programs used to configure workstation and
server network operations request the entry of the gateway network address,
even though it is more appropriate today to use the term router.
Figure 5.13 illustrates the operation of a gateway with respect to the ISO
OSI Reference Model. Unfortunately, the term gateway has been used loosely
to describe a range of products, ranging from bridges and routers that interconnect two or more local area networks to protocol converters that provide
asynchronous dial-up access into an IBM SNA network. Thus, a definition of
the term is warranted.
Definition
In this book, we will use the term gateway, unless otherwise noted, to
describe a product that performs protocol conversion through all seven layers
of the ISO OSI Reference Model. As such, a gateway performs all of the
functions of a router, as well as any protocol conversions required through
the application layer.
Application
Application
Application
Application
Presentation
Presentation
Presentation
Presentation
Session
Session
Session
Session
Transport
Transport
Transport
Transport
Network
Network
Network
Network
Data link
Data link
Data link
Data link
Physical
Physical
Physical
Physical
Figure 5.13 Gateway operation. A gateway operates at all seven layers of the
ISO OSI Reference Model.
214
chapter five
One of the most common types of gateways is an electronic mail (e-mail)
gateway, which converts documents from one e-mail format to another. For
example, an internal corporate network might be operating Lotus Development
Corporation’s CC:MAIL. If they require connectivity with the Internet, the
gateway must convert internal CC:MAIL documents so that they can be
transported via the Simple Mail Transport Protocol (SMTP), which is a TCP/IP
electronic mail application. Similarly, SMTP mail delivered via the Internet to
the organization’s gateway must be converted by the gateway into the format
used by CC:MAIL. A second type of popular gateway enables workstations
on a network to communicate with a mainframe as if the workstation was
a specific type of terminal device. This type of gateway also operates at all
seven layers of the ISO Reference Model.
Operation
Gateways are protocol-specific in function, typically used to provide access to
a mainframe computer. Some vendors manufacture multiprotocol gateways.
Such products are normally manufactured as adapter cards, containing separate processors that are installed in the system unit of a personal computer
or in a specially designed vendor hardware platform. When used in conjunction with appropriate vendor software, this type of gateway is actually an
N-in-1 gateway, where N is the number of protocol conversions and separate
connections the gateway can perform.
Figure 5.14 shows a multiprotocol gateway used to link LAN stations to
an IBM mainframe via an SDLC link and an X.25 connection to a packet
switching network. Once connected to the packet switching network, LAN
traffic may be further converted by gateway facilities built into that network,
or traffic may be routed to a packet network node and transmitted from that
node to its final destination in its X.25 packet format.
Gateways are primarily designed and used for LAN-WAN connections and
not for inter-LAN communications. Because they perform more sophisticated
functions than routers, they are slower in providing network throughput. In
addition, because the configuration of a gateway requires the consideration of
a large number of protocol options, a gateway installation is considerably more
difficult than the setup of a router. Refer to Chapter 8 for specific information
concerning the operation and utilization of gateways.
File Servers
The file server is the central repository of information upon which a local
area network is normally constructed. The file server is normally a personal
networking hardware and software
215
Gateway
A
A
SDLC
Modem
Packet
Network
IBM
mainframe
Modem
Legend:
= Workstation
A = Adapter card
Figure 5.14 Multiprotocol gateway operation. A multiprotocol gateway can
be used to provide local area network stations access to different computational facilities, either directly or through the use of a packet network.
computer or workstation that has a very powerful microprocessor, such as
an Intel Pentium, Digital Alpha, or a RISC chip, and a large amount of fastaccess on-line storage. The on-line storage is used to hold the local area
network operating system and the application programs that other stations
on the network may use. Several software vendors have split application
programs, enabling portions of a program to be run on a network station,
while other portions, such as the updating of a database, occur on the server.
This technique is known as client–server processing.
In comparison, the general term client/server operations references the
access of one computer by another on a network. The computer being accessed,
while normally considered to represent a file server, can also be a mainframe
or minicomputer. For either situation, the client uses a network to access
one or more features of the server. Those features can range in scope from a
spreadsheet program stored on a file server to a database query program on a
mainframe that provides access to billions of records.
Connectivity Functions
As the heart of a local area network, the server supports every network application. It is therefore in an ideal position to perform connectivity functions.
Some servers are designed to function as asynchronous gateways, providing
access to one or more modems that individual network stations can access
216
chapter five
for communications via the switched telephone network. Here, software on
the server considers the group of modems as a pool for access purposes.
Hence, the term modem pooling is used to refer to a server used to provide
this service.
Other servers contain software and hardware to provide access to mainframe
computers via a specific protocol, such as SDLC, X.25, or TCP/IP. Depending
on the capabilities of the network operating system used on the server, the
server may support server-to-server communications on an intra- and interLAN basis. When supporting intra-LAN server-to-server communications, you
can assign network users to a specific default server for an initial network
log-on, distributing the use of servers so that no one server becomes overused.
Then, the network operating system would support the transfer of information
between servers, such as an electronic mail message created by a network user
assigned to one server, who addresses the message to another employee
assigned to a different server. If your network operating system supports
inter-LAN server-to-server communications, the server in effect functions as a
remote bridge.
Types of Servers
The first type of server developed for use with local area networks primarily
supported file and printer sharing. Thus, although referred to as a file server,
the server also permitted a few printers to be connected to the computer as
network-accessible devices. As the concept of local area networking gained
acceptance, servers were developed to perform specific functions. Today,
popular types of servers include application servers, communications servers,
print servers, and remote access servers (RASs). Figure 5.15 illustrates an
Ethernet network that contains four distinct types of servers.
The RAS can be considered to represent a hybrid device that consists of
a modem pool and router. This type of server permits the support of serial
communications in the form of analog modems, digital ISDN calls, and the
connection of a T1 circuit that contains 24 individual connections grouped
together on a common circuit. Incoming communications are prompted for
access validation, typically in the form of a user ID and password, to gain
access from the RAS onto the network. Once connected to the network,
they then obtain access privileges commensurate with the network account
they have.
Although a communications server also provides connectivity between
a network and a serial communications facility, there are some distinct
differences between this server and an RAS. First, a communications server is
networking hardware and software
217
Switched network
Printer
Printer
Print
server
Legend:
Figure 5.15
Application
server
M M
M M
Remote
access
server
Communications
server
= Workstations
Servers can be acquired to perform specific functions.
primarily used for outbound traffic from network users, although if should be
Noted that it supports full-duplex communications. Thus, it may not support
inbound traffic, and if it does, it probably just places the traffic onto the network
without checking the validity of the user. Secondly, most communications
servers function as replacements for users having individual modems. Thus,
another popular name for a limited capability communications server is a
modem pooler.
Location Considerations
Figure 5.16 illustrates a file server used both in its traditional role as a
repository for application programs and for modem pooling and gateway
operations. The transmission of information between workstations and servers
far exceeds transmission between workstations. The location of file servers
on an Ethernet local area network is therefore critical for ensuring that
network performance remains at an acceptable level. In the example shown
in Figure 5.16, the server is located at a point on an Ethernet bus-based local
area network in which the transmission distance between each workstation
and the server is maximized. This location results in the greatest propagation
delay for the transmission of data between each workstation and the server.
218
chapter five
Server
Legend:
= Workstation
Figure 5.16 Server placement. The
location of a server on a network
is critical for ensuring network performance. In this example, because
each workstation requires access to
the server, its location requires the
greatest transmission distance between individual workstations and
the server. This, therefore, is probably a poor server location.
This is probably a poor location with respect to network performance if each
workstation requires a similar level of access to the server.
The reason for this representing a poor placement is the fact that the longer
the distance between two network devices on a CSMA/CD access protocol
network, the greater the probability that one station will listen to the network
and not hear a signal while another has just began transmitting. This, in turn,
results in an increased number of collisions that adversely affects network
performance. Because collisions result in the occurrence of a random back-off
time interval during which no data can be transferred on the network, the
proper location of a server can increase network transmission efficiency.
Wire Hubs
As mentioned in Chapter 3, 10BASE-T, 100BASE-T, and the half-duplex
version of Gigabit Ethernet use wiring concentrators, or hubs. Hubs are
typically located in telephone wiring closets, with stations cabled to the hub
in the form of a star. Hubs are then connected to one another to form a bus.
Only if the closest workstations to the server require a disproportionate level
of server access, in comparison to other workstations, is the server location
shown in Figure 5.16 correct. If each workstation has an equal level of server
access, the server should be relocated to the middle of the network. This
move would minimize propagation delays between the workstations and the
server — which, in turn, would minimize the probability of collisions.
Advantages
Ethernet hubs employ standard wiring between stations and the hub in the
form of twisted-pair cable. Because workstations are connected to a single
networking hardware and software
219
point, administration of a hub-based network is normally simple and economical, because a central point allows network configuration and reconfiguration,
monitoring, and management. Due to the value of hubs in local area networking, several vendors introduced products commonly referred to as intelligent
hubs. This product provides users with the ability to build local area networks, ranging in scope from a single LAN with a small number of nodes
to mixed-protocol LANs with several thousand nodes that can be monitored,
modified, and analyzed from a single point.
Intelligent Hubs
An intelligent hub represents an advance in functionality and capability over
conventional wire hubs. The intelligent hub includes its own microprocessor
and memory, which not only provide a network management capability, but,
in addition, may provide the ability to interconnect separate networks through
integrated bridges and routers.
Most intelligent hubs include a multibus backplane, which enables bridges
and routers to access and route data originating over multiple types of media.
When operating in this manner, the intelligent hub can be viewed as a PBX,
providing connectivity between any station on any network connected through
the hub, while providing a centralized management and control capability.
Through the use of an intelligent hub, the administrator can enable or disable
network ports from a network management workstation, segment a network
to balance traffic better and improve performance, and facilitate troubleshooting and maintenance operations. As networks became more sophisticated,
intelligent hubs evolved to facilitate their operation and management. With
the ability of some intelligent hubs to bring together various media flavors
of different local area networks, this device also provides a mechanism for
the integration of prior piece-by-piece planning, or for the correction of
a plan gone astray. Thus, the use of an intelligent hub may provide you
with an umbrella mechanism to bring together separate networks and media
previously installed in a building.
Switching Hubs
Improvements in the design of intelligent hubs to include faster backplanes,
port memory buffers, and one or more mechanisms to control the flow
of information resulted in the development of a series of Ethernet hardware
products. Collectively referred to as switching hubs or LAN switches, products
chapter five
220
in this broad category of equipment examine the destination address of each
frame as a decision criteria for invoking a switching operation.
A switching hub can be considered to represent a sophisticated bridge that
has the ability to support multiple bridging operations at the same time. This
is illustrated in Figure 5.17, which shows a 6-port switch with workstations
connected to ports 3 and 4 having frames transported to servers connected to
ports 5 and 6. If each port operates at 10 Mbps then the two cross-connections
provide 20 Mbps of bandwidth. In comparison, a conventional shared media
10BASE-T network only permits one frame at a time to be on the network,
limiting bandwidth to 10 Mbps.
Some switching hubs are limited to working with individual workstations
connected to each port, while other switching hubs are designed to switch
frames from all workstations on a segment connected to the hub. The first
type of switching is referred to as port switching, while the second type
of switching is referred to as segment-based switching. The development
of 100-Mbps high-speed Ethernet technology resulted in several vendors
introducing multiple-speed switching hubs during 1994 and 1995, which
could be used to significantly improve the performance of many types of
networks. This improvement primarily resulted from the connection of 100Mbps ports to file servers, while 10-Mbps ports are connected to 10BASE-T
workstations and conventional hubs. Similarly, the first series of Gigabit Ethernet switches, which reached the market during 1997, have 100-Mbps and
1-Gbps ports, with the 100-Mbps ports used to support Fast Ethernet connections. Continuing the switching evolution, during 2001 several switches
Server
Server
5
6
Switching
hub
1
Legend:
2
3
4
= Workstations
Figure 5.17 A switching hub supports multiple
simultaneous cross-connection operations, which
enhance available network bandwidth.
networking hardware and software
221
were introduced that included one or two 10 Gigabit Ethernet ports, while
the remainder of the switch was limited to supporting Gigabit transmission.
Another technology incorporated into switching hubs is a full-duplex operating capability, permitting the simultaneous transmission and reception of
data on a connection. Although many switching hub manufacturers denote
this capability for a 10BASE-T connection; for example, as a 20-Mbps Ethernet port, in reality you obtain a 20-Mbps transmission rate only when
a device is simultaneously transmitting and receiving information. Because
file servers benefit more than workstations from a full-duplex transmission
capability, most network designers connect their file servers to the fullduplex port on a switching hub that supports this capability, and use the
conventional ports for connections to workstations. Readers are referred to
Chapter 6 for an in-depth examination of the operation and utilization of
switching hubs.
5.2 Wireless Network Hardware Components
Because this chapter is focused upon network hardware and software, this
author would be remiss if he did not review wireless Ethernet LAN components. However, prior to doing so, a brief description of the two types of
networking infrastructure supported via wireless LANs is in order.
Network Topologies
The ability to communicate via the use of the air requires computers to have a
wireless LAN network adapter card. That network adapter card functions similarly to a wired LAN adapter card, taking application data and forming frames.
However, those frames are created to flow over the air and the physical layer
of the wireless LAN adapter card supports radio frequency communications
instead of communications over a metallic or fiber-optic media.
There are two types of network topologies supported by wireless LANs.
Those topologies are referred to as ad hoc or simultaneous and infrastructure. An ad hoc network is formed when two clients with wireless
LAN network adapters come into close proximity with one another. The
second type of wireless LAN topology is referred to as an infrastructure.
Under this networking topology, wireless LAN clients communicate with one
another, as well as with their wired relatives, through an access point.
Thus, in order to obtain an appreciation of the infrastructure topology,
the reader requires an understanding of the manner by which an access
point operates.
222
chapter five
Access Point
An access point can be considered to represent a two-port bridge. However,
instead of each port being connected to a wired LAN, one port is connected
to an antenna.
Figure 5.18 illustrates the use of an access point to provide a wireless infrastructure communications topology. Note that in this example two wireless
clients with wireless LAN adapter cards first access the access point in order
to obtain access to the wired LAN. Because the access point functions as
a bridge it constructs a port-address table in a manner similar to that of a
conventional bridge. However, the addresses for the second port represent
those destination addresses associated with wireless clients whose addresses
it learns as the clients access the access point while the latter devices note
their source address.
In addition to wireless LAN network adapter cards and access points, there
are two additional wireless devices that warrant attention. Those devices
include a wireless router and a wireless bridge. Thus, in concluding our
examination of the devices used to create a wireless LAN infrastructure let’s
focus our attention upon each of these.
Wireless Router
A wireless router represents a router with a built-in access point. This device
may also be constructed with two or more built-in Ethernet switch ports.
Hub
Hub
Access
point
Wireless
client
Wireless
client
Figure 5.18 Using an access point enables wireless LAN clients to gain access
to a wired LAN.
networking hardware and software
223
The most common use of the wireless router is in the home or smalloffice environment. In this environment a cable or DSL modem is connected
to an Ethernet port built into the router. Because the router includes a
network address translation capability, it can be used to enable many wireless
clients to simultaneously access the Internet even though the Internet Service
Provider may only allocate a single IP address to the cable modem or DSL
modem connection. The Ethernet switch ports provide connectivity to other
workstations or a wired LAN. Figure 5.19 illustrates the potential use of a
wireless router to create an infrastructure where both wired and wireless
clients can share access to the Internet.
Wireless Bridge
We previously noted that an access point represents a bridge that provides
connectivity between a wired and an over-the-air interface. Thus, you may
now be a bit curious as to what a wireless bridge represents, so let’s discuss
its operation.
A wireless bridge represents an access point and an antenna that are
physically separated from one another. The access point is connected to a
wired LAN while the antenna is commonly mounted on the top or side of a
building. The access point is then cabled to the antenna. Through the use of
Internet
Modem
Wireless
router
Hub
Wireless
client
Wireless
client
Direct
connect
workstation
Wired
workstations
Figure 5.19 Using a wireless router to provide wireless and wired clients
with shared access to each other and the Internet.
chapter five
224
very sensitive highly directional antennas it becomes possible to interconnect
a pair of wired LANs via a pair of wireless LAN bridges at distances up to
approximately ten miles. However, to achieve this range the antennas used
for each bridge must be within a line of sight of each other. Readers are
referred to Chapter 8 for detailed information concerning wireless Ethernet
LAN operations.
5.3 Networking Software
The installation of a local area network requires a variety of hardware and
software products. At minimum, each workstation on the network normally
requires the installation of an interface card into its system unit. This interface
card contains a number of ROM modules and specialized circuitry, as well
as a microprocessor that implements the access method used to communicate
on the common cable.
In some local area networks, one personal computer must be reserved
to process the network commands and functions. Because it services these
commands and functions, it is normally called the network server, although
the term file server is also commonly used to reference this computer. A
combination of specialized software that overlays the operating system on
each workstation and software placed on the server governs the operation of
the network; this relationship provides a client–server processing capability.
In comparison, the flow of data from one workstation to another without the
use of a server is known as peer-to-peer processing. Because the vast majority
of Ethernet LANs use servers, we will focus our attention in this section on the
software required to support client/server operations. To understand the role
of the network server and the functions of LAN software, let us first review
how a personal computer operates under a conventional version of the disk
operating system (DOS) used in the IBM PC personal computer environment.
DOS
DOS is a single-user operating system designed to provide access to peripherals
attached to the system and control of those peripherals, interpretation of
commands to perform various functions on the system, and management of
disk storage and memory. Under DOS, your keyboard is the standard input
device and your display is the standard output device, and control of the
personal computer is limited to one user. When you turn on your PC, one of the
functions performed by DOS is to load a command processor into a predefined
networking hardware and software
225
area of your computer’s memory. The command processor contains coding
that examines your keyboard entries, and coding that performs predefined
functions when certain commands are recognized from your keyboard entries.
The commands for which the command processor immediately executes code
are known as internal commands. When other commands are recognized, the
command processor loads a predefined file into memory and then executes
its contents. These commands are known as external commands, because the
code used to perform the functions associated with these commands resides
on a file external to the command processor’s code in memory.
Network Software Components
As soon as a networked personal computer is initialized, its network software
routines are added to DOS, permitting the computer to interact with the rest
of the network. Before the introduction of DOS Version 3.1, this software
was normally an overlay to DOS that served to filter commands, and that
translated those commands requiring network access into code to transmit
data via the LAN adapter card. Unfortunately, each network operating system
vendor developed code to interface the adapter card in a proprietary manner,
and software developed to work with one manufacturer’s adapter card may
not necessarily work with another manufacturer’s adapter card.
When a command is issued on the PC, the software overlay permits the
command to pass directly to DOS for execution. If a command is issued that
refers to the network, the software overlay intercepts or filters the command
from reaching DOS and, in conjunction with the adapter board, transmits
the command onto the network. If the network is server-based, the nonlocal
commands must be sent to a server for additional processing. The left-hand
portion of Figure 5.20 illustrates the hardware and software components
required when LAN software was originally designed as an overlay to DOS.
Before the introduction of DOS 3.1, most LAN operating system vendors developed proprietary methods to access the LAN adapter card, and
either developed proprietary methods to lock files and records or ignored
incorporating such features. In effect, this limited their networks to simple
file-swapping and printer-sharing applications. Because there was no Network Basic Input/Output System (NetBIOS), a proprietary network BIOS was
developed and accessed via the vendor’s LAN overlay software to send and
receive data from the LAN adapter card. Here, NetBIOS is the lowest level of
software on a local area network, translating commands to send and receive
data via the adapter card into the instructions that actually perform the
requested functions.
226
chapter five
Application program(s)
LAN overlay
software
DOS 3.0
or below
BIOS
Proprietary
network BIOS
LAN adapter card
IBM PC network program
DOS 3.1
or above
PC adapter
with NETBIOS
IBM PC LAN
program
DOS 3.1
or above
BIOS
NETBIOS
BIOS
Token-ring
Ethernet
adapter card adapter card
Figure 5.20 Original PC LAN hardware and software relationships in an IBM
PC environment.
With the introduction of IBM’s first local area network, referred to as the
PC Network, in August 1984, IBM released all three components required
to implement an IBM local area network using IBM equipment: the IBM PC
Network Program, PC DOS 3.1, and the IBM PC Network Adapter. The IBM PC
Network Program was actually a tailored version of Microsoft Corporation’s
Microsoft Networks (MS-NET) software, which is essentially a program that
overlays DOS and permits workstations on a network to share their disks and
peripheral devices. DOS 3.1, also developed by Microsoft, added file- and
record-locking capabilities to DOS, permitting multiple users to access and
modify data. Without file- and record-locking capabilities in DOS, custom
software was required to obtain these functions — without them, the last
person saving data onto a file would overwrite changes made to the file by other
persons. Thus, DOS 3.1 provided networking and application programmers
with a set of standards they could use in developing network software.
Included on the IBM PC Network Adapter card in ROM is an extensive
number of programming instructions, known as NetBIOS. The middle portion
of Figure 5.18 illustrates the hardware and software components of an IBM
PC LAN network.
When the IBM Token-Ring Network was introduced, NetBIOS was removed
from the adapter card and incorporated as a separate software program,
activated from DOS. The right-hand column of Figure 5.18 illustrates this new
relationship between hardware and software. At first, NetBIOS was designed
to operate with Token-Ring adapter cards. Later, IBM extended NetBIOS to
work with CSMA/CD Ethernet adapter cards.
networking hardware and software
227
Due to the standardization of file-and-record locking under DOS 3.1, any
multiuser software program written for DOS Version 3.1 or later will run on
any LAN that supports this version of DOS. Although DOS 3.1 supports many
networking functions, it was not a networking operating system. In fact, a
variety of networking operating systems support DOS 3.1 and later versions
of DOS, including MS-NET, IBM’s PC Network Program, IBM’s Token-Ring
Program, Microsoft’s Windows NT, and Novell’s NetWare. You can therefore
select a third-party network operating system to use with IBM or non-IBM
network hardware, or you can consider obtaining both third-party hardware
and software to construct your local area network.
Network Operating Systems
A modern network operating system operates as an overlay to the personal
computer’s operating system, providing the connectivity that enables personal computers to communicate with one another and share such network
resources as hard disks, CD-ROM jukebox drives, and printers, and even obtain
access to mainframes and minicomputers. Four of the more popular LAN
operating systems are Microsoft Corporation’s Windows NT, its successors,
Windows 2000 and Windows XP, and Novell Corporation’s NetWare.
Both versions of Windows and NetWare are file server–based network
operating systems. This means that most network modules reside on the file
server. A shell program loaded into each workstation works in conjunction
with the server modules. The shell program workstation filters commands,
directing user-entered commands to DOS or to the network modules residing
on the server. Communications between the shell and the server modules
occur at the OSI Reference Model’s Network Layer. Microsoft’s Windows
uses NetBIOS Extended User Interface, commonly referred to as NetBEUI,
which is automatically installed when the operating system is installed, while
Novell’s NetWare uses its Internetwork Packet Exchange (IPX) protocol as the
language in which the workstation communicates with the file server. Both
Windows and NetWare support the concurrent use of multiple protocols. For
example, Windows includes built-in support for TCP/IP, NWLink, and Data
Link control. Until the mid-1980s, it was difficult to support more than one
protocol at a time due to the manner by which network software residing on
a workstation or server communicated with one or more software modules
known as the protocol stack. Once we examine the manner by which a client
gains access to a server and obtain an overview of NetWare and Windows, we
will then turn our attention to the method by which multiple stacks can be
employed to support multiple protocols.
228
chapter five
Services
The process by which the shell enables a workstation to communicate with
a set of services residing on a server is known as a client/server relationship.
Services provided by network modules on the server can range in scope from
file access and transfer, shared printer utilization, and printer queuing to
electronic mail. Other features available in most network operating systems
include the ability to partition disk storage and allocate such storage to
different network users, and the assignment of various types of security levels
to individual network users, groups of users, directories, files, and printers.
Some network operating systems include a disk mirroring feature and a remote
console dial-in capability.
Because file information in the form of updated accounting, payroll, and
engineering data can be critical to the health of a company, it is often very
important to have duplicate copies of information in case a hard disk should
fail. Disk mirroring is a feature that duplicates network information on two
or more disks simultaneously. Thus, if one disk fails, network operations
can continue.
A remote console dial-in capability enables a network user to gain access
to the network from a remote location. This feature can be particularly
advantageous for people who travel and wish to transmit and receive messages
with people back at the office or obtain access to information residing on
the network. Because the administration of a network can be a complex
process, a remote dial-in feature may also make life less taxing for a network
administrator. Working at home or at another location, the administrator can
reassign privileges and perform other network functions that may not be
possible in an eight-hour day.
Looking at NetWare
Because the best way to obtain information concerning the relationship of a
network operating system to network hardware is to examine the software,
we will do so. We will discuss Novell Corporation’s NetWare and Microsoft’s
Windows, as those network operating systems (NOS) are by far the most
popular of all network operating systems used.
Architecture
The architecture or structure of NetWare can be mapped to the OSI Reference
Model. It provides an indication of the method by which this network operating system provides support for different types of hardware, and includes the
networking hardware and software
OSI
reference
model
Application
Presentation
Session
229
Netware
Netware
Application
Net BIOS
emulation
Netware
shell
(workstation)
Core protocol
(NCP)
(on server)
Transport
Sequenced packet exchange (SPX)
Network
Internet packet exchange
(IPX)
Data link
Token-ring Ethernet
ARCnet
Others
Physical
Figure 5.21
NetWare and the OSI Reference Model.
capability for the routing of packets between networks. Figure 5.21 illustrates
the general relationship between NetWare and the OSI Reference Model.
In examining Figure 5.21, note that NetWare supports numerous types of
local area networks. This means that you can use NetWare as the network
operating system on Ethernet, Token-Ring, ARCnet, and other types of networks. In fact, NetWare also supports different types of personal computer
operating systems, such as DOS, OS/2, different versions of Windows, UNIX,
and Macintosh. This means that NetWare is capable of supporting different types of local area networks as well as workstations that use different
operating systems.
Using NetWare on a PC requires the loading of two Novell files whenever
you turn on your computer or perform a system reset. Those files are IPX and
NETx, where x indicates a specific version of the NET file used with a specific
version of DOS, such as NET3 used with DOS 3.
The use of IPX and NETx are required through NetWare Release 3.11.
In Release 3.12 and in NetWare Version 4.X and later versions of this network operating system, NETx was replaced by the use of a virtual loadable
module (VLM). Later in this section, we will discuss the use of NetWare’s
VLM.EXE program.
Both IPX and NET are workstation shell programs that interpret and filter
commands entered from the keyboard and provide a mechanism for communications between the workstation and the server. Before NetWare Version 2.1,
the shell was known as ANET3.COM, and was combined with IPX and NETx
into one file. Later versions of NetWare separated IPX from NETx.
230
chapter five
To automate the loading of NetWare on your workstation to establish a
network connection, you would normally insert appropriate commands into
your computer’s AUTO-EXEC.BAT file. Those commands would include:
IPX
NETx
F:
LOGIN <servername/username>
IPX The command IPX would cause Novell’s IPX.COM file to be loaded
and executed. This file is a driver that establishes communications with the
network interface board, and it is responsible for providing communications
with servers and other stations on the network using a network protocol known
as IPX. At the network layer, Novell’s IPX protocol performs addressing and
internet routing functions. To accomplish this, an IPX packet contains both
the source and destination network addresses. Those addresses are assigned
by a network administrator, and they provide the mechanism for the routing
of data between networks by routers which examine the network layer.
IPX is a connectionless network layer protocol that does not guarantee the
delivery of data. To provide a reliable delivery mechanism, Novell developed its Sequenced Packet eXchange (SPX) — a transport level interface that
provides a connection-oriented packet delivery service.
NCP At the session and presentation layers, NetWare uses a NetBIOS emulator, which provides an interface between application programs written in
compliance with NetBIOS and NetWare. As previously mentioned, the NetWare shell operates on each workstation and communicates with a core set
of modules that reside on servers. That core set of modules is known as
the NetWare Core Protocol (NCP). NCP provides such functions as workstation and network naming management, file partitioning, access and locking
capabilities, accounting, and security.
NET The command NETx loads NETx.COM, which is the true workstation
shell, because it interprets and filters commands entered from the keyboard.
In addition, NETx supports a large number of NetWare commands, which,
when entered, are converted into IPX packets and transmitted to the server
for processing. The NetWare Core Protocol decodes the command request,
processes the request, and then transmits a response to the workstation using
one or more IPX packets. The workstation’s NET module then processes and
displays the response. For example, typing the NetWare command CHKVOL at
the workstation transmits a request to the server to obtain statistics concerning
the logical driver (volume) assigned to the workstation user. The results of that
networking hardware and software
231
request are transmitted back to the workstation and displayed on its screen
the same way a DOS CHKDSK command is displayed.
When the shell (NETx) is loaded, it normally establishes a connection to a
network server by sending a request to IPX to broadcast a Get Nearest Server
command. The first server that responds to the request then establishes a
connection to the workstation and displays the message ‘‘Attached to server
<servername>’’ on your computer’s console. You can also specify a preferred
server by adding the PS = parameter to the NETx command; this provides you
with the ability to distribute workstation server usage over a number of servers.
Once a connection to a NetWare server occurs, the command F: in the
AUTOEXEC.BAT file moves the workstation user to the server’s SYS:LOGIN
directory. That directory is designated or mapped to drive F: on your DOSoperated workstation. Once this is accomplished, the command LOGIN
initiates the LOGIN module on the server. If you include the servername
and username, LOGIN will then request only your password to obtain access
to the server.
Versions Several versions of NetWare have been marketed during the past
ten years. NetWare 286, which was renamed NetWare 2.2, was designed to
operate on Intel 286–based servers. This operating system supported up to
100 users. NetWare 386 (renamed NetWare 3.1), operated on Intel 386–based
servers. This network operating system supported up to 250 users.
The introduction of NetWare 4.0 and the release of NetWare 4.1, followed
by releases 5.0 and 6.0, extended Novell’s NetWare support to local area
networks consisting of up to several thousand workstations. As previously
discussed, NetWare 3.12 as well as all versions of NetWare 4.X resulted
in the replacement of NETx by the virtual loadable module VLM.EXE. By
including the command VLM.EXE in your AUTOEXEC.BAT file, you would
cause the executable virtual loadable module to be loaded. This executable file
will automatically load a number of files with the .VLM extension, tailoring
NetWare to your workstation.
A second change to NetWare is the fact that in November 1991 Novell
ceased supporting its dedicated IPX driver. IPX was specific to the network
interface card and version of NetWare being used on a workstation, and
required you to create a new version each time you installed a new network
card. A second problem associated with IPX is the fact that once used
with an adapter card, you cannot use another protocol with that card. For
example, if you want to communicate using TCP/IP to a UNIX server with
the same card, you would have to change your AUTOEXEC.BAT file, remove
or comment out via REM statements your invocation of IPX and NETx, add
232
chapter five
your TCP/IP commands, and reboot your computer. Obviously this was not a
pleasant situation.
Recognizing the preceding problems, Novell released a new architecture
known as the Open Data-Link Interface (ODI) in 1989. By 1991, ODI became
the only IPX standard interface supported by Novell. Through the use of
ODI, you can support multiple protocols through a common adapter without
requiring the rearrangement of statements in your AUTOEXEC.BAT file and
rebooting your computer. To do so, you must obtain the following special
files — LSL, IPXODI, and an interface driver. LSL is a link support layer
program that you must obtain from Novell. The interface driver is provided
by the manufacturer of the adapter card, while IPXODI is furnished by both
Novell and the adapter card manufacturer.
Figure 5.22 illustrates the relationship of the three previously mentioned
programs when a multiprotocol or dual stack operation is desired. The interface driver provides low-level I/O operations to and from the adapter card,
and passes information received from the LAN to the Link Support Program.
That program examines incoming data to determine if it is NetWare (IPX) or IP
(TCP/IP) in the example illustrated in Figure 5.22. LSL then passes received
data to the appropriate stack. Thus, IPXODI represents a modification to IPX,
which permits it to interface Novell’s LSL program.
Although LSL resides on top of the interface driver, you must load it before
loading that driver. Thus, your AUTOEXEC.BAT file would have the following
generic entries to support ODI on your workstation:
LSL
HRDRIVER
IPXODI
NETx
TCP/IP
IPXODI
Link support layer
(LSL)
Interface driver
Network adapter card
Figure 5.22
ell’s ODI.
Multiprotocol support using Nov-
networking hardware and software
233
NETX
F:
LOGIN
In examining the preceding entries, note that HRDRIVER would be replaced
by the actual name of your adapter card’s interface driver. In addition, under
NetWare 3.12 and 4.X and later versions of this operating system, you would
replace NETx with VLM.
To add the TCP/IP protocol stack under DOS you would add the appropriate
statements to your AUTOEXEC.BAT file. Those statements must follow the
execution of LSL.COM but can either precede or succeed the statements used
to invoke the NetWare protocol stack. For example, assume NetWare files
are located in the NetWare directory and the appropriate packet driver is
contained in the file ODIPKT and the TCP/IP program is contained in the file
TCPIP, while both the ODIPKT and TCP/IP files are located in the directory
TCP. Then, the AUTOEXEC.BAT file would contain the following statements
with the REM(ark) statements optionally added for clarity.
REM *Install NetWare*
C:\NETWARE\LSL.COM
C:\NETWARE\LANDRIVER
C:\NETWARE\IPXODI.COM
C:\NETWARE\NETx.EXE
F:
LOGIN GHELD
REM *Install TCP/IP*
C:\TCP\ODIPKT
C:\TCP\TCPIP
NET.CFG One important file not shown in Figure 5.22 and until now not
discussed is NET.CFG. This file describes the network adapter card configuration to the ODI driver and should be located in the same directory as the
IPXODI and NETx files. However, through the use of an appropriate PATH
statement you can actually locate NET.CFG anywhere you desire.
NET.CFG is an ASCII text file that can contain up to four main areas of
information, which describe the environment of a workstation. Those areas
include a link support area, protocol area, link driver area, and parameter area.
Link Support Area The link support area is used to define the number
of communications buffers and memory set aside for those buffers. This
area is required to be defined when running TCP/IP, however, because
234
chapter five
IPX does not use buffers or memory pools maintained by LSL you can
skip this section if you are only using a NetWare protocol stack. The
following illustration represents an example of the coding of the link support area in the NET.CFG file to support TCP/IP. The actual coding you
would enter depends upon the network adapter card to be used and you
would obtain the appropriate information from the manual accompanying the
adapter card.
LINK SUPPORT
BUFFERS 8 1144
MemPool 4096
MaxStacks 8
Protocol Area The protocol area is used to bind one or more protocols
to specific network adapter cards. By default, IPXODI binds to the network adapter in the lowest system expansion slot as it scans slots in their
numeric order. If you have two or more network adapter cards in a workstation, you can use the protocol area to specify which protocols you want
to bind to each card. You can also accomplish this at the link driver area
by specifying Slot n, where n is the slot number of the network adapter
card you are configuring. Assuming you wish to bind IPX to an adapter
card whose address is h123, you would add the following statements to the
NET.CFG file.
Protocol
PROTOCOL IPX
BIND h123
Because each computer using TCP/IP requires an IP address, the IP address
information must be included in the NET.CFG file if you intend to use
the TCP/IP protocol stack. For example, if the network administrator assigned
your computer the IP address 133.49.108.05, the IP address information would
be entered as follows:
PROTOCOL TCP/IP
ip− address 133.49.108.05
When using TCP/IP, each workstation on the network is assigned the
address of a default router (commonly referred to as a gateway) by the network
administrator. Thus, another statement commonly added to the NET.CFG file
includes the address of the router that the workstation will use. For example,
networking hardware and software
235
if the router’s address is 133.49.108.17, then you would add the following
statement to the NET.CFG file in its protocol area.
ip− router 133.49.108.17
The ip− address and ip− router statements can be avoided if the network
administrator sets up a Reverse Address Resolution Protocol (RARP) server
configured with IP and hardware addresses for workstations on the network.
Then, when the workstation is powered on it will broadcast an RARP packet
that will contain its hardware address. The RARP server will respond with
the workstation’s IP address associated with the hardware address.
Link Driver Area The link driver area is used to set the hardware configuration of the network adapter card so it is recognized by LAN drivers. If you are
only using Novell’s IPX, the first line of your NET.CFG file is a LINK DRIVER
statement which tells NETX the type of LAN card installed in the workstation,
such as
Link Driver 3C5X9
The reason this statement becomes the first statement is because the link
support area is omitted and, if you only have one adapter card, you do not
require a protocol area.
If you’re using an NE 2000 Ethernet card, your link driver area would appear
as follows:
Link Driver NE2000
INT 5
PORT 300
Frame Ethernet− 802.3
Frame Ethernet− II
Protocol IPX 0 Ethernet− 802.3
Protocol IP 8137 Ethernet− II
In this example the frame statements define the types of frames that will be
supported by the adapter cards. Although most adapter cards include software
that automatically construct or modify the NET.CFG file, upon occasion you
may have to customize the contents of that file. To do so you can use the
manual accompanying the network adapter card, which will normally indicate
the statements required to be placed in the file.
236
chapter five
Virtual Loadable Modules The introduction of NetWare 4.0 resulted in the
replacement of NETX by VLMs that sit behind DOS. In comparison, NETX
sat in front of DOS and acted as a filter to identify and act upon network
requests entered from the keyboard. VLMs are referred to as the NetWare DOS
Requester as they use DOS redirection to satisfy file and print service requests.
Because VLMs replace NETX.EXE, you would load VLM.EXE in the position
previously used for NETX.EXE. That is, the sequence of commands placed
in your AUTOEXEC.BAT file to initialize the NetWare protocol stack would
appear as follows:
C:\NETWARE\LSL
C:\NETWARE\LANDRIVER
C:\NETWARE\IPXODI
C:\NETWARE\VLM.EXE
F:
LOGIN GHELD
To modify the AUTOEXEC.BAT file to support dual-stack operations you
could add the appropriate commands either after invoking LSL or after the
‘‘Login’’ statement.
Looking at Windows
Windows, to include workstation and server versions of NT, 2000 and XP,
represents both a computer operating system and network operating system
that can function together or independently. The basic networking protocol
used by each version of Windows is NetBEUI, which provides a network user
interface for local workstations and servers.
NetBIOS The NetBIOS Extended User Interface (NetBEUI) represents an
extension of PC BIOS to the network. NetBIOS was originally developed by
IBM as a simple network protocol for interconnecting PCs on a common
network. The naming structure of the protocol results in names assigned to
devices being translated into network adapter card (that is, MAC) addresses.
This results in NetBIOS operating at the data link layer. In addition, because
the NetBIOS naming structure is nonhierarchical, there is no provision for
specifying network addresses. Due to this, NetBIOS is considered to be
nonroutable. Thus, the initial method used to join two or more NetBIOS
networks together was restricted to bridging.
NetBEUI Recognizing the routability problem of NetBIOS, NetBEUI allows
data to be transported by a transport protocol to obtain the ability to
networking hardware and software
237
interconnect separate networks. In fact, NetBEUI can be transported by
TCP/IP and even IPX/SPX. To accomplish this, NetBEUI maintains a table
of NAMES that are associated with TCP/IP addresses when TCP/IP is
used as a transport protocol, and a similar table matched to NetWare network addresses and station MAC addresses when NetBEUI is transported
via IPX/SPX.
To illustrate the operation of a few of the capabilities of Windows networking, we will briefly use a Windows NT workstation and a Windows NT
server to illustrate the installation of network software and adapter cards.
In addition, we will use a Windows NT workstation to display the servers
on a network where both NT and NetWare servers reside, transferring a file
from an NT workstation to a Novell file server. Both NetWare and Windows NT can communicate on a common network, because NT supports the
NWLink protocol that provides communications compatibility with NetWare’s
IPX/SPX protocol.
Adapter and Software Support Windows workstation and server products
use common methods to add support for network software and adapter cards.
Although the screen display for configuring network software and adapter
cards varies between versions of Windows, the basic methods remain the
same. Thus, although Figure 5.23 illustrates the network settings screen for
Version 3.51 of NT, the basic methods we will describe are also applicable to
other versions of NT, Windows 2000 and Windows XP.
In examining Figure 5.23, note that five network software modules are
shown in the upper box labeled Installed Network Software, and one adapter
card is shown as being installed in the lower box labeled Installed Adapter
Card. Windows supports the binding of multiple protocols to a common
adapter via the use of the network driver interface specification (NDIS), which
will be described at the end of this section. You can add network software,
such as TCP/IP, by clicking on the Add Software button shown in Figure 5.23.
This action will result in the display of a list of networking software directly
supported by Windows. Similarly, if you want to add another adapter you
would click on the Add Adapter button. If the adapter you wish to add is
not directly supported by Windows, you can select the option ‘‘Other — have
disk’’ at the end of the list of supported adapters. This will allow you to add
support for a wide range of NICs that are commonly shipped with Windows
drivers, but which are not directly supported by the version of Windows you
are using.
Network Operation Figure 5.24 illustrates the use of File Manager on a
Windows NT workstation to view the names of devices on both a Windows
238
chapter five
Figure 5.23 Using the Windows NT dialog box to review, add, or change
network software and adapter card support.
Figure 5.24 Viewing devices on both a Windows and a Novell network
through the Windows NT File Manager.
networking hardware and software
239
Figure 5.25 Selecting a path to a directory on a Novell server that will be
mapped to drive E on a local workstation.
network and a NetWare network. Figure 5.25 illustrates the result obtained by
first selecting an appropriate NetWare server and then selecting a directory
on that server that we wish to access. This action will result in the mapping
of drive E on the local workstation to the path shown in Figure 5.25. Once we
enter the appropriate connection information, drive E on the local Windows
NT workstation will be mapped to the directory FRED located under the
directory SYS on the server MDPC-1.
After we correctly log onto the server, we can run network applications
or transfer data to or from the server. Figure 5.26 illustrates how you could
select ‘‘Move’’ from the File menu and enter the command c:\funds\*.* to
move all files under the subdirectory FUNDS on the local workstation to the
network server.
NDIS Operation Considerations Similar to the manner by which Novell
developed an architecture for supporting multiple protocols via a common
adapter, Microsoft developed a competing standard referred to as NDIS. In this
chapter five
240
Figure 5.26 Using File Manager to move all files in the directory FUNDS on
the local workstation to the directory FRED on the file server.
section we will focus our attention upon obtaining an overview of the structure
of NDIS, even though it is well-hidden from view when you use a Windows
operating environment. Although NDIS provides a dual-stack capability similar to that provided by ODI, its setup for operation varies considerably from
the previously discussed dual-stack mechanism. Figure 5.27 illustrates the
Protocol
stack
N
D
I
S
Protocol
stack
LAN support program(s)
NDIS protocol manager
Network adapter driver
Network adapter card
Figure 5.27
NDIS architecture.
networking hardware and software
241
relationship between NDIS software modules, upper-layer protocol stacks,
and the network adapter card.
CONFIG.SYS Usage Unlike ODI, which represents a series of files loaded
from an AUTOEXEC.BAT file, NDIS was designed as a series of device drivers
that are loaded through the CONFIG.SYS file. In a DOS environment the first
statement in the CONFIG.SYS file required for NDIS is:
DEVICE=drive:\path\PROTMAN.DOS
PROTMAN.DOS represents the NDIS Protocol Manager for each workstation
operating DOS. The Protocol Manager reads the file PROTOCOL.INI, which
contains initialization parameters and stores the contents of that file in memory
for use by other NDIS drivers. Thus, a short discussion of PROTOCOL.INI file
is in order.
PROTOCOL.INI Overview The PROTOCOL.INI file can be considered to
represent the NDIS equivalent of the NET.CFG file associated with ODI.
Although most network products including various versions of Windows will
automatically create or modify the PROTOCOL.INI file, some products require
users to create or modify that file. In addition, you may be able to enhance
network performance by modifying an existing parameter set by a network
program that does not consider your total user environment.
Entries in PROTOCOL.INI occur in sections, with each section name surrounded in brackets ([]). Under each section name are one or more named
configuration entries, which appear in the format ‘‘name = value’’. Although
configuration entries can appear anywhere in a line under the section name,
normal practice is to indent each entry three character positions to enhance
readability.
Depending upon the version of Windows you are using, the first section in
the PROTOCOL.INI file may have the heading [PROTMAN− MOD]. The first
configuration entry for DOS is the device name PROTMAN$. Thus, the first
section entry becomes:
[PROTMAN− MOD]
DriverName = PROTMAN
Other versions of Windows may commence with an NDIS help section
followed by the PROTMAN− MOD section.
Other entries in the [PROTMAN− MOD] section are optional and can include
keywords Dynamic, Priority, and Bindstatus in assignment statements. The
242
chapter five
Dynamic statement can be set to ‘‘YES’’ (Dynamic = YES) to support both static
and dynamic binding or ‘‘NO’’ (Dynamic = NO) to set the Protocol Manager
to operate only in static mode, which is its default. In static mode protocol
drivers are loaded once at system initialization and remain in memory. In
the dynamic mode drivers load at the point in time when they are bound by
Protocol Manager. In addition, if the drivers support a dynamic unloading
capability they can be unloaded if the software unbinds them when they are
not needed, freeing memory.
The Priority keyword is used to specify the order of priority of protocol
processing modules. Under NDIS an incoming LAN packet is first offered to
the protocol with the highest priority. Other protocols will see the packet only
if a higher protocol does not first recognize and process the packet. Protocols
not specified in a priority list are the last to inspect incoming packets.
The Bindstatus keyword is used to specify whether Protocol Manager can
optimize memory and can be set to ‘‘YES’’ or ‘‘NO’’. If the keyword is not
used, a default of ‘‘NO’’ is assumed.
The second communications statement included in a CONFIG.SYS file for
NDIS operations invokes the network adapter card driver. For example, if you
were using the NE2000 adapter, you would include the following statement
in the CONFIG.SYS file.
DEVICE=[drive:]\path\NE2000.DOS
NDIS Adapter Support The adapter driver, which is compatible with the
NDIS Protocol Manager, is referred to as an NDIS MAC driver. The NDIS
MAC driver is normally contained on a diskette that is included in a box
in which your NDIS-compatible network adapter is packaged. When using
Windows NT the operating system includes built-in NDIS support for approximately 30 adapter cards. As previously explained, if the adapter you are
using is not directly supported by Windows NT, you would select the
Other option from the install adapter card entry from the network configuration display obtained from the Windows Control Panel. Then you
would use the diskette that accompanies your adapter card to install the
required driver.
Once you install your adapter card and appropriate communications protocols under Windows, the operating system will automatically connect the
software layers as required to form appropriate protocol stacks. Microsoft
refers to this as network bindings, and Figure 5.28 illustrates an example
of the NT Network Bindings display after a large number of protocols
were installed.
networking hardware and software
Figure 5.28
display.
243
Viewing an example of the Windows NT Network Bindings
Application Software
The third major component of software required for productive work to
occur on a local area network is application software. These application
programs support electronic mail, multiple access to database records, or
the use of spreadsheet programs; they operate at the top layer of the OSI
Reference Model.
Until the mid-1980s, most application programs used on LANs were not
tailored to operate correctly in a multiuser environment. A large part of their
inability to work correctly was due to the absence of file- and record-locking
capabilities on PC operating systems — a situation that was corrected with
the introduction of DOS 3.1. A second problem associated with application
programs occurred when the program was written to bypass the personal
computer’s BIOS. Although this action in many instances would speed up
screen displays, disk access, and other operations, in this case it resulted in
nonstandardized program actions. This made it difficult, if not impossible, for
some network operating systems to support ill-defined programs, because an
interrupt clash could bring the entire network to a rapid halt.
Today, most application programs use BIOS calls and are well defined. Such
programs are easily supported by network operating systems. A few programs
that bypass BIOS may also be supported, because the application program
that caused operating system vendors to tailor their software to support such
applications was so popular.
244
chapter five
5.4 The TCP/IP Protocol Suite
No discussion of networking hardware and software related to Ethernet
would be complete without covering the TCP/IP protocol suite. Although
the development of TCP/IP occurred at the Advanced Research Projects
Agency (ARPA), which was funded by the U.S. Department of Defense, while
Ethernet traces its origin to the Xerox Palo Alto Research Center, within a
short period of time the two were linked together. Ethernet frames provide
the data link (layer 2) transportation mechanism for the delivery of network
layer (layer 3) IP and transport layer (layer 4) TCP packets that transport such
application data as file transfer, remote access, and Web server information on
an intra-LAN basis. In comparison, TCP/IP provides the mechanism to route
data between LANs and convert IP addresses used by the protocol suite to
MAC addresses used by Ethernet so that TCP/IP packets can be delivered by
Ethernet frames.
Overview
TCP/IP represents a collection of network protocols that provide services at
the network and transport layers of the ISO’s OSI Reference Model. Originally
developed based upon work performed by the U.S. Department of Defense
Advanced Research Projects Agency Network (ARPANET), TCP/IP is also
commonly referred to as the DOD protocols or the Internet protocol suite.
Protocol Development
In actuality, a reference to the TCP/IP protocol suite includes applications that
use the TCP/IP protocol stack as a transport mechanism. Such applications
range in scope from a remote terminal access program known as Telnet to a file
transfer program appropriately referred to as FTP, as well as the Web browser
transport mechanism referred to as the HyperText Transport Protocol (HTTP).
The effort behind the development of the TCP/IP protocol suite has its roots
in the establishment of ARPANET. The research performed by ARPANET
resulted in the development of three specific protocols for the transmission of
information — the Transmission Control Protocol (TCP), the Internet Protocol
(IP), and the User Datagram Protocol (UDP). Both TCP and UDP represent
transport layer protocols. Transmission Control Protocol provides end-to-end
reliable transmission while UDP represents a connectionless layer 4 transport
protocol. Thus, UDP operates on a best-effort basis and depends upon higher
layers of the protocol stack for error detection and correction and other
networking hardware and software
245
functions associated with end-to-end reliable transmission. Transmission
Control Protocol includes such functions as flow control, error control, and
the exchange of status information, and is based upon a connection being
established between source and destination before the exchange of information
occurs. Thus, TCP provides an orderly and error-free mechanism for the
exchange of information.
At the network layer, the IP protocol was developed as a mechanism to
route messages between networks. To accomplish this task, IP was developed
as a connectionless mode network layer protocol and includes the capability
to segment or fragment and reassemble messages that must be routed between
networks that support different packet sizes than the size supported by the
source and/or destination networks.
The TCP/IP Structure
TCP/IP represents one of the earliest developed layered communications protocols, grouping functions into defined network layers. Figure 5.29 illustrates
ISO Layers
FTP
5−7
TELNET
SMTP
DNS
TCP
4
NFS
SNMP
UDP
ICMP
3
IP
ARP
Ethernet 802.3
2
Token ring 802.5
Legend:
ARP = Address Resolution Protocol
DNS = Domain Name Service
FDDI = Fiber Data Distributed Interface
FTP = File Transfer Protocol
NSF = Network File System
SMTP = Simple Mail Transfer Protocol
SNMP = Simple Network Management Protocol
Figure 5.29
TCP/IP protocols and services.
FDDI
246
chapter five
the relationship of the TCP/IP protocol suite and the services they provide
with respect to the OSI Reference Model. In examining Figure 5.29 note that
only seven of literally hundreds of TCP/IP application services are shown.
Because TCP/IP preceded the development of the OSI Reference Model, its
developers grouped what are now session, presentation, and application layers that correspond to layers 5 through 7 of the OSI Reference Model into
one higher layer. Thus, TCP/IP applications, when compared with the OSI
Reference Model, are normally illustrated as corresponding to the upper three
layers of that model. Continuing our examination of Figure 5.29, you will note
that the subdivision of the transport layer indicates which applications are
carried via TCP and those that are transported by UDP.
As we will note later in this section, TCP represents a connection-oriented
error-free transport protocol. This means that it is well suited for transporting
applications that require the acknowledgement of the availability of a distant
device prior to the actual transfer of data, such as a file transfer application. In
comparison, UDP represents a best-effort, unreliable transport protocol. This
means that UDP can immediately be used to transport data without requiring
a prior handshaking operation to be successful. This also means that data is
transmitted under UDP without error detection and correction, making the
application responsible for deciding if this is needed.
Thus, FTP, Telnet, HTTP, and SMTP represent applications transported
by TCP that require a connection to be established prior to data being transported and need an error detection and correction capability. Domain Name
Service (DNS), Network File System (NFS), and Simple Network Management
Protocol (SNMP) represent applications that do not require a prior connection
and occur on a best effort basis. Thus, DNS, NFS and SNMP are transported
via UDP.
While the prior examples of TCP and UDP usage are well defined, it should
be noted that some applications, such as Internet Telephony, use both transport protocols. For example, call control information such as a dialed number
must flow through the Internet error-free and are carried via TCP. In comparison, real-time digitized voice cannot be retransmitted when errors are detected
since this would result in awkward delays at the receiver. Thus, the actual
digitized voice portion of an Internet Telephony call is transported via UDP.
Although not officially layer 3 protocols, both the Address Resolution Protocol (ARP) and the Internet Control Message Protocol (ICMP) reside in a
‘‘gray’’ area and are commonly shown as residing at that location, so we will
also do this. In addition, because ICMP, as we will shortly note, is transported
with an IP header, it makes sense to consider it residing within layer 3 of the
TCP/IP protocol stack.
networking hardware and software
247
Returning to our examination of Figure 5.29, note that TCP/IP can be
transported at the data link layer by a number of popular LANs, to include
Ethernet, Fast Ethernet, Gigabit Ethernet, Token-Ring, and FDDI frames. Due to
the considerable effort expended in the development of LAN adapter cards to
support the bus structures used in Apple MacIntosh, IBM PCs and compatible
computers, DEC Alphas and SUN Microsystem’s workstations, and even IBM
mainframes, the development of software-based protocol stacks to facilitate
the transmission of TCP/IP on LANs provides the capability to interconnect
LAN-based computers to one another whether they are on the same network
and only require the transmission of frames on a common cable, or if they are
located on networks separated thousands of miles from one another. Thus,
TCP/IP represents both a local and wide area network transmission capability.
Datagrams versus Virtual Circuits
In examining Figure 5.29 you will note that IP provides a common layer 3
transport for TCP and UDP. As briefly noted earlier in this section, TCP
is a connection-oriented protocol that requires the acknowledgment of the
existence of the connection and for packets transmitted once the connection
is established. In comparison, UDP is a connectionless mode service that
provides a parallel service to TCP. Here datagram represents a term used to
identify the basic unit of information that represents a portion of a message
and that is transported across a TCP/IP network.
A datagram can be transported either via an acknowledged connectionoriented service or via an unacknowledged, connectionless service, where
each information element is addressed to its destination and its transmission
is at the mercy of network nodes. IP represents an unacknowledged connectionless service; however, although it is an unreliable transmission method,
you should view the term in the context that delivery is not guaranteed
instead of having second thoughts concerning its use. As a nonguaranteed
delivery mechanism IP is susceptible to queuing delays and other problems
that can result in the loss of data. However, higher layers in the protocol suite,
such as TCP, can provide error detection and correction, which results in the
retransmission of IP datagrams.
Datagrams are routed via the best path available to the destination as the
datagram is placed onto the network. An alternative to datagram transmission
is the use of a virtual circuit, where network nodes establish a fixed path when
a connection is initiated and subsequent data exchanges occur on that path.
TCP implements transmission via the use of a virtual circuit, while IP provides
a datagram-oriented gateway transmission service between networks.
248
chapter five
The routing of datagrams through a network can occur over different paths,
with some datagrams arriving out of sequence from the order in which they
were transmitted. In addition, as datagrams flow between networks they
encounter physical limitations imposed upon the amount of data that can
be transported based upon the transport mechanism used to move data on
the network. For example, the information field in an Ethernet frame is
limited to 1500 bytes, while a 4-Mbps Token-Ring can transport 4500 bytes
in its information field. Thus, as datagrams flow between networks, they
may have to be fragmented into two or more datagrams to be transported
through different networks to their ultimate destination. For example, consider the transfer of a 20,000-byte file from a file server connected to a
Token-Ring network to a workstation connected to an Ethernet LAN via
a pair of routers providing a connection between the two local area networks. The 4-Mbps Token-Ring network supports a maximum information
field of 4500 bytes in each frame transmitted on that network, while the
maximum size of the information field in an Ethernet frame is 1500 bytes.
In addition, depending upon the protocol used on the wide area network
connection between routers, the WAN protocol’s information field could be
limited to 512 or 1024 bytes. Thus, the IP protocol must break up the file
transfer into a series of datagrams whose size is acceptable for transmission
between networks. As an alternative, IP can transmit data using a small
maximum datagram size, commonly 576 bytes, to prevent fragmentation. If
fragmentation is necessary, the source host can transmit using the maximum datagram size available on its network. When the datagram arrives at
the router, IP operating on that communications device will then fragment
each datagram into a series of smaller datagrams. Upon receipt at the destination, each datagram must then be put back into its correct sequence so
that the file can be correctly reformed, a responsibility of IP residing on the
destination host.
Figure 5.30 illustrates the routing of two datagrams from workstation 1 on
a Token-Ring network to server 2 connected to an Ethernet LAN. As the
routing of datagrams is a connectionless service, no call setup is required,
which enhances transmission efficiency. In comparison, when TCP is used, it
provides a connection-oriented service regardless of the lower-layer delivery
system (for example, IP).
TCP requires the establishment of a virtual circuit in which a temporary
path is developed between source and destination. This path is fixed and the
flow of datagrams is restricted to the established path. When UDP, a different
layer 4 protocol in the TCP/IP protocol suite, is used in place of TCP, the flow
of data at the transport layer continues to be connectionless and results in the
networking hardware and software
Token
ring
1
Router
A
Router
C
249
1
2
Router
E
2
Router
B
Router
D
2
Figure 5.30
Ethernet
LAN
Routing of datagrams can occur over different paths.
transport of datagrams over available paths rather than a fixed path resulting
from the establishment of a virtual circuit.
The actual division of a message into datagrams is the responsibility of the
layer 4 protocol, either TCP or UDP, while fragmentation is the responsibility
of IP. In addition, when the TCP protocol is used, that protocol is responsible
for reassembling datagrams at their destination as well as for requesting the
retransmission of lost datagrams. In comparison, IP is responsible for routing
of individual datagrams from source to destination. When UDP is used as the
layer 4 protocol, there is no provision for the retransmission of lost or garbled
datagrams. As previously noted by our discussion of IP, this is not necessarily
a bad situation, as applications that use UDP then become responsible for
managing communications.
Figure 5.31 illustrates the relationship of an IP datagram, UDP datagram, and
TCP segment to a LAN frame. The headers shown in Figure 5.31 represent a
group of bytes added to the beginning of a datagram to allow a degree of control
over the datagram. For example, the TCP header will contain information that
allows this layer 4 protocol to track the sequence of the delivery of datagrams
so they can be placed into their correct order if they arrive out of sequence.
Before focusing our attention on TCP and IP, let’s discuss the role of ICMP
and ARP, two additional network layer protocols in the TCP/IP suite.
ICMP
The Internet Control Message Protocol (ICMP) provides a mechanism for
communicating control message and error reports. Both gateways and hosts
use ICMP to transmit problem reports about datagrams back to the datagram
originator.
250
chapter five
UDP or
TCP header
Layer 4
Layer 3
Layer 2
Application data
IP header
LAN
header
UDP or
TCP header
IP header
Application data
LAN
trailer
UDP datagram
or TCP segment
IP datagram
placed in information field
LAN frame
Figure 5.31
Forming a LAN frame.
An ICMP message is formed by prefixing an IP header to the ICMP message.
Each ICMP message consists of four fields, of which only two are mandatory.
Figure 5.32 illustrates the formation of an IPMP message to include the fields
in the ICMP message.
In Figure 5.32 the Type field defines the type of ICMP message. The code
field can optionally further define the reason for the ICMP message. For
example, a type field value of 3 indicates a Destination Unreachable ICMP
message, which is returned to the originator to inform them that their transmitted datagram cannot reach its destination. The code field value further defines
why the destination was unreachable, with a value of 1 indicating the network was unreachable while a value of 2 indicates the host was unreachable,
ICMP message
IP header
Bits
Figure 5.32
Type
field
(8)
Code
field
(8)
Checksum
field
(16)
Data
field
(32)
Formation and composition of an ICMP message.
networking hardware and software
251
and so on. Not all ICMP messages need further elaboration and as a result
some messages do not have any code field values. The checksum field represents a 16-bit one’s complement of the one’s complement sum of the ICMP
message commencing with the type field. The Data field may or may not be
used depending upon the message. Table 5.2 provides a summary of ICMP
messages to include their type and code values.
In examining the entries in Table 5.3 a few words are in order concerning
their meaning and utilization. ICMP Type 0 and Type 8 messages form the
basis for the application known as Ping. Ping results in the transmission of
TABLE 5.2
Type Value
ICMP Message Type and Code Values
Message/Code Values
0
Echo Reply
3
Destination Unreachable
0 = network unreachable
1 = host unreachable
2 = protocol unreachable
3 = port unreachable
4 = fragmentation needed
5 = source route failed
4
Source Quench
5
Redirect
0 = redirect datagrams for the network
1 = redirect datagrams for the host
2 = redirect datagrams for the type of service and the network
3 = redirect datagrams for the type of service and the host
8
Echo
11
Time Exceeded
0 = time to live exceeded in transit
1 = fragment reassembly time exceeded
12
Parameter Problem
0 = pointer in data field indicates the error
13
Timestamp
14
Timestamp Reply
15
Information Request
16
Information Reply
chapter five
252
TABLE 5.3 Examples of TCP/IP Application Layer Protocol Use of WellKnown Ports
Description
Well-Known
Port
Name
Acronym
Domain Name Protocol
DOMAIN Defines the DNS
53
File Transfer Protocol
FTP
Supports file transfers between
hosts
20,21
Finger Protocol
FINGER
Provides information about a
specified user
79
HyperText
Transmission Protocol
HTTP
Transmits information between a
Web browser and a Web server
80
Post Office Protocol
POP
Enables host users to access mail
from a mail server
110
Simple Mail Transfer
Protocol
SMTP
Provides for the exchange of
network management information
161,162
TELENET Protocol
Telnet
Provides remote terminal access
to a host.
23
a sequence of Echo messages to a host address. If the host is operational
it responds with a series of Echo Reply messages. Because the origination
of the Echo messages sets a timer when each message is transmitted, the
arrival of the response permits the round-trip delay to be computed. Thus,
Ping tells us if a distant host is operational as well as the round-trip delay
to that host. When installing a computer onto a TCP/IP network, it is quite
common to use Ping to insure your computer can reach the outside world
as well as be reached. Thus, Ping plays an important role as a diagnostic
testing tool.
Figure 5.33 illustrates an example of Ping options in a Microsoft Windows
environment as well as the use of the utility application. The top portion of
Figure 5.33 shows the various command options for the program, while the
lower portion illustrates the ‘‘pinging’’ of a Web server.
ARP
The Address Resolution Protocol (ARP) maps the high-level IP address configured via software to a low-level physical hardware address, typically the
NIC’s ROM address. The high-level IP address is 32 bits in length (IP version 4)
networking hardware and software
Figure 5.33
253
Using the Ping utility.
and is commonly represented by four decimal numbers, ranging from 0 to
255 per number, separated from one another by decimals. Thus, another term
used to reference an IP address is the dotted decimal address. The physical
hardware address represents the MAC address. Thus, ARP provides an IP
to MAC address resolution, which enables an IP packet to be transported in
a LAN frame to its appropriate MAC address. Later in this section we will
examine IP addresses in detail.
To illustrate the operation of ARP, consider Figure 5.34, which shows the
format of an Address Resolution Protocol (ARP) packet. The value of the
Hardware field is set to 1 to denote Ethernet. The second field, Protocol,
identifies the protocol address in use and is set to hex 0800 to identify the use
of IP addresses.
The Hardware Length (HLEN) and Protocol Length (PLEN) fields define the
length in bytes of the addresses to be used. In an IP-to-Ethernet environment
the hardware address will be six bytes in length while the protocol will be
four bytes in length. This corresponds to the four-byte IPv4 32-bit address and
the 48-bit or six-byte Ethernet MAC address. The operation field indicates an
ARP request (1) or ARP Reply (2).
When a layer 3 operating device, such as a router or gateway, receives an IP
packet for LAN delivery it must form a LAN frame. Thus, it must determine the
MAC address that corresponds to the IP destination address. To accomplish
this address resolution, the router transmits an ARP Request message as a
chapter five
254
0
15
Hardware type
Protocol type
HLEN
PLEN
Operation
Sender
hardware address
Sender
protocol address
Target
hardware address
Figure 5.34 The Address Resolution Protocol (ARP)
packet format. HLEN = Hardware Length; PLEN =
Protocol Length.
broadcast to all station on the LAN. Since the hardware address field value is
unknown, it is sent filled with zeros. The station that recognizes its IP address
inserts its hardware address into the message, changes the operation field
value to 2 and transmits the message using a destination frame address equal
to the source address that transported the ARP. When the router receives the
response, it places the hardware and IP addresses in memory, referred to as
an ARP cache, to expedite future datagram deliveries. It then forms a frame
with the appropriate hardware destination address to deliver the IP datagram.
TCP
The Transmission Control Protocol (TCP) represents a layer 4 connectionoriented reliable protocol. TCP provides a virtual circuit connection mode
service for applications that require connection setup, error detection, and
automatic retransmission. In addition, TCP is structured to support multiple
application programs on one host to communicate concurrently with processes
on other hosts, as well as for a host to demultiplex and service incoming traffic
among different applications or processes running on the host.
Each unit of data carried by TCP is referred to as a segment. Segments are
created by TCP subdividing the stream of data passed down by application
layer protocols that use its services, with each segment identified by the use
networking hardware and software
255
of a sequence number. This segment identification process enables a receiver,
if required, to reassemble data segments into their correct order.
Figure 5.35 illustrates the format of the TCP protocol header. To obtain an
appreciation for the functionality and capability of TCP, let’s examine the
fields in its header.
Source and Destination Port Fields
The source and destination ports are each 16 bits in length and identify a
process or service at the host receiver. The source port field entry is optional
and when not used is padded with zeros. Both source and destination port
values are commonly referred to as ‘‘well-known ports,’’ as they typically
identify an application layer protocol or process. Table 5.3 lists the wellknown port numbers associated with eight popular TCP/IP application layer
protocols. In examining the entries in the previously referenced table, note that
some protocols, such as FTP, use two port addresses or logical connections.
In the case of FTP, one address (21) is used for the transmission of commands,
responses, and functions as a control path. In comparison, the second port
address (20) is used for the actual file transfer.
2
Source port
2
Destination port
4
Sequence number
4
Acknowledgment
number
2
Data offset/control flags
2
Window
2
Checksum
2
Urgent pointer
Data
Figure 5.35
TCP protocol header.
256
chapter five
Sequence Fields
The sequence number is used to identify the data segment transported. The
acknowledgment number interpretation depends upon the setting of the ACK
control flag which is not directly shown in Figure 5.35. If the ACK control flag
bit position is set, the acknowledgment field will contain the next sequence
number the sender expects to receive. Otherwise the field is ignored.
Control Field Flags
There are six control field flags that are used to establish, maintain, and
terminate connections. Those flags include URG (urgent), SYN, ACK, RST
(reset), PSH (push), and FIN (finish).
Setting URG = 1 indicates to the receiver urgent data is arriving. The
SYN flag is set to 1 as a connection request and thus serves to establish
a connection. As previously discussed, the ACK flag, when set, indicates
that the acknowledgment flag is relevant. The RST flag, when set, means the
connection should be reset, while the PSH flag tells the receiver to immediately
deliver the data in the segment. Finally, the setting of the FIN flag indicates
the sender is done and the connection should be terminated.
The setting of the SYN bit in the TCP header is used to commence what is
referred to as the beginning of a three-way handshake. When a TCP session
commences the originator sets its SYN bit. The receiver will acknowledge the
TCP segment by setting the SYN and ACK bits in the responding TCP header.
The receiver will also set its Window field value, described next, to denote
the quantity of data it can receive. The originator will then respond by setting
the SYN bit in its header, completing the three-way handshake.
While the three-way handshake permits devices to negotiate the amount of
data to be transferred per segment, it also represents a security vulnerability.
That vulnerability occurs as a computer can only handle a certain amount of
open connections. Thus, one hacker technique that has received a degree of
trade press coverage is to flood a server with initial TCP SYN requests and not
respond to the returned SYN/ACK responses. This action creates numerous
open connections on the server and forms what is referred to as a denial of
service (DOS) attack.
Window Field
The window field is used to convey the number of bytes the sender can accept
and functions as a flow control mechanism. This 16-bit field indicates the
number of octets, beginning with the one in the acknowledgment field, that
the originator of the segment can control. Since TCP is a full-duplex protocol,
networking hardware and software
257
each host can use the window field to control the quantity of data that can
be sent to the computer. This enables the recipient to, in effect, control its
destiny. For example, if a receiving host becomes overloaded with processing
or another reason results in the inability of the device to receive large chunks
of data, it can use the window field as a flow control mechanism to reduce the
size of data chunks sent to it. At the end of our review of TCP header fields,
we will examine a TCP transmission sequence to note the interrelated role of
the sequence, acknowledgment, and window fields.
Checksum Field
The checksum provides error detection for the TCP header and data carried
in the segment. Thus, this field provides the mechanism for the detection of
errors in each segment.
Urgent Pointer Field
The urgent pointer field is used in conjunction with the URG flag as a
mechanism to identify the position of urgent data within a TCP segment.
When the URG flag is set the value in the urgent pointer field represents the
last byte of urgent data.
When an application uses TCP, TCP breaks the stream of data provided
by the application into segments and adds an appropriate TCP header. Next,
an IP header is prefixed to the TCP header to transport the segment via the
network layer. As data arrives at its destination network, it’s converted into
a data link layer transport mechanism. For example, on an Ethernet network
TCP data would be transported within Ethernet frames.
TCP Transmission Sequence Example
To illustrate the interrelationship between the sequence, acknowledgment,
and window fields, let’s examine the transmission of a sequence of TCP
segments between two hosts. Figure 5.36 illustrates via the use of a time chart
the transmission of a sequence of TCP segments.
At the top of Figure 5.36 it was assumed that a window size of 16 segments
is in use. Although TCP supports full-duplex transmission, for simplicity of
illustration we will use a half-duplex model in the time chart.
Assuming the host, whose address is ftp.fbi.gov, is transmitting a program or
performing a similar lengthy file transfer operation, the first series of segments
will have sequence numbers 64 through 79, assuming sequence number 63
was just acknowledged. The ACK value of 28 acknowledges that segments
258
chapter five
Host ftp . fbi . gov
Host ftp .opm. gov
DATA, S
T
I
M
E
EQ = 6 4
ACK =
DATA, S
72
, ACK =
OW =
, WIND
EQ = 8 0 ,
ACK =
88, WIN
D A T A, S
EQ = 8 8 ,
ACK =
Figure 5.36
97
16 units of data
received windows = 8
8
ACK = 2
8
D A T A, S
EQ = 8 8 ,
Timer
expires
retransmission
from SEQ = 79
Windows = 16
28
DOW =
8
ACK = 2
8
ACK = 2
8
OW =
, WIND
8
8 units of data
received
Still expecting
SEQ = 88
8 units of data
received
A TCP transmission sequence.
through number 27 were received by the host ftp.opm.gov and that it next
expects to receive segment 28.
Assuming segments 64 through 79 arrive error-free, the host with the
address ftp.opm.gov returns an ACK value of 80 to indicate the next segment
it expects to receive. At this point in time, let’s assume host ftp.opm.gov
is running out of buffer space and halves the window size to 8. Thus, host
ftp.opm.gov sets the window field value in the TCP header it transmits
to host ftp.fbi.gov to 8. Upon receipt of the TCP segment, host ftp.fbi.gov
reduces the number of segments it will transmit to 8 and uses an initial SEQ value of 80, increasing that value by 1 each time it transmits a
new segment until 8 new segments are transmitted. Assuming all 8 segments
were received error-free, host ftp.opm.gov then returns an ACK value of 88,
which acknowledges the receipt of segments with sequence field numbers
through 87.
Next, host ftp.fbi.gov transmits to host ftp.opm.gov another sequence of 8
segments using sequence field values of 88 to 95. However, let’s assume a
transmission impairment occurs that results in the segments being incorrectly
networking hardware and software
259
received or perhaps not even received at all at their intended destination.
If host ftp.opm.gov does not receive anything, it does not transmit anything back to host ftp.fbi.gov. Instead of waiting forever for a response, the
TCP/IP protocol stack includes an internal timer that clicks down to zero
while host ftp.fbi.gov waits for a response. When that value is reached,
the timer expires and the transmitting station retransmits its sequence of
8 segments. On the second time around, the sequence of 8 segments are
shown acknowledged at the bottom of Figure 5.36. If the impairment continued, the transmitting station would attempt a predefined number of
retransmissions after which it would terminate the session if no response
was received.
The altering of window field values provides a sliding window that can
be used to control the flow of information. That is, by adjusting the value of
the window field, a receiving host can inform a transmitting station whether
or not an adjustment in the number of segments transmitted is required. In
doing so there are two special window field values that can be used to further
control the flow of information. A window field value of 0 means a host has
shut down communications, while a window field value of 1 requires an
acknowledgment for each unit of data transmitted, limiting transmission to a
segment-by-segment basis.
UDP
The User Datagram Protocol (UDP) represents a second layer 4 transport
service supported by the TCP/IP protocol suite. UDP is a connectionless
service, which means that the higher-layer application is responsible for
the reliable delivery of the transported message. Figure 5.37 illustrates the
composition of the UDP header.
Octet
Field
1
2
3
4
5
6
7
8
Source port
Destination port
Datagram length
Data checksum
Data
Figure 5.37
The UDP header.
chapter five
260
Source and Destination Port Fields
The source and destination port fields are each 16 bits in length and, as
previously described for TCP, identify the port number of the sending and
receiving process, respectively. Here each port number process identifies an
application running at the corresponding IP address in the IP header prefixed
for the UDP header. The use of a port number provides a mechanism for
identifying network services as they denote communications points where
particular services can be accessed. For example, a value of 161 in a port field
is used in UDP to identify SNMP.
Length Fields
The length field indicates the length of the UDP packets in octets to include
the header and user data. The checksum, which is one’s complement arithmetic sum, is computed over a pseudoheader and the entire UDP packet. The
pseudoheader is created by the conceptual prefix of 12 octets to the header
previously illustrated in Figure 5.37. The first 8 octets are used by source and
destination IP addresses obtained from the IP packet. This is followed by a
zero-filled octet and an octet that identifies the protocol. The last 2 octets in
the pseudoheader denote the length of the UDP packet. By computing the
UDP checksum over the pseudoheader and user data, a degree of additional
data integrity is obtained.
IP
As previously mentioned, IP provides a datagram-oriented gateway service
for transmission between subnetworks. This provides a mechanism for hosts
to access other hosts on a best-effort basis but does not enhance reliability
as it relies on upper-layer protocols for error detection and correction. As a
layer 3 protocol, IP is responsible for the routing and delivery of datagrams.
To accomplish this task IP performs a number of communications functions
to include addressing, status information, management, fragmentation, and
reassembly of datagrams when necessary.
IP Header Format
Figure 5.38 illustrates the IP header format while Table 5.4 provides a brief
description of the fields in the IP header.
networking hardware and software
Version
261
Header length
Type of service
Total length
Identification
Flags
Fragment
offset
Time to live
Protocol
Checksum
Source address
Destination address
IP options
Figure 5.38
IP header format.
Version Field
The four-bit version field identifies the version of the IP protocol used to create
the datagram. The current version of the IP protocol is 4 and is encoded as
0100 in binary. The next-generation IP protocol is version 6, which is encoded
as 0110 in binary. In our discussion of IP we will focus on IPv4 in this section.
Header Length and Total Length Fields
The header length field follows the version field and is also 4 bits in length.
This field indicates the length of the header in 32-bit words. In comparison,
the total length field indicates the total length of the datagram to include its
header and higher-layer information. The use of 16 bits for the total length
field enables an IP datagram to be up to 216 or 65,535 octets in length.
chapter five
262
TABLE 5.4
IP Header Fields
Field
Description
Version
The version of the IP protocol used to create the datagram.
Header length
Header length in 32-bit words.
Type of service
Specifies how the datagram should be handled.
0
1
2 3 4 5
6
7
PRECEDENCE D T R UNUSED
PRECEDENCE indicates importance of the datagram
D When set requests low delay
T When set requests high throughput
R When set requests high reliability
Total length
Specifies the total length to include header and data.
Identification
Used with source address to identify fragments belonging to
specific datagrams.
Flags
Middle bit when set disables possible fragmentation.
Low-order bit specifies whether the fragment contains data
from the middle of the original datagram or the end.
Fragment offset
Specifies the offset in the original datagram of data being
carried in a fragment.
Time to live
Specifies the time in seconds a datagram is allowed to
remain in the internet.
Protocol
Specifies the higher-level protocol used to create the
message carried in the data field.
Header checksum
Protects the integrity of the header.
Source IP address
The 32-bit IP address of the datagram’s sender.
Destination IP address
The 32-bit IP address of the datagram’s intended recipient.
IP options
Primarily used for network testing or debugging.
0
1
2
3 4
5
6
7
COPY OPTION CLASS OPTION NUMBER
When copy bit set it tells gateways that the option should be
copied into all fragments.
networking hardware and software
TABLE 5.4
263
(Continued)
Field
Description
When set to 0 the option is copied into the first fragment.
Option Class
Meaning
0
1
2
3
Datagram or network control
Reserved for future use
Debugging
Reserved for future use
The option number defines a specific option within a class.
Type of Service Field
The type of service field identifies how the datagram is handled. Three of the
eight bits in this field are used to denote the precedence or level of importance
assigned by the originator. Thus, this field provides a priority mechanism for
routing IP datagrams.
Identification and Fragment Offset Fields
To obtain an appreciation for the role of the fragment offset field a brief discussion of IP fragmentation is in order. When an IP protocol stack transmits
a datagram, its size is limited by what is referred to as the Maximum Transmission Unit (MTU). The MTU is usually set when a device driver initializes
an interface and the MTU represents the payload portion of a datagram (its
length less header length).
Most TCP/IP protocol stacks support MTUs up to 64K − 1 bytes, or 65,535
bytes. The MTU supported by a protocol stack for an interface is also known
as an interface MTU. Another MTU worth noting is a route MTU, which
represents the MTU that can be used from source to destination without
requiring fragmentation. The route MTU is usually maintained as a field
value in a host’s routing table and set either manually or via an MTU
discovery process.
When a route MTU is smaller than the length of a datagram to be transferred, the routing entity will either fragment the datagram or drop it. All
implementations of IP must support IP fragmentation based upon RFC 791.
If the DON’T− FRAGMENT bit is set in the header, then fragmentation is not
allowed and the datagram is dropped.
264
chapter five
IP fragmentation places the burden of the effort on the receiving station
and the routing entity. When a station receives an IP fragment it must fully
reassemble the complete IP datagram prior to being able to extract the TCP
segment, resulting in a requirement for additional buffer memory and CPU
processing power at the receiver. In addition, if any fragments are dropped,
the original datagram must be resent. Due to this, most vendors set the
DON’T− FRAGMENT bit in the header.
Setting the DON’T− FRAGMENT bit causes an oversized IP datagram to be
dropped, resulting in an ICMP DESTINATION UNREACHABLE-FRAGMENTATION NEEDED message to be sent to the sender. This will result in the
MTU discover algorithm selecting a smaller MTU for the path and using that
MTU for subsequent transmissions.
The identification field enables each datagram or fragmented datagram to be
identified. If a datagram was previously fragmented, the fragment offset field
specifies the offset in the original datagram of the data being carried. In effect,
this field indicates where the fragment belongs in the complete message. The
actual value in this field is an integer that corresponds to a unit of eight octets,
providing an offset in 64-bit units.
Time to Live Field
The time to live (TTL) field specifies the maximum time that a datagram
can live. Because an exact time is difficult to measure, almost all routers
decrement this field by one as a datagram flows between networks, with the
datagram being discarded when the field value reaches zero. Thus, this field
more accurately represents a hop count field. You can consider this field to
represent a fail-safe mechanism, as it prevents misaddressed datagrams from
continuously flowing on the Internet.
The time-to-live (TTL) field is the key to the operation of a utility program
included in the TCP/IP protocol suite that provides you with the ability to
trace a route through an IP network. That utility program is traceroute, which
in a Microsoft Windows environment is called tracert.
When you use traceroute the program sets a timer and then sets a TTL
value of 1 in the IP datagram prior to transmitting it to the intended recipient.
The first router in the path decrements the value of the TTL field by 1 and,
since it is now 0, throws it into the great bit bucket in the sky and returns a
destination unreachable ICMP message. This action provides the round-trip
delay and address of the first router on the path to the destination. The
program then increments the value of the TTL field to 2 and transmits another
datagram to locate the second router in the path to the destination, and so on
until the destination is reached.
networking hardware and software
265
Flags Field
The flags field contains two bits that indicate how fragmentation occurs while
a third bit is currently unassigned. The setting of one bit can be viewed as a
direct fragment control mechanism as a value of zero indicates the datagram
can be fragmented, while a value of one denotes don’t fragment. The second bit
is set to zero to indicate that a fragment in a datagram is the last fragment and
set to a value of one to indicate more fragments follow the current protocol.
Protocol Field
The protocol field specifies the higher-level protocol used to create the message
carried in the datagram. For example, a value of decimal 6 would indicate
TCP, while a value of decimal 17 would indicate UDP.
Source and Destination Address Fields
The source and destination address fields are both 32 bits in length. As
previously discussed, each address represents both a network and a host
computer on the network.
In examining the IP header a common network problem relates to the
IP address carried in the source and destination address fields. Thus, a
description of IP addressing is warranted as it forms the basis for network
addressing as well as the domain name service translation of English-type
mnemonics, representing computers or host names, into what is known as
dotted decimal IP addresses.
IP Addressing
The IP addressing scheme uses a 32-bit address, which is divided into an
assigned network number and a host number. The latter can be further
segmented into a subnet number and a host number. Through subnetting you
can construct multiple networks while localizing the traffic of hosts to specific
subnets, a technique I will shortly illustrate.
IP addressing numbers are assigned by the InterNIC network information
center and can fall into one of five unique network classes, referenced as
Classes A through E. Figure 5.39 illustrates the IP address formats for Class
A, B, and C networks. Class D addresses are reserved for multicast groups,
while Class E addresses are reserved for future use and are considered as
experimental.
In examining Figure 5.39, note that by examining the first bit in the IP
address you can distinguish a Class A address from Class B and C addresses.
266
chapter five
Class A
Octet 1
Octet 2
Octet 3
Octet 4
0
Network
identifier
Host identifier
Class B
Octet 1
Octet 2
Octet 3
Octet 4
1 0
Network
identifier
Host identifier
Class C
Octet 1
Octet 2
Octet 3
Octet 4
1 1 0
Network identifier
Figure 5.39
Host
identifier
IP Classfull address formats.
Thereafter, examining the composition of the second bit position enables a
Class B address to be distinguished from a Class C address.
An IP 32-bit address is expressed as four decimal numbers, with each
number ranging in value from 0 to 255 and separated from another number by
a dot (decimal point). This explains why an IP address is commonly referred
to as a dotted decimal address.
To facilitate determining the decimal value of an eight-bit binary number
you can use the following positional bit values for a byte:
128 64 32 16 8 4 2 1
Based upon the above, a byte with the bit composition 10000011 would have
the decimal value 128 + 2 + 1, or 131 decimal. Similarly, a byte with the bit
composition 01010111 would have the decimal value 64 + 16 + 4 + 2 + 1, or
87 decimal.
networking hardware and software
267
Class A In examining Figure 5.39, note that a Class A address has three
bytes available for identifying hosts on one network or on subnets, which
provide support for more hosts than other address classes. Thus, Class A
addresses are only assigned to large organizations or countries. Since the first
bit in a Class A address must be 0, the first byte ranges in value from 1 to
127 instead of to 255. Through the use of 7 bits for the network portion and
24 bits for the host portion of the address, 128 networks can be defined with
approximately 16.78 million hosts capable of being addressed on each Class
A network. In actuality, network 127.0.0.0 is not used. This is because any IP
address commencing with 127 with a host value of 1 or greater represents a
loopback address. Thus, you could ping 127.0.0.1 to test the operation of a
TCP/IP protocol stack on your computer.
Class B A Class B address uses two bytes for the network identifier and two
for the host or subnet identifier. This permits up to 65,636 hosts and/or subnets
to be assigned; however, since the first 2 bits of the network portion of the
address are used to identify a Class B address, the network portion is reduced
to a width of 14 bits. Thus, up to 16,384 Class B networks can be assigned.
Due to the manner by which Class B network addresses are subdivided into
network and host portions, such addresses are normally assigned to relatively
large organizations with tens of thousands of employees.
Class C In a Class C address three octets are used to identify the network,
leaving one octet to identify hosts and/or subnets. The use of 21 bits for
a network address enables approximately two million distinct networks to
be supported by the Class C address class. Because one octet only permits
256 hosts or subnets to be identified, many small organizations with a requirement to provide more than 256 hosts with access to the Internet must obtain
multiple Class C addresses.
Host Restrictions In actuality the host portion of an IP address has two
restrictions, which reduces the number of hosts that can be assigned to a
network. First, the host portion cannot be set to all-zero bits, as an all-zeros
host number is used to identify a base network. Secondly, an all-ones host
number represents the broadcast address for a network or subnetwork. Thus,
the maximum number of hosts on a network must be reduced by two. For a
Class C network, a maximum of 254 hosts can then be configured for operation.
Subnetting Through the use of subnetting you can use a single IP address
as a mechanism for connecting multiple physical networks. This action
268
chapter five
extends the use of scarce IP network addresses as well as reduces the number
of routing table entries required. To accomplish subnetting you logically
divide the host portion of an IP address into a network address and a
host address.
Figure 5.40 illustrates an example of the IP subnet addressing format for a
Class B address. In this example, all traffic routed to the address XY, where
X and Y represent the value of the first two Class B address octets, flows to
a common location connected to the Internet, typically a router. The router
in turn connects two or more Class B subnets, each with a distinct address
formed by the third decimal digit, which represents the subnet identifier.
Figure 5.41 illustrates a Class B network address location with two physical
networks using subnet addressing.
Subnet Masks The implementation of a subnet addressing scheme is accomplished by the partitioning of the host identifier portion of an IP address. To
accomplish this a 32-bit subnet mask must be created for each network,
with bits set to 1 in the subnet mask to indicate the network portion of the
IP address, while bits are set to 0 to indicate the host identifier portion.
Thus, the Class B subnet address format illustrated in the lower portion of
Figure 5.40 would require the following 32-bit subnet mask if there were no
subnets:
11111111 11111111 00000000 00000000
Class B address format
Octet 1
Octet 2
Octet 3
Octet 4
1 0
Network
identifier
Host identifier
Class B subnet address format
Octet 1
Octet 2
Octet 3
Octet 4
1 0
Network identifier
Figure 5.40
Subnet
identifier
Class B subnetting.
Host
identifier
networking hardware and software
269
Network X . Y .1.0
Router
Internet
Access
via X. Y. Z. Z
X . Y .1.1
X . Y .1. 2
Network X . Y . 2.0
X . Y . 2.1
X . Y . 2. 2
Figure 5.41 A Class B network address location with two physical networks
using subnet addressing. IP datagrams with the destination address X.Y.Z.Z,
where Z can be any decimal value representing a Class B network address that
can consist of 256 subnets, with 256 hosts on each subnet.
The prior mask would then be entered as 255.255.0.0 in dotted decimal representation into a router configuration screen as well as in software
configuration screens on TCP/IP program stacks operating on each subnet.
Concerning the latter, you must then configure each station to indicate its
subnet and host identifier so that each station obtains a full 4-digit dotted
decimal address.
Because a Class B network address by default uses two bytes for the network
address and two for the host the prior mask does not allow for any subnets. To
allow for subnets we must internally (to our organization) extend the subnet
mask into the host portion of the address. Each one-bit extension permits 2n
subnets to be defined. For example, consider the following subnet:
11111111 11111111 11100000 00000000
This subnet is 255.255.224 in dotted decimal and would support up to
eight (23 ) subnets. Note that internally to our organization the number of bit
positions for hosts on a subnet becomes 16 − 3, or 13 positions. Thus, each
subnet can have a maximum of 213 − 2 hosts as each subnet cannot be all 0’s
or all 1’s, similar to regular network restrictions.
Domain Name Service
Addressing on a TCP/IP network occurs through the use of four decimal numbers ranging from 0 to 255, which are separated from one another by a dot.
270
chapter five
This dotted decimal notation represents a 32-bit address, which consists of an
assigned network number and a host number as previously described during
our examination of IP addressing. Because numeric addresses are difficult to
work with, TCP/IP also supports a naming convention based upon English
words or mnemonics that are both easier to work with and remember. The
translation of English words or mnemonics to 32-bit IP addresses is performed
by a domain name server. Each network normally has at least one domain
name server, and the communications established between such servers on
TCP/IP networks connected to the Internet are referred to as a domain name
service (DNS).
The DNS is the naming protocol used in the TCP/IP protocol suite, which
enables IP routing to occur indirectly through the use of names instead of IP
addresses. To accomplish this, DNS provides a domain name to IP address
translation service.
A domain is a subdivision of a wide area network. When applied to
the Internet where the capitalized I references the collection of networks
interconnected to one another, there are six top-level and seven recently
added domain names, which were specified by the Internet Corporation for
Assigned Names and Numbers (ICANN) at the time this book was prepared.
Those top-level domains are listed in Table 5.5.
Under each top-level domain the InterNIC will register subdomains, which
are assigned an IP network address. An organization receiving an IP network
address can further subdivide their domain into two or more subdomains. In
addition, instead of using dotted decimal notation to describe the location of
each host, they can assign names to hosts as long as they follow certain rules
and install a name server that provides IP address translation between named
hosts and their IP addresses.
To illustrate the operation of a name server, consider the network domain
illustrated in Figure 5.42. In this example we will assume that a well-known
government agency has a local area network with several computers that
will be connected to the Internet. Each host address will contain the specific
name of the host plus the names of all of the subdomains and domains
to which it belongs. Thus, the computer warrants would have the official address:
telnet.warrants.fbi.gov
Similarly, the computer cops would have the address:
ftp.cops.fbi.gov
networking hardware and software
TABLE 5.5
271
Internet Top-Level Domain Names
Domain Name
.COM
.EDU
.GOV
.MIL
.NET
.ORG
Recently added
.aevo
.biz
.coop
.info
.museum
.name
.pro
Assignment
Commercial organization
Educational organization
Government agency
Department of Defense
Networking organization
Not-for-profit organization
Air-transport industry
Businesses
Cooperatives
Information/services (unrestricted
use)
Museums
For registration by individuals
Accountants, lawyers, physicians,
and other professionals
FBI . GOV
Telnet
ftp
Warrants
Cops
Figure 5.42
Robbers
A domain-naming hierarchy.
In examining the domain-naming structure illustrated in Figure 5.39 note
that computers were placed in subdomains using common Internet application
names, such as telnet and ftp. This is a common technique that organizations
use to make it easier for network users within and outside of the organization
to remember mnemonics that represent specific hosts on their network.
272
chapter five
Although domain names provide a mechanism for identifying objects connected to wide area networks, hosts in a domain require network addresses
to transfer information. Thus, another host functioning as a name server is
required to provide a name-to-address translation service.
Name Server
The name server plays an important role in TCP/IP networks. In addition to
providing a name-to-IP address translation service, it must recognize that an
address is outside its administrative zone of authority. For example, assume
a host located on the domain illustrated in Figure 5.42 will use the address
fred.microwear.com to transmit a message. The name server must recognize
that that address does not reside in the current domain and must forward
the address to another name server for translation into an appropriate IP
address. Because most domains are connected to the Internet via an Internet
service provider, the name server on the domain illustrated in Figure 5.42
would have a pointer to the name server of the Internet service provider
(ISP) and forward the query to that name server. The ISP’s name server will
either have an entry in its table-in-cache memory or forward the query to
another higher-level name server. Eventually, a name server will be reached
that has administrative authority over the domain containing the host name to
resolve and will return an IP address through a reversed hierarchy to provide
the originating name server with a response to its query. Most name servers
cache the results of previous name queries, which can considerably reduce
off-domain or Internet DNS queries. In the event a response is not received,
possibly due to an incorrect name or the entry of a name no longer used, the
local name server will generate a ‘‘failure to resolve’’ message after a period
of time that will be displayed on the requesting host’s display.
TCP/IP Configuration
The configuration of a station on a TCP/IP network normally requires the
specification of four IP addresses as well as the station’s host and domain
names. To illustrate the configuration of a TCP/IP station, Figures 5.43 through
5.45 show the screen settings on a Microsoft Windows NT server used to
configure the station as a participant on a TCP/IP network.
Figure 5.44 illustrates the Windows NT Network Settings dialog box with
the TCP/IP protocol selected in the installed network software box. Note that
at the top of that box the entry NWLink IPX/SPX Compatible Transport is
shown. Windows NT has the ability to operate multiple protocol stacks to
networking hardware and software
273
Figure 5.43 Using the Windows NT Network Settings dialog box to configure
the use of the TCP/IP protocol.
Figure 5.44 The Windows NT TCP/IP configuration dialog box with entries
for the IP address of the network interface, subnet mask, and default gateway.
274
chapter five
Figure 5.45 Using the Windows NT DNS Configuration dialog box to specify
the station’s name and domain name as well as two name server addresses.
include NWLink, which is Microsoft’s implementation of the Novell IPX and
SPX protocols. Also note that we previously configured the server to use a
3 Com Etherlink III Adapter as indicated in the box labeled Installed Adapter
Cards in the lower left portion of the screen display.
Clicking on the button labeled Configure in Figure 5.44 results in the display
of another dialog box, this one labeled TCP/IP Configuration. Figure 5.45
illustrates the TCP/IP Configuration dialog box with three address entries
shown. Those address entries indicate the IP address of the interface assigned
to the selected adapter card, the subnet mask, and the IP address of the
default gateway. Note that a computer can have multiple adapter cards, thus
IP addresses are actually assigned to network interfaces. Also note that the
term gateway dates to when such devices routed packets to other networks
if their address was not on the local network. Thus, a more modern term for
gateway is router.
After configuring the entries shown in Figure 5.44, you will require a few
more entries. Those entries include the address of the name server used to
translate near-English mnemonics into IP addresses as well as the name of
your computer and domain. To configure the address of the name server, you
networking hardware and software
275
would first click on the button labeled DNS in Figure 5.44. This action will
result in the display of a dialog box labeled DNS Configuration, which is
shown in Figure 5.45.
The Windows NT DNS Configuration dialog box enables you to specify
your host computer name and your domain name. Those entries are optional;
however, if you do not include those entries and your local DNS uses this
configuration information, other TCP/IP users either on your network or
a distant network will not be able to access your computer by entering
your computer’s near-English address name, which in Figure 5.45 would be
www.xyz.com. Instead, users would have to know your numeric IP address.
The DNS entry area in Figure 5.45 allows you to specify up to three name
servers in the order they should be searched. Many organizations operate two
name servers, so the ability to enter three should suffice for most organizations.
Operating Multiple Stacks
In concluding this chapter we will use three Windows NT screens to expand
upon illustrating the flexibility you can obtain from operating multiple
Figure 5.46 Using the Windows NT Connect Network Drive dialog box to
view Windows and NetWare network devices.
276
chapter five
protocol stacks. In the example we will shortly view, we will use Microsoft’s
Windows NT to show how to access both Windows NT and NetWare networks
as well as operate a Netscape browser to run a TCP/IP application.
Figure 5.46 illustrates the use of the Windows NT Connect Network Drive
dialog box to view both Microsoft Windows and Novell NetWare network
devices that can operate on a common network or network infrastructure. By moving a highlight bar over a particular entry and clicking on
the entry, you obtain the ability to log into NT or NetWare servers or
access-shared directories on NT devices. For those readers from Missouri,
Figure 5.47 illustrates the execution of a NetWare server program viewed
on a Windows NT workstation. In this example, the NT workstation is running Microsoft’s NWLink IPX/SPX compatible protocol, which enables it
to communicate in a client/server environment as a participant on a NetWare LAN.
Figure 5.47 Viewing the execution of a NetWare server program on a Windows NT workstation.
networking hardware and software
277
Figure 5.48 Using the Microsoft Internet Explorer browser to access the
home page of the author’s publisher.
For our last stack example, since we previously configured TCP/IP we
should be able to operate a TCP/IP application. Figure 5.48 illustrates the use
of the Microsoft Internet Explorer browser to access the home page of this
author’s publisher. Thus, this short section illustrates the worth of operating
multiple protocol stacks to access NT and NetWare servers as well as a
Web server.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter six
Bridging and Switching
Methods and Performance
Issues
In Chapter 5, an overview of bridge operations was presented, along with
information concerning the functionality of other local area network hardware and software components. That chapter deferred until now a detailed
examination of bridging methods, to include their network use and performance issues. In this chapter, we will focus our attention on those issues,
examining different methods that bridges use for routing frames, performance
issues that govern their ability to examine and forward frames without introducing network bottlenecks, and their typical employment for interconnecting
LANs. Because LAN switches represent a special type of multiport bridge, we
will also focus our attention upon this topic later in this chapter. Thus, once
we have an appreciation for the operation and utilization of bridges, we will
turn our attention to LAN switches.
6.1 Bridging Methods
Bridges operate by examining MAC layer addresses, using the destination
and source addresses within a frame as a decision criterion to make their
forwarding decisions. Operating at the MAC layer, bridges are not addressed,
and must therefore examine all frames that flow on a network. Because bridges
operate at the MAC layer, they in effect terminate a collision domain. That is,
if a collision is detected upon one port of a bridge, it is not propagated onto
any output port. This means that, unlike a repeater, a bridge can be used to
extend the span of a LAN.
279
280
chapter six
Address Issues
Since bridges connect networks, it is important to ensure that duplicate MAC
addresses do not occur on joined internal networks — a topology we will refer
to as an intranet. While duplicate addresses will not occur when universally
administered addressing is used, when locally administered addressing is
used duplicate addresses become possible. Thus, the addresses assigned to
stations on separate networks joined to form an intranet should be reviewed
before using bridges to connect two or more separate networks.
Two primary routing methods are used by bridges for connecting wired local
area networks: transparent or self-learning and source routing. Transparent
bridges were originally developed to support the connection of Ethernet
networks, as briefly described in Chapter 5.
Transparent Bridging
A transparent bridge examines MAC frames to learn the addresses of stations
on the network, storing information in internal memory in the form of an
address table. Thus, this type of bridge is also known as a self-learning
bridge. To understand the operation of a transparent bridge in more detail
and realize some of the limitations associated with the use of this device,
consider the simple intranet illustrated in Figure 6.1. This intranet consists
C
A
Ethernet 3
Ethernet 2
Ethernet 1
E
D
F
B
Port 1
Bridge 1
Bridge 1 port / address table
Port 1
A
B
Port 2
C
D
E
F
Port 2
Port 1
Bridge 2
Port 2
Bridge 2 port / address table
Port 1
A
B
C
D
Port 2
E
F
Figure 6.1 Transparent bridge operation. A transparent or self-learning
bridge examines the source and destination addresses to form port/address or
routing tables in memory.
bridging and switching methods and performance issues
281
of three Ethernet local area network segments connected through the use of
two self-learning bridges. For simplicity of illustration, only two workstations
are shown and labeled on each local area network. Those labels represent the
48-bit MAC address of each station.
Port/Address Table Construction
As previously noted in Chapter 5, a bridge constructs its port/address table
by using what can be referred to as the ‘‘three F’s’’ — flooding, filtering, and
forwarding. If a bridge encounters a frame with a destination address that is
not in its port/address table, it transmits the frame onto all other ports except
the port it was received on. If the destination address is in its port/address
table and does not represent the port the frame was received on, the bridge
forwards the frame onto the port corresponding to the entry in the table for
the destination address. If the destination address is in the port/address table
and represents the port the frame was received on, there is no need to forward
the frame. Thus, the frame is filtered by the bridge.
In examining the construction of bridge port/address tables for the network
shown in Figure 6.1, we will assume that each bridge operates as a transparent
bridge. As frames flow on the Ethernet, bridge 1 examines the source address
of each frame. Eventually, after both stations A and B have become active, the
bridge associates their address as being on port 1 of that device. Any frames
with a destination address other than stations A or B are considered to be on
another network. Thus, bridge 1 would eventually associate addresses C, D,
E, and F with port 2, once it receives frames with those addresses in their
destination address fields. Similarly, bridge 2 constructs its own port/address
table. Since frames from Ethernet 1 and 1 Ethernet 2 can have source addresses
of A, B, C, or D, eventually the port/address table of bridge 2 associates those
addresses with port 1 of that device. Since frames from Ethernet 1 or Ethernet 2
with a destination address of E or F are not on those local area networks,
bridge 2 then associates those addresses with port 2 of that device.
The port/address tables previously shown in Figure 6.1 are normally stored
in bridge memory sorted by MAC address. In addition, the time the entry
occurred is also added to the table, resulting in a three-column table. The
time of occurrence is used by bridges to periodically purge old entries. Entry
purging is important because inactive entries both use finite memory and
extend the search time associated with the reading of each frame received
on a bridge port and its comparison to entries in the port/address table. This
searching is required to determine if the frame is to be forwarded along with
the port onto which the frame should be placed.
282
chapter six
Advantages
One of the key advantages of a transparent bridge is that it operates independently of the contents of the information field and is protocol-independent.
Because this type of bridge is self-learning, it requires no manual configuration
and is essentially a ‘‘plug and play’’ device. Thus, this type of bridge is attractive for connecting a few local area networks together, and is usually sufficient
for most small and medium-sized businesses. Unfortunately, its use limits the
development of certain interconnection topologies, as we will soon see.
Disadvantages
To see the disadvantages associated with transparent bridges, consider
Figure 6.2, in which the three Ethernet local area networks are interconnected
through the use of three bridges. In this example, the interconnected networks
form a circular or loop topology. Because a transparent bridge views stations
as being connected to either port 1 or port 2, a circular or loop topology will
create problems. Those problems can result in an unnecessary duplication
of frames, which not only degrades the overall level of performance of the
Ethernet 1
Ethernet 2
B
A
E
Port 1
Bridge 1
F
Port 2
Port 1
Port 1
Bridge 2
Bridge 3
Port 2
Port 2
Ethernet 3
C
D
Figure 6.2 Transparent bridges do not support network loops. The construction of a circular or loop topology with transparent bridges can result in an
unnecessary duplication of frames, and may confuse end stations. To avoid
these problems, the Spanning Tree Protocol (STP) opens a loop by placing
one bridge in a standby mode of operation.
bridging and switching methods and performance issues
283
interconnected networks, but will quite possibly confuse the end stations. For
example, consider a frame whose source address is A and whose destination
address is F. Both bridge 1 and bridge 2 will forward the frame. Although
bridge 1 will forward the frame to its appropriate network using the most
direct route, the frame will also be forwarded via bridge 2 and bridge 3
to Ethernet 2, resulting in a duplicate frame arriving at workstation F. At
station F, a mechanism would be required to reject duplicate frames. Even if
such a mechanism is available, the additional traffic flowing across multiple
internet paths would result in an increase in network usage. This, in turn,
would saturate some networks, while significantly reducing the level of
performance of other networks. For these reasons, transparent bridging is
prohibited from creating a loop or circular topology. However, transparent
bridging supports concurrently active multiple bridges, using an algorithm
known as the spanning tree to determine which bridges should forward and
which bridges should only filter frames.
Spanning Tree Protocol
The problem of active loops was addressed by the IEEE Committee 802 in
the 802.1D standard with an intelligent algorithm known as the Spanning
Tree Protocol (STP). The STP, based on graph theory, converts a loop into a
tree topology by disabling a link. This action ensures there is a unique path
from any node in an intranet to every other node. Disabled nodes are then
kept in a standby mode of operation until a network failure occurs. At that
time, the STP will attempt to construct a new tree using any of the previously
disabled links.
Operation
To illustrate the operation of the STP, we must first become familiar with
the difference between the physical and active topology of bridged networks.
In addition, there are a number of terms associated with the spanning tree
algorithm, as defined by the protocol, that we should become familiar with.
Thus, we will also review those terms before discussing the operation of
the algorithm.
Physical versus Active Topology
In transparent bridging, a distinction is made between the physical and active
topology resulting from bridged local area networks. This distinction enables
the construction of a network topology in which inactive but physically
284
chapter six
constructed routes can be placed into operation if a primary route should fail,
and in which the inactive and active routes would form an illegal circular
path violating the spanning tree algorithm if both routes were active at the
same time.
The top of Figure 6.3 illustrates one possible physical topology of bridged
networks. The cost (C) assigned to each bridge will be discussed later in this
chapter. The lower portion of Figure 6.3 illustrates a possible active topology
for the physical configuration shown at the top of that illustration.
When a bridge is used to construct an active path, it will forward frames
through those ports used to form active paths. The ports through which
frames are forwarded are said to be in a forwarding state of operation. Ports
that cannot forward frames because their operation forms a loop are said to be
in a blocking state of operation.
Under the spanning tree algorithm, a port in a blocking state can be placed
into a forwarding state to provide a path that becomes part of the active
network topology. This new path usually occurs because of the failure of
another path, bridge component, or the reconfiguration of interconnected
networks, and must not form a closed loop.
Spanning Tree Algorithm
The basis for the spanning tree algorithm is a tree structure, since a tree forms
a pattern of connections that has no loops. The term spanning is used because
the branches of a tree structure span or connect subnetworks.
As a review for readers unfamiliar with graph theory, let’s examine the
concept behind spanning trees. To appropriately do so we need a point
of reference, so let’s begin with the graph structure shown at the top of
Figure 6.4. A spanning tree of a graph is a subgraph that connects all nodes and
represents a tree. The graph shown at the top of Figure 6.4 has eight distinct
spanning trees. The lower portion of Figure 6.4 illustrates the spanning trees
associated with the graph structure illustrated at the top of the previously
referenced figure.
Minimum Spanning Tree
Suppose the links connecting each node are assigned a length or weight. Then,
the weight of a tree represents the sum of its links or edges. If the weight
or length of the links or tree edges differ, then different tree structures will
have different weights. Thus, the identification of the minimum spanning tree
requires us to examine each of the spanning trees supported by a graph and
identify the structure that has the minimum length or weight.
bridging and switching methods and performance issues
285
Token-ring 1
B1
C =10
Ethernet 2
Ethernet 1
B2
C =10
B4
C =15
B3
C =15
Token-ring 2
B5
C =10
Ethernet 3
(a) Physical topology
Token-ring 1
B2
B1
B3
Ethernet 1
Token-ring 2
Legend: B = Bridge
C = Cost
Ethernet 2
B4
B5
Ethernet 3
(b) Active topology
Figure 6.3 Physical versus active topology. When transparent bridges are
used, the active topology cannot form a closed loop in the intranet.
286
chapter six
(a) Network graph
(b) Possible spanning trees
Figure 6.4
Forming spanning trees from a network graph.
The identification of the minimum spanning tree can be accomplished
by listing all spanning trees and finding the minimum weight or length
associated with the list. This is a brute force method that always works but
is not exactly efficient, especially when a graph becomes complex and can
contain a significant number of trees. A far better method is obtained by the
use of an appropriate algorithm.
Kruskal’s Algorithm
There are several popular algorithms developed for solving the minimum
spanning tree of a graph. One of those algorithms is the Kruskal algorithm
which is relatively easy to understand and will be used to illustrate the
computation of a minimum spanning tree. Because we need weights or
lengths assigned to each edge or link in a graph, let’s revise the network graph
previously shown in Figure 6.4 and add some weights. Figure 6.5 illustrates
the weighted graph.
bridging and switching methods and performance issues
8
B
A
1
3
4
6
C
287
D
Figure 6.5
A weighted network graph.
Kruskal’s algorithm can be expressed as follows:
1. Sort the edges of the graph (G) in their increasing order by weight
or length.
2. Construct a subgraph (S) of G and initially set it to the empty state.
3. For each edge (e) in sorted order:
If the endpoints of the edges (e) are disconnected in S, add them to S.
Using the graph shown in Figure 6.5, let’s apply Kruskal’s algorithm
as follows:
1. The sorted edges of the graph in their increasing order by weight or
length produces the following table:
Edge
Weight/Length
A-C
1
B-D
3
C-B
4
C-D
6
A-B
8
2. Set the subgraph of G to the empty state. Thus, S = null.
3. For each edge add to S as long as the endpoints are disconnected. Thus,
the first operation produces:
A
S = A,C
or
C
288
chapter six
The next operation produces:
S = (A,C) + (B,D)
A
B
C
D
or
The third operation produces:
S = (A,B) + (B,D) + (C,B)
A
B
C
D
or
Note that we cannot continue as the endpoints in S are now all connected.
Thus, the minimum spanning tree consists of the edges or links (A, B) +
(B, D) + (C, B) and has the weight 1 + 4 + 3, or 7. Now that we have an
appreciation for the method by which a minimum spanning tree is formed, let’s
turn our attention to its applicability in transparent bridge-based networks.
Similar to the root of a tree, one bridge in a spanning tree network will
be assigned to a unique position in the network. Known as the root bridge,
this bridge is assigned as the top of the spanning tree, and because of this
position, it has the potential to carry the largest amount of intranet traffic due
to its position.
Because bridges and bridge ports can be active or inactive, a mechanism
is required to identify bridges and bridge ports. Each bridge in a spanning
tree network is assigned a unique bridge identifier. This identifier is the MAC
address on the bridge’s lowest port number and a two-byte bridge priority
level. The priority level is defined when a bridge is installed and functions
as a bridge number. Similar to the bridge priority level, each adapter on
a bridge that functions as a port has a two-byte port identifier. Thus, the
unique bridge identifier and port identifier enable each port on a bridge to be
uniquely identified.
Path Cost
Under the spanning tree algorithm, the difference in physical routes between
bridges is recognized, and a mechanism is provided to indicate the preference
for one route over another. That mechanism is accomplished by the ability
bridging and switching methods and performance issues
289
to assign a path cost to each path. Thus, you could assign a low cost to a
preferred route and a high cost to a route you only want to be used in a
backup situation.
Once path costs are assigned to each path in an intranet, each bridge will
have one or more costs associated with different paths to the root bridge. One
of those costs is lower than all other path costs. That cost is known as the
bridge’s root path cost, and the port used to provide the least path cost toward
the root bridge is known as the root port.
Designated Bridge
As previously discussed, the spanning tree algorithm does not permit active
loops in an interconnected network. To prevent this situation from occurring,
only one bridge linking two networks can be in a forwarding state at any
particular time. That bridge is known as the designated bridge, while all other
bridges linking two networks will not forward frames and will be in a blocking
state of operation.
Constructing the Spanning Tree
The spanning tree algorithm employs a three-step process to develop an active
topology. First, the root bridge is identified. To accomplish this, each bridge
in the intranet will initially assume it is the root bridge. To determine which
bridge should actually act as the root bridge, each bridge will periodically
transmit bridge protocol data unit (BPDU) frames that are described in the
following section. BPDU frames under Ethernet version 2 are referred to as
HELLO frames or messages and are transmitted on all bridge ports. Each
BPDU frame includes the priority of the bridge defined at installation time. As
the bridges in the intranet periodically transmit their BPDU frames, bridges
receiving a BPDU with a lower priority value than its own cease transmitting
their BPDUs; however, they forward BPDUs with a lower priority value.
Thus, after a short period of time the bridge with the lowest priority value
is recognized as the root bridge. In Figure 6.3b we will assume bridge 1 was
selected as the root bridge. Next, the path cost from each bridge to the root
bridge is determined, and the minimum cost from each bridge becomes the
root path cost. The port in the direction of the least path cost to the root
bridge, known as the root port, is then determined for each bridge. If the root
path cost is the same for two or more bridges linking LANs, then the bridge
with the highest priority will be selected to furnish the minimum path cost.
Once the paths are selected, the designated ports are activated.
290
chapter six
In examining Figure 6.3a, let us now use the cost entries assigned to
each bridge. Let us assume that bridge 1 was selected as the root bridge,
since we expect a large amount of traffic to flow between Token-Ring 1 and
Ethernet 1 networks. Therefore, bridge 1 will become the designated bridge
between Token-Ring 1 and Ethernet 1 networks. Here the term designated
bridge references the bridge that has the bridge port with the lowest-cost path
to the root bridge.
In examining the path costs to the root bridge, note that the path through
bridge 2 was assigned a cost of 10, while the path through bridge 3 was
assigned a cost of 15. Thus, the path from Token-Ring 2 via bridge 2 to TokenRing 1 becomes the designated bridge between those two networks. Hence,
Figure 6.3b shows bridge 3 inactive by the omission of a connection to the
Token-Ring 2 network. Similarly, the path cost for connecting the Ethernet 3
network to the root bridge is lower by routing through the Token-Ring 2 and
Token-Ring 1 networks. Thus, bridge 5 becomes the designated bridge for the
Ethernet 3 and Token-Ring 2 networks.
Bridge Protocol Data Unit
As previously noted, bridges obtain topology information by the use of
bridge protocol data unit (BPDU) frames. Once a root bridge is selected, that
bridge is responsible for periodically transmitting a ‘‘HELLO’’ BPDU frame
to all networks to which it is connected. According to the spanning tree
protocol, HELLO frames must be transmitted every 1 to 10 seconds. The
BPDU has the group MAC address 800143000000, which is recognized by
each bridge. A designated bridge will then update the path cost and timing
information and forward the frame. A standby bridge will monitor the BPDUs,
but will not update nor forward them. If the designated bridge does not
receive a BPDU on its root port for a predefined period of time (default is
20 seconds), the designated bridge will assume that either a link or device
failure occurred. That bridge, if it is still receiving configuration BPDU frames
on other ports, will then switch its root port to a port that is receiving the best
configuration BPDUs.
When a standby bridge is required to assume the role of the root or designated
bridge, the HELLO BPDU will indicate that a standby bridge should become
a designated bridge. The process by which bridges determine their role in
a spanning tree network is iterative. As new bridges enter a network, they
assume a listening state to determine their role in the network. Similarly,
when a bridge is removed, another iterative process occurs to reconfigure the
remaining bridges.
bridging and switching methods and performance issues
291
Although the STP algorithm procedure eliminates duplicate frames and
degraded intranet performance, it can be a hindrance for situations where
multiple active paths between networks are desired. In addition, if a link or
device fails, the time required for a new spanning tree to be formed via the
transmission of BPDUs can easily require 45 to 60 seconds or more. Another
disadvantage of STP occurs when it is used in remote bridges connecting
geographically dispersed networks. For example, returning to Figure 6.2,
suppose Ethernet 1 were located in Los Angeles, Ethernet 2 in New York, and
Ethernet 3 in Atlanta. If the link between Los Angeles and New York were
placed in a standby mode of operation, all frames from Ethernet 2 routed to
Ethernet 1 would be routed through Atlanta. Depending on the traffic between
networks, this situation might require an upgrade in the bandwidth of the links
connecting each network to accommodate the extra traffic flowing through
Atlanta. Since the yearly cost of upgrading a 56- or 64-Kbps circuit to a 128Kbps fractional T1 link can easily exceed the cost of a bridge or router, you
might wish to consider the use of routers to accommodate this networking
situation. In comparison, when using local bridges, the higher operating
rate of local bridges in interconnecting local area networks normally allows
an acceptable level of performance when LAN traffic is routed through an
intermediate bridge.
Protocol Dependency
Another problem associated with the use of transparent bridges concerns the
differences between Ethernet and IEEE 802.3 frame field compositions. As
noted in Chapter 4, the Ethernet frame contains a type field that indicates
the higher-layer protocol in use. Under the IEEE 802.3 frame format, the type
field is replaced by a length field, and the data field is subdivided to include
logical link control (LLC) information in the form of destination (DSAP) and
source (SSAP) service access points. Here, the DSAP and SSAP are similar to
the type field in an Ethernet frame: they also point to a higher-level process.
Unfortunately, this small difference can create problems when you are using
a transparent bridge to interconnect Ethernet and IEEE 802.3 networks.
The top portion of Figure 6.6 shows the use of a bridge to connect an
AppleTalk network supporting several Macintosh computers to an Ethernet
network on which a Digital Equipment Corporation VAX computer is located.
Although the VAX may be capable of supporting DecNet Phase IV, which
is true Ethernet, and AppleTalk if both modules are resident, a pointer is
required to direct the IEEE 802.3 frames generated by the Macintosh to
the right protocol on the VAX. Unfortunately, the Ethernet connection used
292
chapter six
Apple
Dec
phase IV talk
Ethernet NIC
Ethernet
B
IEEE 802.3
M
M
Frame differences
Ethernet
Type
Information
IEEE 802.3
Length
Information
DSAP SSAP Control
Information
Legend:
= Workstation
M = Macintosh
Figure 6.6 Protocol differences preclude linking IEEE 802.3 and Ethernet
networks using transparent bridges. A Macintosh computer connected on an
IEEE 802.3 network using AppleTalk will not have its frame pointed to the
right process on a VAX on an Ethernet. Thus, the differences between Ethernet
and IEEE 802.3 networks require transparent bridges for interconnecting
similar networks.
by the VAX will not provide the required pointer. This explains why you
should avoid connecting Ethernet and IEEE 802.3 networks via transparent
bridges. Fortunately, almost all Ethernet NICs manufactured today are IEEE
802.3–compatible to alleviate this problem; however, older NICs may operate
as true Ethernets and result in the previously mentioned problem.
Source Routing
Source routing is a bridging technique developed by IBM for connecting
Token-Ring networks. The key to the implementation of source routing is the
bridging and switching methods and performance issues
293
use of a portion of the information field in the Token-Ring frame to carry
routing information and the transmission of discovery packets to determine
the best route between two networks.
The presence of source routing is indicated by the setting of the first bit
position in the source address field of a Token-Ring frame to a binary 1. When
set, this indicates that the information field is preceded by a route information
field (RIF), which contains both control and routing information.
The RIF Field
Figure 6.7 illustrates the composition of a Token-Ring RIF. This field is
variable in length and is developed during a discovery process, described
later in this section.
Field format
Up to 16 bytes
2 bytes
Control
Ring #
Bridge #
B B B L L L L L D LF LF LF LF R R R
Ring #
Ring #
12 bits
Bridge #
Bridge #
4 bits
B are broadcast bits
Bit settings
Designator
Nonbroadcast
0XX
All-routes broadcast
10X
Single route broadcast
11X
L are length bits which denote length of the RIF in bytes
D is direction bit
LF identifies largest frame
Bit settings
Size in bytes
000
001
010
011
100
101
110
111
R are reserved bits
516
1500
2052
4472
8191
Reserved
Reserved
Used in all-routes broadcast frame
Figure 6.7 Token-Ring route information field. The Token-Ring RIF is variable in length.
294
chapter six
The control field contains information that defines how information will be
transferred and interpreted and what size the remainder of the RIF will be. The
three broadcast bit positions indicate a nonbroadcast, all-routes broadcast, or
single-route broadcast situation. A nonbroadcast designator indicates a local
or specific route frame. An all-routes broadcast designator indicates that a
frame will be transmitted along every route to the destination station. A
single-route broadcast designator is used only by designated bridges to relay
a frame from one network to another. In examining the broadcast bit settings
shown in Figure 6.7, note that the letter X indicates an unspecified bit setting
that can be either a 1 or 0.
The length bits identify the length of the RIF in bytes, while the D bit
indicates how the field is scanned, left to right or right to left. Since vendors
have incorporated different memory in bridges which may limit frame sizes,
the LF bits enable different devices to negotiate the size of the frame. Normally,
a default setting indicates a frame size of 512 bytes. Each bridge can select
a number, and if it is supported by other bridges, that number is then used
to represent the negotiated frame size. Otherwise, a smaller number used
to represent a smaller frame size is selected, and the negotiation process is
repeated. Note that a 1500-byte frame is the largest frame size supported
by Ethernet IEEE 802.3 networks. Thus, a bridge used to connect Ethernet
and Token-Ring networks cannot support the use of Token-Ring frames
exceeding 1500 bytes.
Up to eight route number subfields, each consisting of a 12-bit ring number
and a 4-bit bridge number, can be contained in the routing information field.
This permits two to eight route designators, enabling frames to traverse up
to eight rings across seven bridges in a given direction. Both ring numbers
and bridge numbers are expressed as hexadecimal characters, with three hex
characters used to denote the ring number and one hex character used to
identify the bridge number.
Operation Example
To illustrate the concept behind source routing, consider the intranet illustrated in Figure 6.8. In this example, let us assume that two Token-Ring
networks are located in Atlanta and one network is located in New York.
Each Token-Ring and bridge is assigned a ring or bridge number. For simplicity, ring numbers R1, R2, and R3 are used here, although as previously
explained, those numbers are actually represented in hexadecimal. Similarly, bridge numbers are shown here as B1, B2, B3, B4, and B5 instead of
hexadecimal characters.
bridging and switching methods and performance issues
295
0 R3 B2 R1 B1 R1 A
A
C
A
B1 R1 A
B
R1
R3
B2
B1
D
New York
B3
Atlanta
B5
0 R3 B5 R2 B4 R2 B3 R1 A
B3 R1 A
R2
B4
B4 R2 B3 R1 A
Figure 6.8 Source routing discovery operation. The route discovery process
results in each bridge entering the originating ring number and its bridge
number into the RIF.
When a station wants to originate communications, it is responsible for
finding the destination by transmitting a discovery packet to network bridges
and other network stations whenever it has a message to transmit to a new
destination address. If station A wishes to transmit to station C, it sends a
route discovery packet containing an empty RIF and its source address, as
indicated in the upper left portion of Figure 6.8. This packet is recognized
by each source routing bridge in the network. When a source routing bridge
receives the packet, it enters the packet’s ring number and its own bridge
identifier in the packet’s routing information field. The bridge then transmits
the packet to all of its connections except the connection on which the packet
was received, a process known as flooding. Depending on the topology of the
interconnected networks, it is more than likely that multiple copies of the
discovery packet will reach the recipient. This is illustrated in the upper right
corner of Figure 6.8, in which two discovery packets reach station C. Here, one
packet contains the sequence R1B1R1B2R30 — the zero indicates that there is
no bridging in the last ring. The second packet contains the route sequence
R1B3R2B4R2B5R30. Station C then picks the best route, based on either the
most direct path or the earliest arriving packet, and transmits a response to
296
chapter six
the discover packet originator. The response indicates the specific route to
use, and station A then enters that route into memory for the duration of the
transmission session.
Under source routing, bridges do not keep routing tables like transparent
bridges. Instead, tables are maintained at each station throughout the network.
Thus, each station must check its routing table to determine what route frames
must traverse to reach their destination station. This routing method results
in source routing using distributed routing tables instead of the centralized
routing tables used by transparent bridges.
Advantages
There are several advantages associated with source routing. One advantage is
the ability to construct mesh networks with loops for a fault-tolerant design;
this cannot be accomplished with the use of transparent bridges. Another
advantage is the inclusion of routing information in the information frames.
Several vendors have developed network management software products that
use that information to provide statistical information concerning intranet
activity. Those products may assist you in determining how heavily your
wide area network links are being used, and whether you need to modify the
capacity of those links; they may also inform you if one or more workstations
are hogging communications between networks.
Disadvantages
Although the preceding advantages are considerable, they are not without
a price. That price includes a requirement to identify bridges and links
specifically, higher bursts of network activity, and an incompatibility between
Token-Ring and Ethernet networks. In addition, because the structure of the
Token-Ring RIF supports a maximum of seven entries, routing of frames is
restricted to crossing a maximum of seven bridges.
When using source routing bridges to connect Token-Ring networks, you
must configure each bridge with a unique bridge/ring number. In addition,
unless you wish to accept the default method by which stations select a frame
during the route discovery process, you will have to reconfigure your LAN
software. Thus, source routing creates an administrative burden not incurred
by transparent bridges.
Due to the route discovery process, the flooding of discovery frames occurs in
bursts when stations are turned on or after a power outage. Depending upon
the complexity of an intranet, the discovery process can degrade network
bridging and switching methods and performance issues
297
performance. This is perhaps the most problematic for organizations that
require the interconnection of Ethernet and Token-Ring networks.
A source routing bridge can be used only to interconnect Token-Ring
networks, since it operates on RIF data not included in an Ethernet frame.
Although transparent bridges can operate in Ethernet, Token-Ring, and mixed
environments, their use precludes the ability to construct loop or mesh
topologies, and inhibits the ability to establish operational redundant paths
for load sharing. Another problem associated with bridging Ethernet and
Token-Ring networks involves the RIF in a Token-Ring frame. Unfortunately,
different LAN operating systems use the RIF data in different ways. Thus,
the use of a transparent bridge to interconnect Ethernet and Token-Ring
networks may require the same local area network operating system on each
network. To alleviate these problems, several vendors introduced source
routing transparent (SRT) bridges, which function in accordance with the
IEEE 802.1D standard approved during 1992.
Source Routing Transparent Bridges
A source routing transparent bridge supports both IBM’s source routing and
the IEEE transparent STP operations. This type of bridge can be considered
two bridges in one; it has been standardized by the IEEE 802.1 committee as
the IEEE 802.1D standard.
Operation
Under source routing, the MAC packets contain a status bit in the source field
that identifies whether source routing is to be used for a message. If source
routing is indicated, the bridge forwards the frame as a source routing frame. If
source routing is not indicated, the bridge determines the destination address
and processes the packet using a transparent mode of operation, using routing
tables generated by a spanning tree algorithm.
Advantages
There are several advantages associated with source routing transparent
bridges. First and perhaps foremost, they enable different networks to use
different local area network operating systems and protocols. This capability enables you to interconnect networks developed independently of one
another, and allows organization departments and branches to use LAN
operating systems without restriction. Secondly, also a very important consideration, source routing transparent bridges can connect Ethernet and
298
chapter six
Token-Ring networks while preserving the ability to mesh or loop TokenRing networks. Thus, their use provides an additional level of flexibility for
network construction.
Translating Operations
When interconnecting Ethernet/IEEE 802.3 and Token-Ring networks, the
difference between frame formats requires the conversion of frames. A bridge
that performs this conversion is referred to as a translating bridge.
As previously noted in Chapter 4, there are several types of Ethernet frames,
such as Ethernet, IEEE 802.3, Novell’s Ethernet-802.3, and Ethernet-SNAP.
The latter two frames represent variations of the physical IEEE 802.3 frame
format. Ethernet and Ethernet-802.3 do not use logical link control, while IEEE
802.3 CSMA/CD LANs specify the use of IEEE 802.2 logical link control. In
comparison, all IEEE 802.5 Token-Ring networks either directly or indirectly
use the IEEE 802.2 specification for logical link control.
The conversion from IEEE 802.3 to IEEE 802.5 can be accomplished by
discarding portions of the IEEE 802.3 frame not applicable to a TokenRing frame, copying the 802.2 LLC protocol data unit (PDU) from one
frame to another, and inserting fields applicable to the Token-Ring frame.
Figure 6.9 illustrates the conversion process performed by a translating bridge
IEEE 802.3
Preamble DA SA Length DSAP SSAP Control
Discard
insert
Discard
insert
Copy
SD AC FC DA SA
RIF
Information
Discard
insert
Copy
DSAP SSAP Control
FCS
Information FCS ED FS
Legend:
DA
SA
AC
FC
RIF
DSAP
SSAP
ED
FS
= Destination Address
= Source Address
= Access Control
= Frame Control
= Routing Information Field
= Destination Service Access Point
= Source Service Access Point
= End Delimiter
= Frame Status Field
Figure 6.9
IEEE 802.3 to 802.5 frame conversion.
bridging and switching methods and performance issues
299
Ethernet
Preamble DA SA
Tokenring
Discard
insert
Type
Copy
Copy
SD AC FC DA SA
Data
RIF
FCS
FCS
Discard
insert
DSAP SSAP Control OC Type Data FCS
ED FS
Legend:
DA
SA
AC
FC
RIF
DSAP
SSAP
ED
FS
OC
Figure 6.10
= Destination Address
= Source Address
= Access Control
= Frame Control
= Routing Information Field
= Destination Service Access Point
= Source Service Access Point
= End Delimiter
= Frame Status Field
= Organization Code
Ethernet to Token-Ring frame conversion.
linking an IEEE 802.3 network to an IEEE 802.5 network. Note that fields
unique to the IEEE 802.3 frame are discarded, while fields common to both
frames are copied. Fields unique to the IEEE 802.5 frame are inserted by
the bridge.
Since an Ethernet frame, as well as Novell’s Ethernet-802.3 frame, does not
support logical link control, the conversion process to IEEE 802.5 requires
more processing. In addition, each conversion is more specific and may or
may not be supported by a specific translating bridge. For example, consider
the conversion of Ethernet frames to Token-Ring frames. Since Ethernet does
not support LLC PDUs, the translation process results in the generation of a
Token-Ring-SNAP frame. This conversion or translation process is illustrated
in Figure 6.10.
6.2 Bridge Network Utilization
In this section, we will examine the use of bridges to interconnect separate
local area networks and to subdivide networks to improve performance.
In addition, we will focus our attention on how we can increase network
300
chapter six
availability by employing bridges to provide alternate communications paths
between networks.
Serial and Sequential Bridging
The top of Figure 6.11 illustrates the basic use of a bridge to interconnect two
networks serially. Suppose that monitoring of each network indicates a high
level of intranetwork use. One possible configuration to reduce intra-LAN
traffic on each network can be obtained by moving some stations off each of
the two existing networks to form a third network. The three networks would
then be interconnected through the use of an additional bridge, as illustrated
in the middle portion of Figure 6.11. This extension results in sequential or
cascaded bridging, and is appropriate when intra-LAN traffic is necessary but
minimal. This intranet topology is also extremely useful when the length of an
Ethernet must be extended beyond the physical cabling of a single network. By
locating servers appropriately within each network segment, you may be able
to minimize inter-LAN transmission. For example, the first network segment
could be used to connect marketing personnel, while the second and third
segments could be used to connect engineering and personnel departments.
This might minimize the use of a server on one network by persons connected
to another network segment.
A word of caution is in order concerning the use of bridges. Bridging
forms what is referred to as a flat network topology, because it makes its
forwarding decisions using layer 2 MAC addresses, which cannot distinguish
one network from another. This means that broadcast traffic generated on
one segment will be bridged onto other segments which, depending upon
the amount of broadcast traffic, can adversely affect the performance on
other segments.
The only way to reduce broadcast traffic between segments is to use a
filtering feature included with some bridges or install routers to link segments. Concerning the latter, routers operate at the network layer and forward
packets explicitly addressed to a different network. Through the use of network addresses for forwarding decisions, routers form hierarchical structured
networks, eliminating the so-called broadcast storm effect that occurs when
broadcast traffic generated from different types of servers on different segments
are automatically forwarded by bridges onto other segments.
Both serial and sequential bridging are applicable to transparent, source
routing, and source routing transparent bridges that do not provide redundancy nor the ability to balance traffic flowing between networks. Each of these
deficiencies can be alleviated through the use of parallel bridging. However,
bridging and switching methods and performance issues
301
Serial bridging
B
Sequential or cascaded bridging
B
B
Parallel bridging
B
B
Figure 6.11
Legend:
= Workstation
B = Bridge
Serial, sequential, and parallel bridging.
this bridging technique creates a loop and is only applicable to source routing
and source routing transparent bridges.
Parallel Bridging
The lower portion of Figure 6.11 illustrates the use of parallel bridges to
interconnect two Token-Ring networks. This bridging configuration permits
302
chapter six
one bridge to back up the other, providing a level of redundancy for linking
the two networks as well as a significant increase in the availability of one
network to communicate with another. For example, assume the availability
of each bridge used at the top of Figure 6.11 (serial bridging) and bottom
of Figure 6.11 (parallel bridging) is 90 percent. The availability through two
serially connected bridges would be 0.9 × 0.9 (availability of bridge 1 ×
availability of bridge 2), or 81 percent. In comparison, the availability through
parallel bridges would be 1 − (0.1 × 0.1), which is 99 percent.
The dual paths between networks also improve inter-LAN communications
performance, because communications between stations on each network can
be load balanced. The use of parallel bridges can thus be expected to provide a
higher level of inter-LAN communications than the use of serial or sequential
bridges. However, as previously noted, this topology is not supported by
transparent bridging.
Star Bridging
With a multiport bridge, you can connect three or more networks to form a
star intranet topology. The top portion of Figure 6.12 shows the use of one
bridge to form a star topology by interconnecting four separate networks.
This topology, or a variation on this topology, could be used to interconnect
networks on separate floors within a building. For example, the top network
could be on floor N + 1, while the bottom network could be on floor N − 1 in
a building. The bridge and the two networks to the left and right of the bridge
might then be located on floor N.
Although star bridging permits several networks located on separate floors
within a building to be interconnected, all intranet data must flow through
one bridge. This can result in both performance and reliability constraints to
traffic flow. Thus, to interconnect separate networks on more than a few floors
in a building, you should consider using backbone bridging.
Backbone Bridging
The lower portion of Figure 6.12 illustrates the use of backbone bridging. In
this example, one network runs vertically through a building with Ethernet
ribs extending from the backbone onto each floor. Depending upon the amount
of intranet traffic and the vertical length required for the backbone network,
the backbone can be either a conventional Ethernet bus-based network or a
fiber-optic backbone.
bridging and switching methods and performance issues
303
Star bridging
B
Backbone bridging
B
B
B
Legend:
B
= Workstation
= Bridge
Figure 6.12
Star and backbone bridging.
6.3 Bridge Performance Issues
The key to obtaining an appropriate level of performance when interconnecting networks is planning. The actual planning process will depend upon
several factors, such as whether separate networks are in operation, the type
of networks to be connected, and the type of bridges to be used — local
or remote.
Traffic Flow
If separate networks are in operation and you have appropriate monitoring
equipment, you can determine the traffic flow on each of the networks to be
304
chapter six
interconnected. Once this is accomplished, you can expect an approximate
10- to 20-percent increase in network traffic. This additional traffic represents
the flow of information between networks after an interconnection links previously separated local area networks. Although this traffic increase represents
an average encountered by the author, your network traffic may not represent
the typical average. To explore further, you can examine the potential for
intranet communications in the form of electronic messages that may be transmitted to users on other networks, potential file transfers of word processing
files, and other types of data that would flow between networks.
Network Types
The types of networks to be connected will govern the rate at which frames
are presented to bridges. This rate, in turn, will govern the filtering rate at
which bridges should operate so that they do not become bottlenecks on
a network. For example, the maximum number of frames per second will
vary between different types of Ethernet and Token-Ring networks, as well as
between different types of the same network. The operating rate of a bridge
may thus be appropriate for connecting some networks while inappropriate
for connecting other types of networks.
Type of Bridge
Last but not least, the type of bridge — local or remote — will have a considerable bearing upon performance issues. Local bridges pass data between
networks at their operating rates. In comparison, remote bridges pass data
between networks using wide area network transmission facilities, which
typically provide a transmission rate that is only a fraction of a local area
network operating rate. Now that we have discussed some of the aspects
governing bridge and intranet performance using bridges, let’s probe deeper
by estimating network traffic.
Estimating Network Traffic
If we do not have access to monitoring equipment to analyze an existing
network, or if we are planning to install a new network, we can spend some
time developing a reasonable estimate of network traffic. To do so, we should
attempt to classify stations into groups based on the type of general activity
performed, and then estimate the network activity for one station per group.
This will enable us to multiply the number of stations in the group by the
bridging and switching methods and performance issues
305
station activity to determine the group network traffic. Adding up the activity
of all groups will then provide us with an estimate of the traffic activity for
the network.
As an example of local area network traffic estimation, let us assume that our
network will support 20 engineers, 5 managers, and 3 secretaries. Table 6.1
shows how we would estimate the network traffic in terms of the bit rate
for each station group and the total activity per group, and then sum up the
network traffic for the three groups that will use the network. In this example,
which for the sake of simplicity does not include the transmission of data to
a workstation printer, the total network traffic was estimated to be slightly
below 50,000 bps.
TABLE 6.1
Estimating Network Traffic
Activity
Engineering workstations
Request program
Load program
Save files
Send/receive e-mail
Total engineering activity
= 1,613 × 20 = 32,260 bps
Managerial workstations
Request program
Load program
Save files
Send/receive e-mail
Total managerial activity
= 1,590 × 5 = 7,950 bps
Secretarial workstations
Request program
Load program
Save files
Send/receive e-mail
Total secretarial activity
= 3,112 × 3 = 9,336 bps
Total estimated network activity
= 49,546 bps
∗ Note: Bit rate is computed
by 3,600 seconds/hour.
Message Size
(Bytes)
Frequency
Bit Rate∗
1,500
480,000
120,000
2,000
1/hour
1/hour
2/hour
2/hour
4
1,067
533
9
1,613
1,500
320,000
30,000
3,000
2/hour
2/hour
2/hour
4/hour
7
1,422
134
27
1,590
1,500
640,000
12,000
3,000
4/hour
2/hour
8/hour
6/hour
14
2,844
214
40
3,112
by multiplying message rate by frequency by 8 bits/byte and dividing
306
chapter six
To plan for the interconnection of two or more networks through the use
of bridges, our next step should be to perform a similar traffic analysis for
each of the remaining networks. After this is accomplished, we can use the
network traffic to estimate inter-LAN traffic, using 10 to 20 percent of total
intranetwork traffic as an estimate of the intranet traffic that will result from
the connection of separate networks.
Intranet Traffic
To illustrate the traffic estimation process for the interconnection of separate
LANs, let us assume that network A’s traffic was determined to be 50,000 bps,
while network B’s traffic was estimated to be approximately 100,000 bps.
Figure 6.13 illustrates the flow of data between networks connected by a local
bridge. Note that the data flow in each direction is expressed as a range, based
on the use of an industry average of 10 to 20 percent of network traffic routed
between interconnected networks.
Network Types
Our next area of concern is to examine the types of networks to be interconnected. In doing so, we should focus our attention on the operating rate of
Network A
Local traffic
50,000 bps
Total traffic 60,000 to 70,000 bps
10,000 to 20,000 bps
B
Total traffic 105,000 to 110,000 bps
Network B
Local traffic
100,000 bps
Legend:
5,000 to 10,000 bps
= Workstation
Figure 6.13 Considering intranet data flow. To determine the traffic flow on
separate networks after they are interconnected, you must consider the flow
of data onto each network from the other network.
bridging and switching methods and performance issues
307
each LAN. If network A’s traffic was estimated to be approximately 50,000 bps,
then the addition of 10,000 to 20,000 bps from network B onto network A will
raise network A’s traffic level to between 60,000 and 70,000 bps. Similarly,
the addition of traffic from network A onto network B will raise network B’s
traffic level to between 105,000 and 110,000 bps. In this example, the resulting
traffic on each network is well below the operating rate of all types of local
area networks, and will not present a capacity problem for either network.
Bridge Type
As previously mentioned, local bridges transmit data between networks at
the data rate of the destination network. This means that a local bridge will
have a lower probability of being a bottleneck than a remote bridge, since the
latter provides a connection between networks using a wide area transmission
facility, which typically operates at a fraction of the operating rate of a LAN.
In examining the bridge operating rate required to connect networks, we
will use a bottom-up and a top-down approach. That is, we will first determine
the operating rate in frames per second for the specific example previously
discussed. This will be followed by computing the maximum frame rate
supported by an Ethernet network.
For the bridge illustrated in Figure 6.13, we previously computed that its
maximum transfer rate would be 20,000 bps from network B onto network A.
This is equivalent to 2500 bytes per second. If we assume that data is transported in 512-byte frames, this would be equivalent to 6 frames per second — a
minimal transfer rate supported by every bridge manufacturer. However, when
remote bridges are used, the frame forwarding rate of the bridge will more
than likely be constrained by the operating rate of the wide area network
transmission facility.
Bridge Operational Considerations
A remote bridge wraps a LAN frame into a higher-level protocol packet
for transmission over a wide area network communications facility. This
operation requires the addition of a header, protocol control, error detection,
and trailer fields, and results in a degree of overhead. A 20,000-bps data flow
from network B to network A, therefore, could not be accommodated by a
transmission facility operating at that data rate.
In converting LAN traffic onto a wide area network transmission facility,
you can expect a protocol overhead of approximately 20 percent. Thus, your
actual operating rate must be at least 24,000 bps before the wide area network
communications link becomes a bottleneck and degrades communications.
chapter six
308
Now that we have examined the bridging performance requirements for two
relatively small networks, let us focus our attention on determining the
maximum frame rates of an Ethernet network. This will provide us with the
ability to determine the rate at which the frame processing rate of a bridge
becomes irrelevant, since any processing rate above the maximum network
rate will not be useful. In addition, we can use the maximum network
frame rate when estimating traffic, because if we approach that rate, network
performance will begin to degrade significantly when use exceeds between 60
to 70 percent of that rate.
Ethernet Traffic Estimation
An Ethernet frame can vary between a minimum of 72 bytes and a maximum
of 1526 bytes. Thus, the maximum frame rate on an Ethernet will vary with
the frame size.
Ethernet operations require a dead time between frames of 9.6 µsec. The bit
time for a 10-Mbps Ethernet is 1/107 or 100 nsec. Based upon the preceding,
we can compute the maximum number of frames/second for 1526-byte frames.
Here, the time per frame becomes:
9.6 µsec + 1526 bytes × 8 bits/byte
or
9.6 µsec + 12,208 bits × 100 nsec/bit
or
1.23 msec
Thus, in one second there can be a maximum of 1/1.23 msec or 812 maximumsize frames. For a minimum frame size of 72 bytes, the time per frame is:
9.6 µsec + 72 bytes × 8 bits/byte × 100 nsec/bit
or
67.2 × 10−6 sec.
Thus, in one second there can be a maximum of 1/67.2 × 10−6 or 14,880
minimum-size 72-byte frames. Since 100BASE-T Fast Ethernet uses the same
frame composition as Ethernet, the maximum frame rate for maximum- and
minimum-length frames are ten times that of Ethernet. That is, Fast Ethernet
supports a maximum of 8120 maximum-size 1526-byte frames per second
and a maximum of 148,800 minimum-size 72-byte frames per second. Similarly, Gigabit Ethernet uses the same frame composition as Ethernet but is
100 times faster. This means that Gigabit Ethernet is capable of supporting a maximum of 81,200 maximum-length 1526-byte frames per second
bridging and switching methods and performance issues
309
and a maximum of 1,488,000 minimum-length 72-byte frames per second.
As you might expect, 10 Gigabit Ethernet expands support by an order of
magnitude beyond the frame rate of Gigabit Ethernet. For both Gigabit and
10 Gigabit Ethernet the maximum frame rates are for full-duplex operations.
Table 6.2 summarizes the frame processing requirements for a 10-Mbps Ethernet, Fast Ethernet, Gigabit Ethernet, and 10 Gigabit Ethernet under 50 percent
and 100 percent load conditions, based on minimum and maximum frame
lengths. Note that those frame processing requirements define the frame examination (filtering) operating rate of a bridge connected to different types of
Ethernet networks. That rate indicates the number of frames per second a
bridge connected to different types of Ethernet local area networks must be
capable of examining under heavy (50-percent load) and full (100-percent
load) traffic conditions.
In examining the different Ethernet network frame processing requirements
indicated in Table 6.2, it is important to note that the frame processing requirements associated with Fast Ethernet, Gigabit Ethernet, and 10 Gigabit Ethernet
commonly preclude the ability to upgrade a bridge by simply changing its
adapter cards. Due to the much greater frame processing requirements associated with very high speed Ethernet networks, bridges are commonly designed
to support those technologies from the ground up to include adapters and a
TABLE 6.2 Ethernet Frame Processing Requirements
(Frames per Second)
Frame Processing Requirements
Average Frame Size (Bytes)
Ethernet
1526
72
Fast Ethernet
1526
72
Gigabit Ethernet
1526
72
10 Gigabit Ethernet
1526
72
50% Load
100% Load
406
7440
812
14,880
4050
74,400
8120
148,800
40,600
744,000
81,200
1,488,000
406,000
7,440,000
812,000
14,880,000
310
chapter six
central processor to support the additional frame processing associated with
their higher operating rate.
We can extend our analysis of Ethernet frames by considering the frame
rate supported by different link speeds. For example, let us consider a pair of
remote bridges connected by a 9.6-Kbps line. The time per frame for a 72-byte
frame at 9.6 Kbps is:
9.6 × 10−6 + 72 × 8 × 0.0001041 s/bit or 0.0599712 seconds per frame
Thus, in one second the number of frames is 1/.0599712, or 16.67 frames per
second. Table 6.3 compares the frame-per-second rate supported by different
link speeds for minimum- and maximum-size Ethernet frames. As expected,
the frame transmission rate supported by a 10-Mbps link for minimum- and
maximum-size frames is exactly the same as the frame processing requirements
under 100 percent loading for a 10-Mbps Ethernet LAN, as indicated in
Table 6.2.
In examining Table 6.3, note that the entries in this table do not consider
the effect of the overhead of a protocol used to transport frames between
two networks. You should therefore decrease the frame-per-second rate by
approximately 20 percent for all link speeds through 1.536 Mbps. The reason
the 10-Mbps rate should not be adjusted is that it represents a local 10Mbps Ethernet bridge connection that does not require the use of a wide
area network protocol to transport frames. Also note that the link speed of
1.536 Mbps represents a T1 transmission facility that operates at 1.544 Mbps.
However, since the framing bits on a T1 circuit use 8 Kbps, the effective line
speed available for the transmission of data is 1.536 Mbps.
TABLE 6.3
Link Speed versus Frame Rate
Frames per Second
Link Speed
Minimum
Maximum
9.6 Kbps
16.67
0.79
19.2 Kbps
33.38
1.58
56.0 Kbps
97.44
4.60
64.0 Kbps
111.17
5.25
2815.31
136.34
14,880.00
812.00
1.536 Mbps
10.0 Mbps
bridging and switching methods and performance issues
311
Predicting Throughput
Until now, we have assumed that the operating rate of each LAN linked
by a bridge is the same. However, in many organizations this may not
be true, because LANs are implemented at different times using different
technologies. Thus, accounting may be using a 10-Mbps LAN, while the
personnel department might be using a 100-Mbps LAN.
Suppose we wanted to interconnect the two LANs via the use of a multimedia bridge. To predict throughput between LANs, let us use the network
configuration illustrated in Figure 6.14. Here, the operating rate of LAN A is
assumed to be R1 bps, while the operating rate of LAN B is assumed to be
R2 bps.
In one second, R1 bits can be transferred on LAN A and R2 bits can be
transferred on LAN B. Similarly, it takes 1/R1 seconds to transfer one bit on
LAN A and 1/R2 seconds to transfer one bit on LAN B. So, to transfer one
bit across the bridge from LAN A to LAN B, ignoring the transfer time at
the bridge:
1
1
1
=
+
RT
R1
R2
or
RT =
1
(1/R1 ) + (1/R2 )
We computed that a 10-Mbps Ethernet would support a maximum transfer of
812 maximum-sized frames per second. If we assume that the second LAN
operating at 100 Mbps is also an Ethernet, we would compute its transfer rate
to be approximately 8120 maximum-sized frames per second. The throughput
in frames per second would then become:
RT =
1
= 738 frames per second
(1/812) + (1/8120)
Knowing the transfer rate between LANs can help us answer many common
questions. It can also provide us with a mechanism for determining whether
LAN A
R1
Bridge
R2
LAN B
Figure 6.14 Linking LANs with different operating rates. When LANs with
different operating rates (R1 and R2 ) are connected via a bridge, access of files
across the bridge may result in an unacceptable level of performance.
312
chapter six
we should alter the location of application programs on different servers. For
example, suppose that a program located on a server on LAN B suddenly
became popular for use by workstations on LAN A. If the program required
1024 K of storage, we could estimate the minimum transfer time required
to load that program and, depending on the results of our calculation, we
might want to move the program onto a server on LAN A. For this example,
the data transfer rate would be 738 frames/second × 1500 bytes/frame or
1,107,000 bytes per second. Dividing the data to be transferred by the data
transfer rate, we obtain:
1024 Kbytes × 1024 bytes/k
= .95 seconds
1,107,000 bytes/seconds
The preceding computation represents a best-case scenario, in which the use
of each network is limited to the user on LAN A loading a program from a
server on LAN B. In reality, the average number of users using the resources
of each network must be used to adjust the values of R1 and R2 . For example,
suppose that through monitoring you determined that the average number of
active users on LAN A was 5 and on LAN B was 10. In that case, you would
adjust the value of R1 by dividing 812 by 5 to obtain 162.4 and adjust the
value of R2 by dividing 8120 by 10 to obtain 812. You would then use the new
values of R1 and R2 to obtain the average value of RT , based on the fact that
the program loading operation would be performed concurrently with other
operations on each network. Thus, you would compute RT as follows:
RT =
1
= 135.4 frames per second
(1/162.5) + (1/812)
6.4 LAN Switches
The incorporation of microprocessor technology into hubs can be considered
as the first step in the development of switching hubs, which are now more
commonly referred to as LAN switches. Through additional programming,
the microprocessor could examine the destination address of each frame;
however, switching capability required the addition of a switching fabric
design into the hub. Once this was accomplished, it became possible to use
the microprocessor to read the destination address of each frame and initiate
a switching action based upon data stored in the hub’s memory, which
associates destination frame addresses with hub ports.
bridging and switching methods and performance issues
313
There are several basic types of LAN switches, with the major difference
between each type resulting from the layer in the ISO Reference Model where
switching occurs. A layer 2 switch looks into each frame to determine the
destination MAC address while a layer 3 switch looks further into the frame to
determine the destination network address. Similarly, a layer 4 switch looks
even further into each frame to focus upon the transport layer header. Depending upon the software that supports switch operations a layer 4 switch may
be programmed to make switching decisions based upon two or more criteria,
such as destination IP address and port number. Thus, a layer 2 switch operates at the MAC layer and can be considered to represent a sophisticated bridge
while a layer 3 switch resembles a router. In comparison, a layer 4 switch that
uses multiple metrics in determining where to forward frames could function
as a traffic load balancer. Layer 2, layer 3, and layer 4 operations will be
covered as we examine the operation and use of LAN switches.
In this section we will first examine the rationale for LAN switches by noting
the bottlenecks associated with conventional and intelligent hubs as network
traffic grows. Once this is accomplished, we will focus upon the operation
and usage of different types of LAN switches.
Rationale
The earliest types of Ethernet LANs were designed to use coaxial cable
configured using a bus topology. The development of the hub-based 10BASET local area network offered a number of networking advantages over the use
of coaxial cable. Some of those advantages included the use of twisted-pair
cable, which is easier to use and less expensive than coaxial cable, and the
ability to reconfigure, troubleshoot, and isolate network problems. By simply
moving a cable connection from one port to another network, administrators
can easily adjust the usage of a hub or interconnect hubs to form a new network
structure. The connection of test equipment to a hub, either to a free port or by
temporarily removing an existing network user, could be accomplished much
easier than with a coaxial-based network. Recognizing these advantages, hub
manufacturers added microprocessors to their hubs, which resulted in the
introduction of a first generation of intelligent Ethernet hubs.
The first generation of intelligent hubs used the capability of a builtin microprocessor to provide a number of network management functions
network administrators could use to better control the operation and usage of
their network. Those functions typically include tracking the network usage
level and providing summary statistics concerning the transmission of frames
by different workstations, as well as providing the network administrator with
chapter six
314
the ability to segment the LAN by entering special commands recognized by
the hub.
Bottlenecks
Both conventional and first-generation intelligent hubs simply duplicate
frames and forward them to all nodes attached to the hub. This restricts
the flow of data to one workstation at a time, since collisions occur when
two or more workstations attempt to gain access to the media at the
same time.
Conventional hubs, to include the first generation of intelligent hubs, create
network bottlenecks, because all network traffic flows through a shared backplane. This results in every workstation connected to the hub competing for a
slice of the backplane’s bandwidth. For example, consider the hub illustrated
in Figure 6.15, in which up to seven workstations and a file server contend
for access to the network. Since only one device can transmit at any point in
time, the average slice of bandwidth that each device receives is 1.25 Mbps
(10 Mbps/8). The actual data transfer capability is less, since attempts by two
or more workstations to simultaneously transmit can result in collisions that
cause jam signals to be placed on the network, precluding other workstations
from transmitting data during the duration of those signals. As more users
are added to a network through the interconnection of hubs, network performance will continue to decrease as the potential for congestion increases.
Thus, manufacturers of Ethernet products, as well as network administrators, focused their efforts upon developing different tools and techniques to
alleviate network congestion.
Congestion-Avoidance Options
There are several techniques you can consider to alleviate network congestion.
Those techniques include splitting a network into two, with each segment
Server
Conventional hub
WS
WS
Figure 6.15 When using a conventional hub, congestion occurs when
several workstations vie for access to
a server.
bridging and switching methods and performance issues
315
connected to a separate server port, using bridges and dual servers, employing
a router to develop a collapsed backbone network linking multiple network
segments, or using one or more intelligent switches.
Network Segmentation
One of the earliest methods used to alleviate network congestion was obtained
through the use of a server with an internal bridging capability. Splitting the
network infrastructure into two and connecting each resulting segment to an
NIC installed in a server provides the capability to reduce traffic on each segment, in comparison to traffic previously carried on a nonsegmented network.
Figure 6.16 illustrates the segmentation of a network into two on one server.
NetWare, as well as other LAN operating systems, includes a capability to
move packets between cable segments. This enables users on each segment
to transmit and receive information from users on other segments, as well as
maintain access to a common server. If server usage is low but network usage
is high, this method of network subdivision represents a cost-effective method
for reducing the effect of network congestion upon LAN performance.
In examining Figure 6.16, note that a workstation on each network segment
can simultaneously transmit data to the server or receive data from the
server. Thus, segmentation not only reduces network congestion, but in
addition can double throughput if the server is capable of supporting sustained
transmission to or receiving data from workstations on both segments.
Bridging
The major problem associated with the use of a server for network segmentation is the fact that it must perform internal bridging in addition to its
file server operations. This usually limits the ability of the server to support
network segments with a large number of workstations. In addition, the workstations on the connected segments still contend for the services of a common
server. Thus, the use of a stand-alone bridge is usually the next step to
Server
Figure 6.16 Network segmentation
using a common server. Through
the use of a file server that can
move packets between segments,
you obtain the ability to subdivide
a network.
316
chapter six
consider when the level of network usage adversely affects LAN performance,
and segmentation through the use of a file server is not a viable option.
Figure 6.17 illustrates the use of a bridge for network segmentation. Although
the segmentation shown in Figure 6.17 is similar to the segmentation shown
in Figure 6.16, there are several distinct differences between the two methods. First, the stand-alone bridge requires file servers to be located on one
or more network segments. Secondly, since a stand-alone bridge is performing the required bridging functions, more workstations can be placed
on each network segment than when a file server performs bridging operations. Workstations on each segment can simultaneously access the server on
each segment, permitting network throughput to double when such traffic is
localized to each segment.
Using a Router
Although primarily thought of as a mechanism to interconnect geographically
dispersed networks, routers can be used as switching devices to interconnect
network segments located within one geographical area, such as a building
or campus. The networking architecture associated with the use of a router
in this manner is referred to as a collapsed backbone, since the older busstructured Ethernet LAN is replaced by LAN segments that communicate
with one another through the router. Until 1994, the primary device used in
the center of a collapsed backbone network was a router. Since then, LAN
switches have gradually taken over the role of routers for reasons we will
discuss later in this chapter.
To other
hubs
To other
hubs
Bridge
Server
Server
Figure 6.17 Using a bridge for network segmentation. The use of a bridge
and one or more servers on each interconnected segment can significantly
increase network capacity by localizing more traffic on each segment.
bridging and switching methods and performance issues
317
Server
Router
Server
Server
Server
Figure 6.18 Collapsed router-based backbone. A collapsed backbone topology based upon the use of a router enables servers to be distributed based
upon network access requirements.
Figure 6.18 illustrates a collapsed backbone formed through the use of a
router. Note that you can locate servers on network segments for which server
access is limited to workstations on the segment, and connect servers that are
accessed by workstations on two or more segments to separate router ports.
Thus, a router provides you with the ability to distribute network traffic based
upon server access requirements. In addition, the construction of a collapsed
backbone offers other benefits over distributed networks. Those advantages
include the centralization of complexity and economy of operations. Rather
than dispersing routers, locating a single router near a network technical
control center can make network management easier. Then, if a routing
problem arises, you may not have to send a technician to every location where
hubs and servers are installed, since router management software may be able
to pinpoint the problem. Since one router may be able to replace a number of
less capable devices, you may also obtain some economic benefits from the
use of a single router.
Three additional benefits associated with the use of a router for creating a
collapsed backbone network are security, protocol support, and the control
of broadcast traffic. Many routers have a sophisticated filtering capability,
permitting you to program templates that enable only predefined applications
from predefined network users to access other network segments or servers
connected to the router. Concerning protocol support, many routers can
318
chapter six
support multiprotocol transmission. Thus, you could connect a router port to
a mainframe and pass SNA traffic between the mainframe and workstations
on different network segments, as well as switching other protocol traffic
between workstations and different types of servers. Since the use of a router
results in a hierarchical network structure where frames are transported based
upon network addresses, this provides the ability to restrict broadcast traffic
generated on one segment to that segment. Thus, service advertisements
commonly generated by different types of servers can be localized to the
segment where they are located instead of flowing onto other segments where
they would consume bandwidth.
Although the advantages associated with the use of routers to create a
collapsed backbone are considerable, their use has several disadvantages.
Those disadvantages include the complexity and cost of routers, as well as
the latency or delay time associated with moving packets from one network
segment to another segment. Concerning router latency, although all devices
have a degree of latency, the delay associated with routers greatly exceeds
that of bridges. The reason for this is the fact that routers operate at the
network layer while bridges operate at the data link layer. This means that a
bridge can inspect the beginning of each frame to determine its destination. In
comparison, a router must look into each frame to determine the destination
network address. Thus, the processing rate of a router is generally 50 to
75 percent of the processing rate of a bridge.
Another problem associated with the use of routers is the eggs in one basket
syndrome, since the failure of the router results in the inability of many
network users to access servers or send messages to users on other network
segments. The latter problem has been addressed by router manufacturers
through redundant logic and power supplies, as well as the use of modular
components that can be easily swapped out without bringing down the router
when a component failure occurs. Although the latter problem is applicable to
LAN switches, this device addresses some of the problems associated with the
use of routers, as well as — through improved functionality — permits LAN
switches to provide an enhanced level of overall network performance. Thus,
let’s focus our attention upon the operation and use of LAN switches in the
remainder of this chapter.
LAN Switch Operations
LAN switches can be categorized in three main ways — by their method of
operation, their support of single or multiple addresses per port, and the layer
in the ISO Reference Model where switching occurs. The method of operation
bridging and switching methods and performance issues
319
is commonly referred to as the switching technique, while the address-per-port
support is normally referred to as the switching method.
The third main way of categorizing a LAN switch is based upon the type of
address used as a switching decision criteria — data link layer MAC addresses,
network layer network addresses or transport layer port number. Since LAN
switches that support layer 3 switching actually represent a switch with a
built-in routing capability, we will discuss layer 3 switches as a separate entity.
We will also examine how a layer 4 LAN switch can function as an application
distributor by making switching decisions based upon the port number within
the transport header. Thus, we will first examine layer 2 switching hubs before
focusing our attention upon layer 3 and layer 4 switches.
Layer 2 Switching
A switch that operates at the data link layer is a multiport bridge that reads
MAC destination addresses and uses those addresses as a decision criterion
for determining where to switch each frame. When operating at the data
link layer, a LAN switch provides several distinct advantages over the use of
conventional and intelligent hubs. First, a LAN switch does not transmit or
regenerate data onto every port. Instead, unicast data flow is restricted to a
routing from one port to a second port. This precludes a network user from
operating a monitor and observing the flow of data on the network. Hence,
LAN switches, although not necessarily secure devices, provide a more secure
environment than a shared Ethernet LAN.
A second advantage of LAN switches is their ability to support simultaneous
communication sessions. For example, consider a simple four-port LAN
switch with port numbers 1, 2, 3, and 4. If a user on port 1 is switched to
a server on port 3, and another user on port 2 is switched to a server on
port 4, you obtain two very short-duration, simultaneous communications
sessions. Thus, in this example, bandwidth is double that of a conventional
hub. If the LAN switch supports only 10BASE-T operating rates, the four-port
switch provides a maximum of two simultaneous cross-connections, doubling
potential bandwidth to 20 Mbps. As we will note later in this chapter, many
LAN switches incorporate full-duplex, Fast Ethernet, Gigabit Ethernet, and
even 10 Gigabit Ethernet technology, providing bandwidth many orders of
magnitude beyond that obtainable from the use of a conventional hub.
Switching Techniques There are three switching techniques used by LAN
switches — cross-point, also referred to as cut-through or on the fly, store-andforward, and a hybrid method which alternates between the first two methods,
320
chapter six
based upon the frame error rate. As we will soon note, each technique has one
or more advantages and disadvantages associated with its operation.
Cross-Point A cross-point switch examines the destination address of each
packet entering the port. It then searches a predefined table of addresses
associated with ports to obtain a port destination. Once the port destination is determined, the switch initiates the cross-connection between the
incoming port and the outgoing port. Figure 6.19 illustrates cross-point/cutthrough switching.
As previously noted in this chapter, a layer 2 switch can be considered to
represent a more sophisticated type of bridge capable of supporting multiple
simultaneous cross-connections. Thus, it should come as no surprise that a
cross-point switch uses a backward learning algorithm to construct a portdestination address table. That is, the switch monitors MAC source addresses
encountered on each port to construct a port-destination address table. If the
destination address resides on the same port the frame was received from,
this indicates that the frame’s destination is on the current network and no
switching operation is required. Thus, the switch discards the frame. If the
destination address resides on a different port, the switch obtains the port
destination and initiates a cross-connection through the switch, transferring
the frame to the appropriate destination port where it is placed onto a network
where a node with the indicated destination address resides. If the destination
address is not found in the table, the switch floods the frame onto all ports
other than the port it was received on. Although flooding adversely affects the
capability of a switch to perform multiple simultaneous cross-connections,
the majority of this type of activity occurs when a switch is powered on and
1
2
Switch
engine
Ports
Look up
table
3
Legend:
1 = Packet address examined
2 = Port destination address obtained
3 = Switch to destination port
Figure 6.19
Cross-point/cut-through switching.
bridging and switching methods and performance issues
321
its port-address table is empty. Thereafter, flooding occurs periodically after
an entry is purged from the table due to aging and a new request to a purged
destination occurs, a new station becomes a recipient of traffic, or when a
broadcast address is encountered.
Cross-point switching minimizes the delay or latency associated with placing a packet received on one port onto another port. Since the switching
decision is made once the destination address is read, this means the full
packet is not examined. Thus, a cross-point switch cannot perform error
checking on a packet. This limitation does not represent a problem for most
networks due to extremely low error rates on local area networks. However,
when errored packets are encountered, they are passed from one network
segment to another. This results in an unnecessary increase in network use on
the destination segment, as a store-and-forward switch would simply discard
packets containing one or more bit errors.
Latency Considerations A cross-point switching method only requires a
small portion of a frame to be stored until it is able to read the destination
address, perform its table lookup operation, and initiate switching to an
appropriate output port. Due to this, latency is minimized.
Latency functions as a brake on two-way frame exchanges. For example, in a
client/server environment the transmission of a frame by a workstation results
in a server response. Thus, the minimum wait time is 2 ∗ latency for each
client-server exchange, lowering the effective throughput of the switch. We
can compute the minimum amount of latency associated with a cross-point
switch as follows. At a minimum, the switch must read 14 bytes (8 bytes for
the preamble and 6 bytes for the destination address) before being able to
initiate a search of its port-destination address table. At 10 Mbps we obtain:
9.6 µs + 14 bytes ∗ 8 bits/byte ∗ 100 ns/bit
or
9.6 × 10−6 + 112 ∗ 100 ∗ 10−9
or
20.8 ∗ 10−6 seconds
Here 9.6 µs represents the Ethernet interframe gap at an operating rate of
10 Mbps, while 100 ns/bit represents the bit duration of a 10-Mbps Ethernet
LAN. Thus, the minimum one-way latency not counting switch overhead
of a cut-through layer 2 switch is 20.8 ∗ 10−6 seconds, while the round-trip
minimum latency would be twice that duration.
At 100 Mbps the interframe gap would be reduced to .96 µs, while the bit
duration would be 10 ns. Thus, at 100 Mbps the minimum one-way latency
chapter six
322
for a cross-point switch becomes:
.96 × 10−6 + 112 ∗ 10 ∗ 10−9
or
2.08 ∗ 10−6 seconds
For Gigabit Ethernet the minimum one-way latency would be further reduced
by an order of magnitude to .208 ∗ 10−6 seconds.
Store-and-Forward A store-and-forward LAN switch stores the full incoming packet in a buffer. This enables the switch to perform a CRC check to
determine if the received packet is error-free. If it is, the switch uses the
destination address of the packet to perform a table lookup to obtain the
destination port address. Once that address is obtained, the switch performs a
cross-connect operation and forwards the packet to its destination. Otherwise,
the frame is considered to have one or more bits in error and will be discarded.
Besides checking the CRC a store-and-forward switch will examine the entire
frame. This enables other errors, such as runts and extended lengths (giant)
frames to be caught and sent to the great bit bucket in the sky instead of
being forwarded.
The storage of frames by a store-and-forward switch permits filtering
against various frame fields to occur. Although a few manufacturers of storeand-forward LAN switches support different types of filtering, the primary
advantage advertised by such manufacturers is data integrity and the ability to
perform translation switching, such as switching a frame between an Ethernet
network and a Token-Ring network. Since the translation process is extremely
difficult to accomplish on the fly due to the number of conversions of frame
data, most switch vendors first store the frame, resulting in store-and-forward
switches supporting translation between different types of connected networks. Concerning the data integrity capability of store-and-forward switches,
whether or not this is actually an advantage depends upon how you view the
additional latency introduced by the storage of a full frame in memory as well
as the necessity for error checking. Concerning the latter, switches should
operate error-free, so a store-and-forward switch only removes network errors,
which should be negligible to start with.
When a switch removes an errored frame, the originator will retransmit the
frame after a period of time. Since an errored frame arriving at its destination
network address is also discarded, many persons question the necessity of
error checking by a store-and-forward switch. However, filtering capability, if
offered, may be far more useful as you could use this capability, for example,
to route protocols carried in frames to destination ports far easier than by
bridging and switching methods and performance issues
323
frame destination address. This is especially true if you have hundreds or
thousands of devices connected to a large switch. You might set up two
or three filters instead of entering a large number of destination addresses
into the switch. When a switch performs filtering of protocols, it really
becomes a router. This is because it is now operating at layer 3 of the OSI
Reference Model.
Figure 6.20 illustrates the operation of a store-and-forward switch. Note
that a common switch design is to use shared buffer memory to store entire
frames, which increases the latency associated with this type of switch. Since
the minimum length of an Ethernet frame is 72 bytes, then the minimum
one-way delay or latency for a 10 mbps Ethernet operation, not counting
the switch overhead associated with the lookup table and switching fabric
operation, becomes:
9.6 µsec + 72 bytes ∗ 8 bits/byte ∗ 100 ns/bit
or
9.6 ∗ 10−6 + 576 ∗ 100 ∗ 10−9
or
67.2 ∗ 10−6 seconds
Once again, 9.6 µs represents the 10-Mbps Ethernet interframe gap, while
100 ns/bit is the bit duration of a 10-Mbps Ethernet LAN. Thus, the
minimum one-way latency of a store-and-forward Ethernet LAN switch is
.0000672 seconds, while a round-trip minimum latency is twice that duration.
For a maximum-length Ethernet frame with a data field of 1500 bytes, the
1
2
3
Shared
RAM
Ports
Switch
engine
Look up
table
4
Legend:
1 = Packet address read as packet enters RAM
2 = Full packet enters RAM; CRC computed
3 = Destination port obtained
4 = Packet forwarded from RAM to destination port
Figure 6.20
Store-and-forward switching.
chapter six
324
frame length becomes 1526 bytes. Thus, the one-way maximum latency at
10 Mbps becomes:
9.6 µs + 1526 bytes ∗ 8 bits/byte ∗ 100 ns/bit
or
9.6 ∗ 10−6 + 12208 ∗ 100 ∗ 10−9
or
.0123404 seconds
At a 100 Mbps operating rate associated with Fast Ethernet the interframe gap
is reduced to .96 µs while the bit duration becomes 10 ns. Thus, the one way
latency for a minimum length frame becomes:
.96 ∗ 10−6 + 576 ∗ 10 ∗ 10 10−9
or
6.72 ∗ 10−6 seconds
Similarly, for a maximum length frame the one way latency at 100
Mbps becomes:
.96 ∗ 10−6 + 12208 ∗ 10−9
or
.00123 seconds
At a Gigabit Ethernet data rate of 1 Gbps both the interframe gap and bit
duration are further reduced by an order of magnitude. This results in the one
way latency for a minimum length frame becoming .672 ∗ 10−6 seconds while
the delay for a maximum length frame becomes:
1.23 ∗ 10−4 seconds
Table 6.4 summarizes the minimum latency for cut-through and store-andforward switches operating at 10 and 100 Mbps and 1 Gbps. You can use
the entries in this table to determine the potential effect upon different
applications, such as real-time video, and select a switch that can satisfy
organizational requirements.
Hybrid A hybrid switch supports both cut-through and store-and-forward
switching, selecting the switching method based upon monitoring the error
rate encountered by reading the CRC at the end of each frame and comparing
its value with a computed CRC performed on the fly on the fields protected
by the CRC. Initially the switch might set each port to a cut-through mode
of operation. If too many bad frames are noted occurring on the port the
switch will automatically set the frame processing mode to store-and-forward,
bridging and switching methods and performance issues
325
permitting the CRC comparison to be performed before the frame being
forwarded. This permits frames in error to be discarded without having them
pass through the switch. Since the switch, no pun intended, between cutthrough and store-and-forward modes of operation occurs adaptively, another
term used to reference the operation of this type of switch is adaptive.
The major advantages of a hybrid switch are that it provides minimal
latency when error rates are low and discards frames by adapting to a storeand-forward switching method so it can discard errored frames when the frame
error rate rises. From an economic perspective, the hybrid switch can logically
be expected to cost a bit more than a cut-through or store-and-forward switch
as its software development effort is a bit more comprehensive. However, due
to the competitive market for communications products upon occasion its
price may be reduced below competitive switch technologies.
Switching Methods LAN switches can be classified with respect to their
support of single or multiple addresses per port. The support of a single
address per port is referred to as port-based switching, while the support of
multiple addresses per port is referred to as segment-based switching.
Port-Based Switching A port-based LAN switch can be considered to operate
similar to an n × n matrix switch, reading the destination address of incoming
frames from a single device connected to the port and using that address
through a table lookup process to initiate a cross-connect to a destination port.
Figure 6.21 illustrates an example of port-based switching. Since each
connected node is isolated from other nodes except when simultaneously
TABLE 6.4
Latency
Cut-through and Store-and-forward Minimum Switch
Latency (seconds)
Switch/Operating Rate
Minimum Length Frame
Minimum Length Frame
Cut-through
10 Mbps
100 Mbps
1 Gbps
20.8 ∗ 10−6
2.08 ∗ 10−6
.208 ∗ 10−6
20.8 ∗ 10−6
2.08 ∗ 10−6
.208 ∗ 10−6
Store-and-forward
10 Mbps
100 Mbps
1 Gbps
67.2 ∗ 10−6
6.72 ∗ 10−6
.672 ∗ 10−6
1.2304 ∗ 10−2
.12304 ∗ 10−2
.012304 ∗ 10−2
326
chapter six
User
1
User
N
Port
switching
Server
1
Server
2
Server
N
Figure 6.21 Port-based switching. In
port-based switching only a single
address per port is supported,
restricting switching to one device
per port.
contending for access to the same destination port, the resulting network
throughput can be considerably higher than a shared Ethernet network. For
example, user 1 could communicate with server 1, user 2 with server 2, and so
on, all simultaneously. In this best case example, with n users and n servers,
the operating rate through the switch becomes n × 10 Mbps, or n times the
operating rate of a conventional Ethernet network.
It is important to compare the maximum potential throughput through a
switch with its rated backplane speed. If the maximum potential throughput
is less than the rated backplane speed, the switch will not cause delays based
upon the traffic being routed through the device. For example, consider a
64-port switch that has a backplane speed of 400 Mbps. If the maximum
port rate is 10 Mbps, then the maximum throughput assuming 32 active
cross-connections were simultaneously established becomes 320 Mbps. In
this example the switch has a backplane transfer capability sufficient to
handle the worst-case data transfer scenario. Now let’s assume that the
maximum backplane data transfer capability was 200 Mbps. This would
reduce the maximum number of simultaneous cross-connections capable of
being serviced to 20 instead of 32 and adversely affect switch performance
under certain operational conditions.
Since a port-based switch only has to store one address per port, search times
are minimized. When combined with a pass-through or cut-through switching
technique, this type of switch results in a minimal latency to include the
overhead of the switch in determining the destination port of a frame.
Segment-Based Switching A segment-based switch permits switched connectivity between multiple LAN segments, by supporting multiple addresses
per port. The key difference between a segment-based switch and a port-based
switch is in their ability to support multiple addresses per port. A port-based
bridging and switching methods and performance issues
327
layer 2 switch only supports one MAC address per port. In comparison, a
segment-based layer 2 switch supports multiple addresses per port, enabling
it to support the direct connection of a single workstation or server as well as
the connection of a segment to each port on the switch.
Figure 6.22 illustrates an example of a segment-based switch. Note that
ports that support the connection of a segment must support switching for
multiple MAC addresses. In examining the segment-based switching example
illustrated in Figure 6.22, also note that workstations on each segment contend
for access to n servers. Since a group of users is connected to the switch via
a conventional hub, throughput is limited by the conventional hubs. In this
example, two hubs would limit throughput of the network to 20 Mbps if a
user on each segment accessed a different server. In comparison, a port-based
switch can provide a higher throughput, since each network user is directly
connected to a port on the switch. The primary advantage of a segment-based
LAN switch is cost, since a few ports can be used to support network segments
containing a large number of LAN users.
Switching Processor The construction of intelligent switches varies both
between manufacturers as well as within some vendor product lines. Most
switches are based upon the use of either reduced instruction set computer (RISC) microprocessors or application-specific integrated circuit (ASIC)
chips, while a few products use conventional complex instruction set computer (CISC) microprocessors.
Although there are a large number of arguable advantages and disadvantages associated with each architecture from the standpoint of the switch
Segment
switching
hub
Conventional
hub
User
1
User
N
Server
1
Server
N
Conventional
hub
User
1
User
N
Figure 6.22 Segment-based switching. A segment-based switching technique
requires each port to support multiple MAC addresses.
328
chapter six
manufacturer that are beyond the scope of this book, there are also some key
considerations that warrant discussion with respect to evolving technology,
such as virtual LANs. Both RISC and CISC architectures enable switches to
be programmed to make forwarding decisions based upon either the data link
layer or network layer address information. In addition, when there is a need
to modify the switch such as to enable it to support a vLAN standard, this
architecture is easily upgradable.
In comparison to RISC- and CISC-based switches, an ASIC-based device represents the use of custom-designed chips to perform specific switch functions
in hardware. Although ASIC-based switches are faster than RISC- and CISCbased switches, there is no easy way to upgrade this type of switch. Instead,
the vendor will have to design and manufacture new chips and install the
hardware upgrade in the switch.
Today most switches use an ASIC architecture, as its speed enables the
support of cut-through switching. While ASIC-based switches provide the
speed necessary to minimize latency, readers should carefully check vendor
upgrade support as support for adding load balancing and other functions to
a LAN switch can be expected to require modifications to existing switches.
Now that we have an appreciation for the general operation and use of LAN
switches, let’s obtain an appreciation for the high-speed operation of switch
ports, which enables dissimilar types of networks to be connected and which
can result in data flow compatibility problems along with methods used to
alleviate such problems.
High-Speed Port Operations There are several types of high-speed port
connections switches may support. Those high-speed connections include
100-Mbps Fast Ethernet, 1-Gbps and 10-Gbps Gigabit Ethernet, 100-Mbps
FDDI, 155-Mbps ATM, full-duplex Ethernet and Token-Ring, and fat pipes,
with the latter referencing a grouping of ports treated as a transmission entity.
Another common name used in place of the term fat pipe is trunk group. The
most common use of one or more high-speed connections on a switch is to
support highly used devices, such as network servers and printers as well as
for obtaining a backbone LAN connection capability. Figure 6.23 illustrates
the use of an Ethernet switch, with two 100BASE-T Fast Ethernet adapters
built into the switch to provide a high-speed connection from the switch to
each server. Through the use of high-speed connections the cross-connection
time from a server to client when the server responds to a client query is
minimized. Since most client queries result in server responses containing
many multiples of characters in the client query, this allows the server to
respond to more queries per unit of time. Thus, the high-speed connection
bridging and switching methods and performance issues
Server
329
Server
100BASE-T
ports
10BASE-T
ports
Figure 6.23 Using high-speed connections to servers.
can enhance client/server response times through a switch. In examining
Figure 6.23, let’s assume a small query results in the server responding by
transmitting the contents of a large file back to the client. If data flows
into the switch at 100 Mbps and flows from the switch to the client at
10 Mbps, any buffer area in the switch used to provide temporary storage for
speed incompatibilities between ports will rapidly be filled and eventually
overflow, resulting in the loss of frames which, when compensated for by
retransmission, compounds the problem. Thus, a mechanism is required to
regulate the flow of data into and out of switch ports. That mechanism is
known as flow control, and specific methods used to implement flow control
will be covered later in this chapter when we discuss switch features.
Thus for now we can note that the LAN switch is a highly versatile device
that may support single or multiple devices per port and whose operation can
vary based upon its architecture. By providing the capability for supporting
multiple simultaneous cross-connections, the LAN switch can significantly
increase network bandwidth, and its ability to support high-speed network
connections enhances its versatility. Now that we have a basic appreciation
for the operational characteristics of generic Ethernet LAN switches, we can
use that information as a base and focus our attention upon the operation of
Ethernet switches.
Switch Operations Although features incorporated into Ethernet switches
considerably differ between vendors as well as within vendor product lines,
330
chapter six
upon occasion we can categorize this communications device by the operating
rate of the ports they support. Doing so results in six basic types of Ethernet
switches which are listed in Table 6.5. Switches that are restricted to operating
at a relatively low data rate are commonly used for departmental operations,
while switches that support a mixed data rate are commonly used in a tiered
network structure at a higher layer in the tier than switches that operate at a
low uniform data rate. Concerning the latter, when used in a tiered network
structure the lower uniform operating rate switch is commonly used at the
lower level in the tier. One item that warrants attention in Table 6.5 is the
last entry in the table. At the time this book was prepared 10 Gbps ports were
restricted to one or two per switch and it could be several years before a
switch is available with ports that all operate at 10 Gbps.
Multi-Tier Network Construction
Figure 6.24 illustrates the generic use of a two-tiered Ethernet switch–based
network, with the switch at the higher tier functioning as a backbone connectivity mechanism, which enables access to shared servers commonly known
as global servers by users across departmental boundaries, while switches in
the lower tier facilitate access to servers shared within a specific department.
This hierarchical networking structure is commonly used with a higher-speed
Ethernet switch such as a Fast Ethernet or Gigabit Ethernet switch, or with
other types of backbone switches, such as FDDI and ATM, as well as with other
types of lower-tier switches. One common variation associated with the use
of a tiered switch–based network is the placement of both departmental and
global servers on an upper-tier switch. This placement allows all servers to be
colocated in a common area for ease of access and control and is commonly
referred to as a server farm. However, if an upper-tier switch should fail,
TABLE 6.5 Types of Ethernet Switches Based
upon Port Operating Rates
All ports operate at 10 Mbps.
Mixed 10-/100-Mbps port operation.
All ports operate at 100 Mbps.
Mixed 10-/100-/1000-Mbps port operation.
All ports operate at 1000 Mbps.
Mixed 10-/100-/1000/10000 Mbps port operation.
bridging and switching methods and performance issues
GS
331
GS
Switch
Higher tier
Lower tier
Switch
Switch
DS
DS
Legend:
DS = Departmental server
GS = Global server
= Workstation
Figure 6.24
network.
Generic construction of a two-tiered Ethernet switch–based
access to all servers could be affected, representing a significant disadvantage
of this design. A second major disadvantage is the fact that all traffic has to
be routed through at least two switches when a server farm is constructed.
In comparison, when servers primarily used by departmental employees are
connected to a switch serving departmental users, most traffic remains local
to the local switch at the bottom of the tier.
With the introduction of Gigabit Ethernet switches it becomes possible to
use this type of switch in either a multitier architecture as previously shown
in Figure 6.24 or as a star-based backbone. Concerning the latter, Figure 6.25
illustrates the potential use of a Gigabit Ethernet switch that supports a
mixture of 100-Mbps and 1-Gbps ports. In this example the Gigabit Ethernet
switch is shown being used to support two fat pipes or trunk groups, with
one trunk group consisting of four 100-Mbps ports, while the second group
332
chapter six
Enterprise
server
Enterprise
server
100 Mbps
1 Gbps
Fast ethernet
hub
Gigabit switch
Trunk
group
400 Mbps
Trunk group
200 Mbps
Fast ethernet
hub
Fast ethernet
hub
Departmental
server
Departmental
server
Figure 6.25
Using a Gigabit Ethernet switch as a star-based backbone switch.
consists of two 100-Mbps ports. To enable each grouping requires the use of
switches from the same vendor since the establishment of fat pipes or trunk
groups is a proprietary feature and not yet standardized.
In examining Figure 6.25 note that enterprise servers are connected to the
Gigabit switch, while department servers are connected to 100-Mbps Fast
Ethernet hubs. By connecting 10BASE-T LAN switches to Fast Ethernet hubs
you could extend the star into a star-tiered network structure.
6.5 Switch Basic Architecture
Regardless of the operating rate of each port on an Ethernet switch, most
devices are designed in a similar manner. That is, most switches consist of a
chassis into which a variety of cards are inserted, similar in many respects to
the installation of cards into the system expansion slots of personal computers.
Components
Modular Ethernet switches that are scalable commonly support CPU, logic,
matrix, and port cards.
bridging and switching methods and performance issues
333
CPU Card
The CPU card commonly manages the switch, identifies the types of LANs
attached to switch ports, and performs self and directed switch tests.
Logic Module
The logic module is commonly responsible for comparing the destination
address of frames read on a port against a table of addresses it is responsible
for maintaining, and instructing the matrix module to initiate a cross-bar
switch once a comparison of addresses results in the selection of a destination
port address.
Matrix Module
The matrix module of a switch can be considered to represent a cross-bar
of wires from each port to each port as illustrated in Figure 6.26. Upon
1
n
Port
Port n
Figure 6.26 The key to the operation of a switch is a matrix module
which enables each port to be cross-connected to other ports. The matrix
module of a switch with n ports can be considered to represent an n × n
star-wired backplane.
334
chapter six
receipt of an instruction from a logic module, the matrix module initiates a
cross-connection between the source and destination port for the duration of
the frame.
Port Module
The port module can be considered to represent a cluster of physical interfaces
to which either individual stations or network segments are connected based
upon whether the switch supports single or multiple MAC addresses per port.
Some port modules permit a mixture of port cards to be inserted, resulting
in, as an example, 10 and 100 Mbps as well as full-duplex connections to be
supported. In comparison, other port modules are only capable of supporting
one type of LAN connection. In addition, there are significant differences
between vendor port modules concerning the number of ports supported.
Some modules are limited to supporting two or four ports, while other
modules may support six, eight, or ten ports. It should be noted that many
switches support other types of LAN port modules such as Token-Ring, FDDI,
and even ATM.
Redundancy
In addition to the previously mentioned modules, most switches also support
single and redundant power supply modules and may also support redundant
matrix and logic modules. Figure 6.27 illustrates a typical Ethernet modular
switch chassis showing the installation of 11 modules to include five 8-port
modules to form a 40-port switch.
C
P
U
L
O
G
I
C
L
O
G
I
C
M
A
T
R
I
X
P
O
W
E
R
P
O
W
E
R
O
P
E
N
O
P
E
N
Figure 6.27 A typical Ethernet modular switch chassis containing a mixture
of port, CPU, logic, matrix, and power cards.
bridging and switching methods and performance issues
335
Switch Features
There are literally an ever expanding number of features being incorporated
into Ethernet switches. Those features range from providing such basic functions as port and segment switching to methods developed by some vendors
to prevent erroneous frames from being transferred through the switch in a
cut-through mode of operation. In this section we will review 19 distinct
switch features, which are summarized in alphabetical order in Table 6.6.
The table of features presented was constructed not only to list features
you should note, but, in addition, as a mechanism to facilitate the evaluation
of switches. That is, you can indicate your requirement for a particular
feature and then note whether or not that requirement can be satisfied by
different vendor products by replacing Vendor A and Vendor B by the names
of switches you are evaluating. By duplicating this table, you can extend the
two rightmost columns to evaluate more than two products. As we examine
each feature listed in Table 6.6, our degree of exploration will be based
upon whether or not the feature was previously described. If the feature
was previously described in this chapter, we will limit our discussion to
a brief review of the feature. Otherwise we will discuss its operation in
considerable detail.
Address Table Size Support
The ability of a switch to correctly forward packets to their intended direction
depends upon the use of address tables. Similarly, the capability of a switch
to support a defined number of workstations depends upon the number of
entries that can be placed in its address table. Thus, the address table size
can be viewed as a constraint that affects the ability of a switch to support
network devices.
There are two address table sizes you may have to consider — the number
of addresses supported per port and the number of addresses supported per
switch. The first address table constraint is only applicable for ports that
support the connection of network segments. In comparison, the total number
of addresses recognized per switch represents a constraint that affects the
entire switch. Many Ethernet switches support up to 1024 addresses per port
for segment-based support. Such switches may only support a total of 8192
addresses per switch. This means that a 16-port switch with 8 fully populated
segments could not support the use of the 8 remaining ports as the switch
would run out of address table entries. Thus, it is important to consider the
number of addresses supported per port and per switch as well as match such
data against your anticipated requirements.
336
chapter six
TABLE 6.6
Ethernet Switch Features
Feature
Address table size support
Addresses/port
Addresses/switch
Aging settings
Architecture
ASIC-based
CISC-based
RISC-based
Autonegotiation ports
Backplane transfer capacity
Error prevention
Fat pipe and trunk group
Filtering/forwarding rate support
Filtering rate
Forwarding rate
Flow control
Backpressure
Software drivers
802.3x flow control
No control
Full-duplex port operation
Jabber control
Latency
Management
Mirrored port
Module insertion
Port buffer size
Port module support
Spanning tree support
Switch type
Port-based switch
Segment-based switch
Requirement
Vendor A
Vendor B
bridging and switching methods and performance issues
TABLE 6.6
337
(Continued)
Feature
Requirement
Vendor A
Vendor B
Switching mode
Cut-through
Store-and-forward
Hybrid
Virtual LAN support
Port-based
MAC-based
Layer 3–based
Rule-based
Aging Settings
As explained earlier in this book when we discussed the operation of bridges,
MAC addresses and their associated ports are stored with a time stamp value.
This provides the bridge with the ability to purge old entries to make room
for new entries in the port-address table. Since a switch is a multiport bridge,
it also uses a timer to purge old entries from its port-address table. Some
switches provide users with the ability to set the aging time within a wide
range of values or to disable aging. Other switches have a series of predefined
aging values from which a user can select.
Architecture
As previously noted, there are three basic methods used to construct LAN
switches. Those methods include the use of application-specific integrated
circuits (ASICs), complex or conventional instruction set computers (CISCs),
and reduced instruction set computers (RISCs). Although the use of an ASICbased architecture commonly results in a very low latency and high level of
performance, upgrades are difficult to accomplish as such circuits must be
replaced. In comparison, both conventional microprocessor and RISC-based
switches use instructions in replaceable ROM. Although the differences may
appear trivial, if an ASIC-based switch, for example, requires an upgrade you
would probably require a technician to visit your site to perform the upgrade.
In comparison, you might be able to simply download a new software release
from the vendor’s World Wide Web site or electronic bulletin board to update
a RISC- or CISC-based switch.
338
chapter six
Autonegotiation Ports
To provide a mechanism to migrate from 10 Mbps to 100 Mbps, National
Semiconductor developed a chip set known as NWay, which provides an
automatic data rate sensing capability as part of an autonegotiation function.
This capability enables a switch port to support either a 10- or 100-Mbps
Ethernet attachment to the port; however, this feature only works when
cabled to a 10-/100-Mbps network adapter card. You may otherwise have to
use the switch console to configure the operating rate of the port or the port
may be fixed to a predefined operating rate.
Backplane Transfer Capacity
The backplane transfer capacity of a switch provides you with the ability to
determine how well the device can support a large number of simultaneous
cross-connections, as well as its ability to perform flooding. For example,
consider a 64-port 10BASE-T switch with a backplane transfer capacity of
400 Mbps. Since the switch can support a maximum of 64/2 or 32 cross
connects, the switch’s backplane must provide at least a 32 ∗ 10 Mbps or
320 transfer capacity. However, when it encounters an unknown destination
address on one port, the switch will output or flood the packet onto all
ports other than the port the frame was received on. Thus, to operate in a
nonblocked mode the switch must have a buffer transfer capacity of 64 ∗
10 Mbps or 640 Mbps.
Error Prevention
Some switch designers recognize that the majority of runt frames (frames
improperly terminated) result from a collision occurring during the time it
takes to read the first 64 bytes of the frame. On a 10-Mbps Ethernet LAN
this is equivalent to a time of 51.2 µs. In a cut-through switch environment
when latency is minimized, it becomes possible to pass runt frames to the
destination. To preclude this from happening, some switch designers permit
the user to introduce a 51.2-µs delay, which provides sufficient time for the
switch to verify that the frame is of sufficient length to have a high degree
of probability that it is not a runt frame. Other switches that operate in the
cut-through mode may simply impose a 51.2-µs delay at 10 Mbps to enable
this error prevention feature. Regardless of the method used, the delay is only
applicable to cut-through switches that support LAN segments, as single user
ports do not generate collisions.
bridging and switching methods and performance issues
339
Fat Pipe and Trunk Group
A fat pipe is a term used to reference a high-speed port. When 10BASE-T
switches were first introduced, the term actually referenced a group of two or
more ports operating as an entity. Today a fat pipe can reference a 100-Mbps
port on a switch primarily consisting of 10-Mbps operating ports or a 1-Gbps
or 10-Gbps port on a switch that contains a mixture of lower-speed and Gigabit
or 10 Gigabit Ethernet ports. In addition, some vendors retain the term fat pipe
as a reference to a group of ports operating as an entity while other vendors
use the term trunk group to represent a group of ports that function as an
entity. However, to support a grouping of ports operating as a common entity
requires the interconnected switches to be obtained from the same company
as the method used to group ports is currently proprietary.
Filtering and Forwarding Rate Support
The ability of a switch to interpret a number of frame destination addresses
during a defined time interval is referred to as its filtering rate. In comparison,
the number of frames that must be routed through a switch during a predefined
period of time is referred to as the forwarding rate. Both the filtering and
forwarding rates govern the performance level of a switch with respect to its
ability to interpret and route frames. When considering these two metrics, it is
important to understand the maximum frame rate on an Ethernet LAN, which
was discussed earlier in this book.
Flow Control
Flow control represents the orderly regulation of transmission. In a switched
network environment, there are a number of situations for which flow control
can be used to prevent the loss of data and subsequent retransmissions, which
can create a cycle of lost data followed by retransmissions. The most common
cause of lost data results from a data rate mismatch between source and
destination ports. For example, consider a server connected to a switch via a
Fast Ethernet 100-Mbps connection, which responds to a client query when
the client is connected to a switch port at 10 Mbps. Without the use of a buffer
within the switch, this speed mismatch would always result in the loss of
data. Through the use of a buffer, data can be transferred into the switch at
100 Mbps and transferred out at 10 Mbps. However, since the input rate is
10 times the output rate, the buffer will rapidly fill. In addition, if the server
is transferring a large quantity of data the buffer could overflow, resulting
in subsequent data sent to the switch being lost. Thus, unless the length of
340
chapter six
the buffer is infinite, an impossible situation, there would always be some
probability that data could be lost.
Another common cause of lost data is when multiple source port inputs
are contending for access to the same destination port. If each source and
destination port operates at the same data rate, then only two source ports
contending for access to the same destination port can result in the loss of
data. Thus, a mechanism is required to regulate the flow of data through a
switch. That mechanism is flow control.
All Ethernet switches this author is familiar with have either buffers in each
port, or centralized memory that functions as a buffer. The key difference
between switch buffers is in the amount of memory used. Some switches have
128K, 256 Kbytes, or even 1 or 2 Mbytes per port, whereas other switches
may support the temporary storage of 10 or 20 full-length Ethernet frames. To
prevent buffer overflow four techniques are used — backpressure, proprietary
software, IEEE 802.3x flow control, and no control. Thus, let’s examine
each technique.
Backpressure Backpressure represents a technique by which a switch generates a false collision signal. In actuality, the switch port operates as if it
detected a collision and initiates the transmission of a jam pattern. The jam
pattern consists of 32 to 48 bits that can have any value other than the CRC
value that corresponds to any partial frame transmitted before the jam.
The transmission of the jam pattern ensures that the collision lasts long
enough to be detected by all stations on the network. In addition, the jam
signal serves as a mechanism to cause nontransmitting stations to wait until
the jam signal ends before attempting to transmit, alleviating additional
potential collisions from occurring. Although the jam signal temporarily stops
transmission, enabling the contents of buffers to be output, the signal also
adversely affects all stations connected to the port. Thus, a network segment
consisting of a number of stations connected to a switch port would result in
all stations having their transmission capability suspended, even when just
one station was directing traffic to the switch.
Backpressure is commonly implemented based upon the level of buffer
memory used. When buffer memory is filled to a predefined level, that level
serves as a threshold for the switch to generate jam signals. Then, once the
buffer is emptied beyond another lower level, that level serves as a threshold
to disable backpressure operations.
Proprietary Software Drivers Software drivers enable a switch to directly
communicate with an attached device. This enables the switch to enable and
bridging and switching methods and performance issues
341
disable the station’s transmission capability. Currently, software drivers are
available as a NetWare Loadable Module (NLM) for NetWare servers, and may
be available for Windows XP by the time you read this book.
IEEE 802.3x Flow Control During 1997 the IEEE standardized a method that
provides flow control on full-duplex Ethernet connections. To provide this
capability, a special pause frame was defined that is transmitted by devices
that want to stop the flow of data from the device at the opposite end of
the link.
Figure 6.28 illustrates the format of the pause frame. Since a full-duplex
connection has a device on each end of the link, the use of a predefined
destination address and operation code (OpCode) defines the frame as a pause
frame. The value following the OpCode defines the time in terms of slot times
that the transmitting device wants its partner to pause. This initial pause time
can be extended by additional pause frames containing new slot time values
or canceled by another pause frame containing a zero slot time value. The
PAD field shown at the end of the pause frame must have each of its 42 bytes
set to zero.
Under the 802.3x standard, the use of pause frames is autonegotiated on
copper media and can be manually configured for use on fiber links. The
actual standard does not require a device capable of sending a pause frame to
actually do so. Instead, it provides a standard for recognizing a pause frame
as well as a mechanism for interpreting the contents of the frame so a receiver
can correctly respond to it.
The IEEE 802.3x flow control standard is applicable to all versions of
Ethernet from 10 Mbps to 10 Gbps; however, the primary purpose of this
standard is to enable switches to be manufactured with a minimal amount
of memory. By supporting the IEEE 802.3x standard, a switch with a limited
amount of memory can generate pause frames to regulate inbound traffic
instead of having to drop frames when its buffer is full.
No Control Some switch vendors rely upon the fact that the previously
described traffic patterns that can result in buffers overflowing and the loss of
Destination
address
01−C2−80−00−00−01
(6 bytes)
Source
address
(6 bytes)
Figure 6.28
Type
8808
(2 bytes)
OpCode
0001
(2 bytes)
Pause.time
(slot times)
(2 bytes)
The IEEE 802.3x pause frame.
PAD
(42
bytes)
342
chapter six
data have a relatively low probability of occurrence for any significant length
of time. In addition, upper layers of the OSI Reference Model will retransmit
lost packets. Thus, some switch vendors rely upon the use of memory buffers
and do not incorporate flow control into their products. Whether or not this
is an appropriate solution will depend upon the traffic you anticipate flowing
through the switch.
Full-Duplex Port Operation
If a switch port only supports the connection of one station, a collision
can never occur. Recognizing this fact, most Ethernet switch vendors now
support full-duplex or bidirectional traffic flow by using two of the four wire
connections for 10BASE-T for transmission in the opposite direction. Fullduplex support is available for 10BASE-T, Fast Ethernet, and Gigabit Ethernet
connections. Since collisions can occur on a segment, switch ports used for
segment-based switching cannot support full-duplex transmission.
In addition to providing a simultaneous bidirectional data flow capability,
the use of full duplex permits an extension of cabling distances. This extension
of cabling distance becomes possible since the CSMA/CD mode of operation
is no longer followed. This means we do not need to be concerned about the
number of bit times in a collision diameter. Thus, the governing parameter
is not the distance a signal can be driven prior to attenuation and distortion
rending the signal unrecognizable. For example, at 100 Mbps the use of a fiber
cable for full-duplex operations can support a distance of 2000 meters, while
only 412 meters is supported using half-duplex transmission via fiber.
Due to the higher cost of full-duplex ports and adapter cards, you should
carefully evaluate the potential use of FDX before using this capability. For
example, most client workstations will obtain a minimal gain through the use
of a full-duplex capability since humans operating computers rarely perform
simultaneous two-way operations. Thus, other than speeding acknowledgments associated with the transfer of data, the use of an FDX connection for
workstations represents an excessive capacity that should only be considered
when vendors are competing for sales, and as such, they provide this capability as a standard. In comparison, the use of an FDX transmission capability
to connect servers to switch ports enables a server to respond to one request
while receiving a subsequent request. Thus, the ability to use the capability
of FDX transmission is enhanced by using this capability on server-to-switch
port connections.
Although vendors would like you to believe that FDX doubles your transmission capability, in actuality you will only obtain a fraction of this advertised
bridging and switching methods and performance issues
343
throughput. This is because most network devices, to include servers that
are provided with an FDX transmission capability, only use that capability a
fraction of the time.
Jabber Control
A jabber is an Ethernet frame whose length exceeds 1518 bytes. Jabbers are
commonly caused by defective hardware or collisions, and can adversely
affect a receiving device by its misinterpretation of data in the frame. A
switch operating in the cut-through mode with jabber control will truncate
the frame to an appropriate length. In comparison, a store-and-forward switch
will normally automatically drop a jabbered frame.
Latency
When examining vendor specifications, the best word of advice is to be
suspicious of latency notations, especially those concerning store-and-forward
switches. Many vendors do not denote the length of the frame used for latency
measurements, while some vendors use what might be referred to as creative
accounting when computing latency. Thus, let’s review the formal definition
of latency. Latency can be defined as the difference in time (t) from the first
bit arriving at a source port to the first bit output on the destination port.
Modern cut-through switches have a latency of approximately 40 µs, while
store-and-forward switches have a latency between 80 and 90 ms for a 72-byte
frame, and between 1250 and 1300 ms for a maximum-length 1500-byte frame.
For a store-and-forward Ethernet switch an entire frame is first stored. Since
the maximum length of an Ethernet frame is 1526 bytes, this means that the
maximum latency for a store-and-forward 10-Mbps Ethernet switch is:
1526 bytes ∗ 8 bits/byte
10 Mbps
or 1.2208 ms
plus the time required to perform a table lookup and cross-connection between
source and destination ports. Since a 10-Mbps Ethernet LAN has a 9.6-µs gap
between frames, this means that the minimum delay time between frames
flowing through a cut-through switch is 20.8 µs. At 100 Mbps the total cutthrough delay would be 2.08 µs, while at 1 Gbps the total delay would be
0.208 µs. Figure 6.29 illustrates the composition of this delay at a 10-Mbps
operating rate. For a store-and-forward 10BASE-T switch, considering the 9.6µs gap between frames results in a maximum latency of 12304 µs plus the time
required to perform a table lookup and initiate a cross-connection between
344
chapter six
PREAMBLE
9.6 µs
gap between
frames
8 bytes
DESTINATION
ADDRESS
6 bytes
Rest of frame
11.2 µs
Minimum latency or
delay between frames
20.8 µs
Figure 6.29 Switch latency includes a built-in delay resulting from the
structure of the Ethernet frame.
source and destination ports. As noted earlier in this section, at 100 Mbps
the maximum latency would be one-tenth, or 1230.4 µs, while at 1 Gbps the
delay would be reduced by another order of magnitude to 123.04 µs.
Management
The most common method used to provide switch management involves the
integration of RMON support for each switch port. This enables an SNMP
console to obtain statistics from the RMON group or groups supported by
each switch. Since the retrieval of statistics on a port-by-port basis can be
time-consuming, most switches that support RMON also create a database of
statistics to facilitate their retrieval.
Mirrored Port
A mirrored port is a port that duplicates traffic on another port. For traffic
analysis and intrusive testing, the ability to mirror data exiting the switch
on one port to a port to which test equipment is connected can be a very
valuable feature.
Module Insertion
Modular switches support two different methods of module insertion — switch
power down and hot. As their names imply, a switch power down method
requires you to first deactivate the switch and literally bring it down. In
comparison, the ability to perform hot insertions enables you to add modules
to an operating switch without adversely affecting users.
bridging and switching methods and performance issues
345
Port Buffer Size
Switch designers incorporate buffer memory into port cards as a mechanism
to compensate for the difference between the internal speed of the switch and
the operating rate of an end station or segment connected to the port. Some
switch designers increase the amount of buffer memory incorporated into port
cards to use in conjunction with a flow control mechanism, while other switch
designers may use port buffer memory as a substitute for flow control. If used
only as a mechanism for speed compensation, the size of port buffers may be
limited to a few thousand bytes of storage. When used in conjunction with a
flow control mechanism or as a flow control mechanism, the amount of buffer
memory per port may be up to 64, 128, or 256 Kbytes, perhaps even up to 1 or
2 Mbytes. Although you might expect more buffer memory to provide better
results, this may not necessarily be true. For example, assume a workstation
on a segment is connected to a port that has a large buffer with just enough
free memory to accept one frame. When the workstation transmits a sequence
of frames only the first is able to be placed into the buffer. If the switch then
initiates flow control as the contents of its port buffer is emptied, subsequent
frames are barred from moving through the switch. When the switch disables
flow control, it is possible that another station with data to transmit is able
to gain access to the port before the station that sent frame one in a sequence
of frames. Due to the delay in emptying the contents of a large buffer, it
becomes possible that subsequent frames are sufficiently delayed as they
move through a switch to a mainframe via a gateway that a time-dependent
session could time out. Thus, you should consider your network structure
in conjunction with the operating rate of switch ports and the amount of
buffer storage per port to determine if an extremely large amount of buffer
storage could potentially result in session time-outs. Fortunately, most switch
manufacturers limit port buffer memory to 128 Kbytes, which at 10 Mbps
results in a maximum delay of
128 ∗ 1024 ∗ 8 bits/byte
10 Mbps
or .10 seconds. At 100 Mbps, the maximum delay is reduced to .01 second,
while at 1 Gbps the delay becomes .001 second.
Port Module Support
Although many Ethernet switches are limited to supporting only Ethernet
networks, the type of networks supported can considerably differ between
346
chapter six
vendor products as well as within a specific vendor product line. Thus, you
may wish to examine the support of port modules for connecting 10BASE-2,
10BASE-5, 10BASE-T, 100BASE-T LANs, and Gigabit and 10 Gigabit Ethernet
devices. In addition, if your organization supports different types of LANs or
is considering the use of switches to form a tier-structured network using a
different type of high-speed backbone, you should examine port support for
FDDI, full-duplex Token-Ring, and ATM connectivity. Many modular Ethernet
switches include the ability to add translating bridge modules, enabling
support for several different types of networks through a common chassis.
Spanning Tree Support
Ethernet networks use the spanning tree algorithm to prevent loops that, if
enabled, could result in the continuous replication of frames. In a bridged
network, spanning tree support is accomplished by the use of bridge protocol
data units (BPDUs), which enable bridges to select a root bridge and agree
upon a network topology that preclude loops from occurring. Since a switch
in effect represents a sophisticated bridge, we would want to preclude the use
of multiple switches from forming a loop. For example, consider Figure 6.30,
which illustrates the use of two switches to interconnect two LANs. If both
switches were active, a frame from the client connected on LAN A destined
to the server on LAN B would be placed back onto LAN A by switch S1,
causing a packet loop and the replication of the packet, a situation we want
to avoid. By incorporating spanning tree support into each switch, they can
communicate with one another to construct a topology that does not contain
loops. For example, one switch in Figure 6.30 would place a port in a blocking
Client
LAN A
Switch
1
Switch
2
LAN B
Server
Figure 6.30
The need for loop control.
bridging and switching methods and performance issues
347
mode while the other switch would have both ports in a forwarding mode
of operation.
To obtain the ability to control the spanning tree, most switches permit a
number of parameters to be altered from their management console. Those
parameters include the forwarding delay that governs the time the switch
will wait before forwarding a packet, the aging time the switch waits for the
receipt of a hello packet before initiating a topology change, the Hello time
interval between the transmission of BPDU frames, and the path cost assigned
to each port.
Switch Type
As previously discussed, a switch will either support one or multiple
addresses per port. If it supports one address per port, it is a port-based
switch. In comparison, if it supports multiple addresses per switch, it is
considered to be a segment-based switch, even if only one end station is
connected to some or all ports on the switch.
Switching Mode
Ethernet switches can be obtained to operate in a cut-through, store-andforward, or hybrid operating mode. As previously discussed in this section,
the hybrid mode of operation represents toggling between cut-through and
store-and-forward based upon a frame error rate threshold. That is, a hybrid
switch might initially be set to operate in a cut-through mode and compute
the CRC for each frame on-the-fly, comparing its computed values with the
CRCs appended to each frame. When a predefined frame error threshold is
reached, the switch would change its operating mode to store-and-forward,
enabling erroneous frames to be discarded. Some switch vendors reference a
hybrid switch mode as an error-free cut-through operating mode.
Virtual LAN Support
A virtual LAN can be considered to represent a broadcast domain created
through the association of switch ports, MAC addresses, or a network layer
parameter. Thus, there are three basic types of vLAN creation methods you can
evaluate when examining the functionality of an Ethernet switch. In addition,
some vendors now offer a rules-based vLAN creation capability, which enables
users to have an almost infinite number of vLAN creation methods with the
ability to go down to the bit level within a frame as a mechanism for vLAN
associations. Although port-based vLANs were standardized by the IEEE
348
chapter six
under the 802.1Q specification during the late 1990s, other vLAN creation
methods currently represent proprietary vendor-specific solutions.
Switched-Based Virtual LANs
As briefly mentioned in our review of switch features, a virtual LAN or
vLAN can be considered to represent a broadcast domain. This means that
transmission generated by one station assigned to a vLAN is only received
by those stations predefined by some criteria to be in the domain. Thus,
to understand how vLANs operate requires us to examine how they are
constructed.
Construction Basics
A vLAN is constructed by the logical grouping of two or more network nodes
on a physical topology. To accomplish this logical grouping you must use a
vLAN-aware switching device. Those devices can include intelligent switches,
which essentially perform bridging and operate at the MAC layer, or routers,
which operate at the network layer, or layer 3, of the OSI Reference Model.
Although a switching device is required to develop a vLAN, in actuality it is
the software used by the device that provides you with a vLAN capability.
That is, a vLAN represents a subnetwork or broadcast domain defined by
software and not by the physical topology of a network. Instead, the physical
topology of a network serves as a constraint for the software-based grouping
of nodes into a logically defined network.
Implicit versus Explicit Tagging The actual criteria used to define the
logical grouping of nodes into a vLAN can be based upon implicit or explicit
tagging. Implicit tagging, which in effect eliminates the use of a special tagging
field inserted into frames or packets, can be based upon MAC address, port
number of a switch used by a node, protocol, or another parameter that nodes
can be logically grouped into. Since many vendors offering vLAN products
use different construction techniques, interoperability between vendors may
be difficult, if not impossible. In comparison, explicit tagging requires the
addition of a field into a frame or packet header. This action can result in
incompatibilities with certain types of vendor equipment as the extension of
the length of a frame or packet beyond its maximum can result in the inability
of such equipment to handle such frames or packets. As noted in Chapter 4
when we examined different types of Ethernet frames, under the IEEE 802.1Q
standard a four-byte field is inserted into the frame header behind the source
address field. This field contains a two-byte tag protocol identifier that is set
bridging and switching methods and performance issues
349
to a value of hex 8100 and three additional subfields. A 3-bit priority subfield
enables eight levels of priority to be assigned to a frame and permits 802.1p
compliant switches and routers to place prioritized traffic into predefined
queues as a mechanism to expedite traffic. A 1-bit canonical format identifier
subfield when set indicates that a Token-Ring frame is being transported
encapsulated within an Ethernet frame. The last subfield is a 12-bit vLAN ID
field. This field contains a value that identifies the vLAN to which the frame
belongs. After we examine the generic operation of port-based vLANs, we will
focus our attention upon the 802.1Q operations.
Port-Grouping vLANs
As its name implies, a port-grouping vLAN represents a virtual LAN created
by defining a group of ports on a switch or router to form a broadcast domain.
Thus, another common name for this type of vLAN is a port-based virtual LAN.
Operation Figure 6.31 illustrates the use of a LAN switch to create two
vLANs based upon port groupings. In this example the switch was configured
to create one vLAN consisting of ports 0, 1, 5, and 6, while a second vLAN
was created based upon the grouping of ports 2, 3, 4, and 7 to form a second
broadcast domain.
0
1
2
3
Switch
matrix
4
Legend :
5
n
6
7
= port n
= network segment
vLAN1 = ports 0, 1, 5, 6
vLAN2 = ports 2, 3, 4, 7
Figure 6.31 Creating port-grouping vLANs using
a LAN switch.
chapter six
350
Advantages associated with the use of LAN switches for creating vLANs
include the ability to use the switching capability of the switch, the ability
to support multiple stations per port, and intranetworking capability. A key
disadvantage associated with the use of a port-based vLAN is the fact they
are commonly limited to supporting one vLAN per port. This means that
moves from one vLAN to another affect all stations connected to a particular
switch port.
Supporting Inter-vLAN Communications The use of multiple NICs provides
an easy-to-implement solution to obtaining an inter-vLAN communications
capability when only a few vLANs must be linked. This method of inter-vLAN
communications is applicable to all methods of vLAN creation; however,
when a built-in routing capability is included in a LAN switch, you would
probably prefer to use the routing capability rather than obtain and install
additional hardware.
Figure 6.32 illustrates the use of a server with multiple NICs to provide
support to two port-based vLANs. Not only does this method of multiple
vLAN support require additional hardware and the use of multiple ports on
a switch or wiring hub, but, in addition, the number of NICs that can be
installed in a station is typically limited to two or three. Thus, the use of a
large switch with hundreds of ports configured for supporting three or more
vLANs may not be capable of supporting inter-vLAN communications unless
a router is connected to a switch port for each vLAN on the switch.
IEEE 802.1Q Operations
When the IEEE developed the 802.1Q specification for supporting port-based
vLANs, it recognized that vLAN aware switches would have to interoperate
Server
vLAN1
vLAN2
0
1
2
3
4
5
6
7
Figure 6.32 Overcoming the port-based
constraint where stations can only join a
single vLAN. By installing multiple network
adapter cards in a server or workstation,
a LAN device can become a member of
multiple vLANs.
bridging and switching methods and performance issues
351
with ‘‘legacy’’ devices that are not aware of vLANs. Due to this, the 802.1Q
specification provides support for both tagged and untagged frames, with each
type of frame associated with different vLANs.
Initially, all switch ports in an 802.1Q environment belonged to a single
port-based vLAN referred to as a port vLAN ID (PVID). The PVID has a
numeric value, with a default of 1. Any untagged frame that enters the switch
generated by a non-aware vLAN device would thus become a member of the
vLAN identified by the PVID for the port through which the frame entered the
switch. If the frame was generated by a vLAN-aware network adapter card,
it would contain a vLAN tag in its header that would identify the vLAN to
which the frame belongs. That value is the vLAN ID or VID.
Each switch port can have one or more VIDs. Those VIDs identify all of
the vLANs that a specific port is a member of. Thus, when a frame enters a
switch port it is identified as belonging to a vLAN either by the VID within
its frame or via the port on which the frame entered the switch. The switch
then consults its vLAN-port table and forwards the frame onto all ports that
correspond to the VID.
Figure 6.33 illustrates an example of the manner by which an 802.1Q aware
LAN switch could be configured to support tagged and untagged frames. In
this example assume the workstation UT transmits an untagged frame into
the switch on port 0. By default the PVID value of 1 is used to tag the frame,
802.1Q aware LAN switch
= 2
= 1
VID
PVID
= 1
= 2
= 1
VID
VID
PVID
0
1
UT
T
= 4
= 6
= 2
VID
VID
PVID
2
= 2
= 5
= 2
VID
VID
PVID
3
Legend:
UT
untagged
T
tagged VID = 2
Figure 6.33
IEEE 802.1Q PVID and VID illustrative example.
352
chapter six
resulting in it being forwarded to port 1. Now let’s assume station T transmits
a tagged frame with a VID value of 2 into the switch on port 1. In this example,
the frame would be forwarded onto ports 0 and 3.
MAC-Based vLANs
Figure 6.34 illustrates the use of an 18-port switch to create two virtual LANs.
In this example, 18 devices are shown connected to the switch via six ports,
with four ports serving individual network segments. Thus, the LAN switch
in this example is more accurately referenced as a segment switch with a
vLAN1
vLAN2
Server
Server
17
18
4
5
LAN
2
2
1
0
1
Switch
5
6
3
4
7
3
8
13
9
10
11
14
15
16
12
Legend:
n = Port n
n = MAC address
Figure 6.34 Layer 2 vLAN. A layer 2 vLAN uses MAC addresses to construct
broadcast domains that form a virtual LAN.
bridging and switching methods and performance issues
353
MAC or layer 2 vLAN capability. This type of switch can range in capacity
from small 8- or 16-port devices capable of supporting segments with up to
512 or 1024 total addresses to large switches with hundreds of ports capable
of supporting thousands of MAC addresses. For simplicity of illustration we
will use the 6-port segment switch to denote the operation of layer 2 vLANs
as well as their advantages and disadvantages.
In turning our attention to the vLANs shown in Figure 6.34, note that
we will use the numeric or node addresses shown contained in circles as
MAC addresses for simplicity of illustration. Thus, addresses 1 through 8 and
17 would be grouped into a broadcast domain representing vLAN1, while
addresses 9 through 16 and 18 would be grouped into a second broadcast
domain to represent vLAN2. At this point in time you would be tempted
to say ‘‘so what,’’ as the use of MAC addresses in creating layer 2 vLANs
resembles precisely the same effect as if you used a port-grouping method
of vLAN creation. For example, using a LAN switch with vLAN creation
based upon port grouping would result in the same vLANs as those shown in
Figure 6.34 when ports 0, 1, and 4 are assigned to one vLAN and ports 2, 3,
and 5 to the second.
To indicate the greater flexibility associated with the use of equipment
that supports layer 2 vLAN creation, let’s assume users with network node
addresses 7 and 8 were just transferred from the project associated with
vLAN1 to the project associated with vLAN2. If you were using a portgrouping method of vLAN creation, you would have to physically recable
nodes 7 and 8 to either the segment connected to port 2 or the segment
connected to port 3. In comparison, when using a segment switch with a
layer 2 vLAN creation capability, you would use the management port to
delete addresses 7 and 8 from vLAN1 and add them to vLAN2. The actual
effort required to do so might be as simple as dragging MAC addresses from
one vLAN to the other when using a graphical user interface (GUI) to entering
one or more commands when using a command line management system.
The top of Figure 6.35 illustrates the result of the previously mentioned node
transfer. The lower portion of Figure 6.35 shows the two vLAN layer 2 tables,
indicating the movement of MAC addresses 7 and 8 to vLAN2.
Although the reassignment of stations 7 and 8 to vLAN2 is easily accomplished at the MAC layer, it should be noted that the partitioning of a segment
into two vLANs can result in upper-layer problems. This is because upperlayer protocols, such as IP, normally require all stations on a segment to have
the same network address. Some switches overcome this problem by dynamically altering the network address to correspond to the vLAN on which the
station resides. Other switches without this capability restrict the creation of
354
chapter six
vLAN1
vLAN2
Server
Server
17
18
4
5
LAN
0
1
2
Switch
5
6
3
4
3
2
1
7
8
13
9
10
11
14
15
16
12
Legent:
n = Port n
n = MAC address n
vLAN1 = 1, 2, 3, 4, 5, 6, 17
vLAN2 = 7, 8, 9, 10, 11,12,13,14,15, 16, 18,
Figure 6.35
Moving stations when using a layer 2 vLAN.
MAC-based vLANs to one device per port, in effect limiting the creation of
vLANs to port-based switches.
Interswitch Communications Similar to the port-grouping method of vLAN
creation, a MAC-based vLAN is normally restricted to a single switch; however, some vendors include a management platform that enables multiple
switches to support MAC addresses between closely located switches. Unfortunately, neither individual nor closely located switches permit an expansion
of vLANs outside of the immediate area, resulting in the isolation of the
bridging and switching methods and performance issues
355
virtual LANs from the rest of the network. This deficiency can be alleviated in
two ways. First, for inter-vLAN communications you could install a second
adapter card in a server and associate one MAC address with one vLAN while
the second address is associated with the second vLAN. While this method
is appropriate for a switch with two vLANs, you would require a different
method to obtain interoperability when communications are required between
a large number of virtual LANs. Similar to correcting the interoperability problem with the port-grouping method of vLAN creation, you would have to use
routers to provide connectivity between MAC-based vLANs and the rest of
your network.
Router Restrictions When using a router to provide connectivity between
vLANs, there are several restrictions you must consider. Those restrictions
typically include a requirement to use a separate switch port connection to the
router for each virtual LAN and the inability to assign portions of segments
to different vLANs. Concerning the former, unless the LAN switch either
internally supports layer 3 routing or provides a trunking or aggregation
capability that enables transmission from multiple vLANs to occur on a
common port to the router, one port linking the switch to the router will
be required for each vLAN. Since router and switch ports are relatively
costly, intranetworking of a large number of vLANs can become expensive.
Concerning the latter, this requirement results from the fact that in a TCP/IP
environment routing occurs between segments. An example of inter-vLAN
communications using a router is illustrated in Figure 6.35.
When inter-vLAN communications are required, the layer 2 switch transmits
packets to the router via a port associated with the virtual LAN workstation
requiring such communications. The router is responsible for determining
the routed path to provide inter-vLAN communications, forwarding the
packet back to the switch via an appropriate router-to-switch interface. Upon
receipt of the packet the switch uses bridging to forward the packet to its
destination port.
Returning to Figure 6.36, a workstation located in vLAN1 requiring communications with a workstation in vLAN2 would have its data transmitted by the
switch on port 5 to the router. After processing the packet the router would
return the packet to the switch, with the packet entering the switch on port 6.
Thereafter, the switch would use bridging to broadcast the packet to ports 2,
3, and 7 where it would be recognized by a destination node in vLAN2 and
copied into an appropriate NIC.
356
chapter six
Router
vLAN1
vLAN2
Server
Server
17
18
4
1
Figure 6.36
2
Hub
1
0
5
6
3
4
7
6
5
Switching
3
2
7
8
13
9
10
11
14
15
16
12
Inter-vLAN communications require the use of a router.
Layer 3–Based vLANs
A layer 3–based vLAN is constructed using information contained in the
network layer header of packets. As such, this precludes the use of LAN
switches that operate at the data link layer from being capable of forming
layer 3 vLANs. Thus, layer 3 vLAN creation is restricted to routers and LAN
switches that provide a layer 3 routing capability.
Through the use of layer 3 operating switches and routers, there are a
variety of methods that can be used to create layer 3 vLANs. Some of the
more common methods supported resemble the criteria by which routers
operate, such as IPX network numbers and IP subnets, AppleTalk domains,
and layer 3 protocols.
The actual creation options associated with a layer 3 vLAN can vary
considerably based upon the capability of the LAN switch or router used
bridging and switching methods and performance issues
357
to form the vLAN. For example, some hardware products permit a subnet to
be formed across a number of ports and may even provide the capability to
allow more than one subnet to be associated with a network segment connected to the port of a LAN switch. In comparison, other LAN switches may
be limited to creating vLANs based upon different layer 3 protocols.
Subnet-Based vLANs Figure 6.37 illustrates the use of a layer 3 LAN switch
to create two vLANs based upon IP network addresses. In examining the
vLANs created through the use of the LAN switch, note that the first vLAN is
associated with the subnet 198.78.55, which represents a Class C IP address,
while the second vLAN is associated with the subnet 198.78.42, which
represents a second Class C IP address. Also note that since it is assumed that
the LAN switch supports the assignment of more than one subnet per port,
port 1 on the switch consists of stations assigned to either subnet. While some
LAN switches support this subnetting capability, it is also important to note
that other switches do not. Thus, a LAN switch that does not support multiple
subnets per port would require stations to be recabled to other ports if it was
desired to associate them to a different vLAN.
198.78.55.XXX
198.78.42.XXX
Server
Server
5
4
LAN Switch
0
1
2
3
198.78.55.XXX 198.78.42.XXX
198.78.42.XXX
198.78.55.XXX
198.78.55.XXX
vLAN1 = Subnet 198.78.55
vLAN2 = Subnet 198.78.42
Figure 6.37
vLAN creation based upon IP subnets.
358
chapter six
Protocol-Based vLANs In addition to forming vLANs based upon a network
address, the use of the layer 3 transmission protocol as a method for vLAN
creation provides a mechanism that enables vLAN formation to be based
upon the layer 3 protocol. Through the use of this method of vLAN creation,
it becomes relatively easy for stations to belong to multiple vLANs. To
illustrate this concept, consider Figure 6.38, which illustrates the creation
of two vLANs based upon their layer 3 transmission protocol. In examining
the stations shown in Figure 6.38, note that the circles with the uppercase
I represent those stations configured for membership in the vLAN based
upon the use of the IP protocol, while those stations represented by circles
containing the uppercase X represent stations configured for membership in
the vLAN that uses the IPX protocol as its membership criteria. Similarly,
stations represented by circles containing the characters I/X represent stations
operating dual protocol stacks, which enable such stations to become members
of both vLANs.
Two servers are shown at the top of the LAN switch illustrated in
Figure 6.38. One server is shown operating dual IPX/IP stacks, which results
Server
X
Server
I/X
4
0
1
I/X
I/X
5
X I/X
2
3
I
I/X
X
X
X
I
I = vLAN1 membership X = vLAN2 membership
I/X = Membership in both LANs
Legend:
n
= Port n
I
= IP protocol used by station
X = IPX protocol used by station
I/X = IPX and IP protocols used by station
Figure 6.38 vLAN creation based
upon protocol.
bridging and switching methods and performance issues
359
in the server belonging to both vLANs. In comparison, the server on the
upper right of the switch is configured to support IPX and could represent a
NetWare file server restricted to membership in the vLAN associated with the
IPX protocol.
Rule-Based vLANs
A recent addition to vLAN creation methods is based upon the ability of LAN
switches to look inside packets and use predefined fields, portions of fields,
and even individual bit settings as a mechanism for the creation of a vLAN.
Capabilities The ability to create vLANs via a rule-based methodology
provides, no pun intended, a virtually unlimited vLAN creation capability.
To illustrate a small number of the almost unlimited methods of vLAN
creation, consider Table 6.7, which lists eight examples of rule-based vLAN
creation methods. In examining the entries in Table 6.7, note that in addition
to creating vLANs via the inclusion of specific field values within a packet,
such as all IPX users with a specific network address, it is also possible to
create vLANs using the exclusion of certain packet field values. The latter
capability is illustrated by the next to last example in Table 6.7, which forms a
vLAN consisting of all IPX traffic with a specific network address but excludes
a specific node address.
Multicast Support One rule-based vLAN creation example that deserves a
degree of explanation to understand its capability is the last entry in Table 6.7.
Although you might be tempted to think that the assignment of a single IP
address to a vLAN represents a typographical mistake, in actuality it represents
TABLE 6.7
Rule-Based vLAN Creation Examples
All IP users with a specific IP subnet address.
All IPX users with a specific network address.
All network users whose adapter cards were manufactured by the XYZ Corporation.
All traffic with a specific Ethernet-type field value.
All traffic with a specific SNAP field value.
All traffic with a specific SAP field value.
All IPX traffic with a specific network address but not a specific node address.
A specific IP address.
360
chapter six
the ability to enable network stations to dynamically join an IP multicast group
without adversely affecting the bandwidth available to other network users
assigned to the same subnet, but located on different segments attached to a
LAN switch. To understand why this occurs, let me digress and discuss the
concept associated with IP multicast operations.
IP multicast references a set of specifications that allows an IP host to
transmit one packet to multiple destinations. This one-to-many transmission
method is accomplished by the use of Class D IP addresses (224.0.0.0 to
239.255.255.255), which are mapped directly to data link layer 2 multicast
addresses. Through the use of IP multicasting, a term used to reference the use
of Class D addresses, the need for an IP host to transmit multiple packets to
multiple destinations is eliminated. This, in turn, permits more efficient use of
backbone network bandwidth; however, the arrival of IP Class D–addressed
packets at a network destination, such as a router connected to an internal corporate network, can result in a bandwidth problem. This is because multicast
transmission is commonly used for audio and/or video distribution of educational information, videoconferencing, news feeds, and financial reports, such
as delivering stock prices. Due to the amount of traffic associated with multicast transmission, it could adversely affect multiple subnets linked together by
a LAN switch that uses subnets for vLAN creation. By providing a registration
capability that allows an individual LAN user to become a single-user vLAN
associated with a Class D address, Class D packets can be routed to a specific
segment even when several segments have the same subnet. Thus, this limits
the effect of multicast transmission to a single segment.
Switch Usage
The basic use of a stand-alone switch is to support a workgroup that requires
additional bandwidth beyond that available on a shared bandwidth LAN.
Figure 6.39 illustrates the use of a switch to support a workgroup or small
organizational department. As a workgroup expands or several workgroups
are grouped together to form a department, most organizations will want to
consider the use of a two-tiered switching network. The first or lower-level
tier would represent switches dedicated to supporting a specific workgroup
to include local servers. The upper tier would include one or more switches
used to interconnect workgroup switches as well as to provide workgroup
users with access to departmental servers whose access crosses workgroup
boundaries. Since the upper-tier switch or switches are used to interconnect
workgroup switches, the upper-tier switches are commonly referred to as
bridging and switching methods and performance issues
S
361
S
10/100 Mbps
Ethernet switch
Legend:
= Segment
S
= Server
= Workstation
= 100-Mbps connection
Figure 6.39
Support for a small department or workgroup.
backbone switches. Figure 6.40 illustrates a possible use of one backbone
switch to interconnect two workgroup switches.
Since the backbone switch provides an interconnection between workgroup
switches as well as access to departmental servers, the failure of a backbone
switch would have a much more significant effect upon communications than
the failure of a workgroup switch. Thus, you should consider using a backbone
switch with redundant power supplies, common logic, and other key modules.
Then, the failure of one module at worst would only make one or a few port
connections inoperative. If you acquire one or a few additional port modules
you would then have the ability to recable around a port failure without
having to wait for a replacement module to be shipped to your location.
When using one or more backbone switches, it is important to note that
these switches directly affect the throughput between workgroups as well
as the transfer of information to and from departmental servers. Due to
this, most organizations will use dedicated 100-Mbps or Gigabit Ethernet
362
chapter six
DS
DS
100 Mps
ethernet switch
or gigabit switch
WS
WS
10/100 Mbps
ethernet switch
10/100 Mbps
ethernet switch
Legend:
DS = Departmental server
WS = Workgroup server
= Segment
= Workstation
= 100-Mbps connection
Figure 6.40
Creating a two-tiered switch-based network.
switches for backbone operations. If this type of switch is not available at an
economical cost, an alternative is to use a 10-/100-Mbps switch with enough
100-Mbps ports to provide connections from workgroup switches as well as
to departmental servers.
Organizational Switching
Building upon the departmental switching previously illustrated in
Figure 6.40, you can use routers to interconnect geographically dispersed
departments. Doing so can result in an organizational Ethernet switching
network that could span different locations in the same city, different cities,
one or more states, a country, or the entire globe. Figure 6.41 illustrates the
attachment of one router to a backbone switch, connecting the backbone
at one location to a wide area network. Although the actual use of one
bridging and switching methods and performance issues
363
Wide
area
network
DS
DS
Router
100/1000 Mbps
ethernet switch
WS
WS
10/100 Mbps
ethernet switch
10/100 Mbps
ethernet switch
Legend:
DS = Departmental server
WS = Workgroup server
= Segment
= Workstation
= 100-Mbps connection
Figure 6.41
works.
Interconnecting geographically dispersed switch-based net-
or more routers will be governed by our specific networking requirements,
Figure 6.41 illustrates how you can connect one switch-based network to
other switch-based networks.
As you design your network infrastructure you should consider the use of
one or more Ethernet switch features previously discussed in this chapter to
enhance the performance of your network. For example, you may wish to use
full-duplex ports for local and departmental server connections. In addition,
364
chapter six
by the time you read this book economical 10-Gigabit switches should be
available whose use could provide you with another option to consider when
constructing a tiered network structure.
Layer 3 and Layer 4 Switching
In concluding this chapter we will briefly turn our attention to LAN switching
at layer 3 and layer 4 in the OSI Reference Model. In a TCP/IP environment
this requires the LAN switch to look further into each frame to locate the IP
header and the TCP or UDP header and then a particular field in each header.
This results in more processing being required to locate the applicable header
and field within the header that is used as a switching decision criterion.
The key advantage associated with the use of layer 3 and layer 4 LAN
switches is the fact that they enable an organization to use the switch to tailor
their network to a specific operational requirement. For example, assume your
organization operates several Web servers. Through the use of a layer 3 LAN
switch you could direct traffic to different servers based upon the destination
IP address. Using a layer 4 LAN switch you could route traffic based upon IP
address, TCP port, or both metrics. Doing so could provide your organization
with the ability to perform a load balance operation.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter seven
Routers
In Chapter 5, we examined the basic operation and use of a variety of local
area networking components, including routers. Information presented in
that chapter will serve as a foundation for the more detailed discussion of the
operation and use of routers presented in this chapter.
7.1 Router Operation
By operating at the ISO Reference Model network layer, a router becomes
capable of making intelligent decisions concerning the flow of information
in a network. To accomplish this, routers perform a variety of functions that
are significantly different from those performed by bridges. Unlike bridges,
routers are addressable. Routers examine frames that are directly addressed to
them by looking at the network address within each frame to make their forwarding decisions. In an IP environment you would configure the IP address
of the router serving your network in a TCP/IP properties dialog box if you are
using Microsoft Windows. However, because routers were originally referred
to as gateways and the latter term is still used as a reference to a router, you
would enter the IP address of the ‘‘gateway’’ serving your network. Thus, if
your workstation needs to transmit a packet to an IP address on a different
network, it will use the preconfigured gateway address to literally route the
packet to the router, which will then send it off the network towards its
ultimate destination.
IP Support Overview
The most popular network layer protocol supported by routers is the Internet
Protocol (IP), whose packet format was described in Chapter 5. Each IP
network has a distinct network address, and each interface on the network
has a unique host address that represents the host portion of a 32-bit address.
365
366
chapter seven
Since the IP address occurs at the network layer while frames that move
data on a LAN use MAC addresses associated with the data link layer, a
translation process is required to enable IP-compatible devices to use the
transport services of a local area network. Thus, any discussion of how routers
support IP requires an overview of the manner by which hosts use the services
of a router.
When a host has a packet to transmit, it will first determine if the destination IP address is on the local network or a distant network, with the
latter requiring the services of a router. To accomplish this, the host will
use the subnet mask bits set in its configuration to determine if the destination is on the local network. For example, assume the subnet mask is
255.255.255.128. This means the mask extends the network portion of an IP
address to 11111111.11111111.11111111.1, or 25 bit positions, resulting in 7
(32–25) bit positions being available for the host address. This also means
you can have two subnets, with subnet 0 containing host addresses 0 to 127
and subnet 1 having host addresses 128 to 255, with the subnet defined by the
value of the 25th bit position in the IP address. However, we need to note that
restrictions concerning IP network addresses mentioned in Chapter 5 are also
applicable to subnets. That is, you cannot have a subnet address that consists
of all 0’s or all 1’s. Noting these restrictions, the host addresses allowable
on subnet 0 then range from 1 to 126 while the allowable host addresses on
subnet 1 range from 129 to 254.
If we assume the base network IP address is 193.56.45.0, then the base
network, two subnets, and the subnet mask are as follows:
Base network:
Subnet 0:
Subnet 1:
Subnet mask:
11000001.00111000.00101101.00000000
11000001.00111000.00101101.00000000
11000001.00111000.00101101.10000000
11111111.11111111.11111111.10000000
=
=
=
=
193.56.45.0
193.56.45.0
193.56.45.128
193.56.45.128
In examining the above base network, subnets and subnet mask, it is
important to remember that you cannot have a subnet of all 0’s or all 1’s.
Thus, as previously noted, the allowable hosts on subnet 0 range from 1 to 126
while the allowable hosts on subnet 1 range from 129 to 254. Now suppose
a host with the IP address 193.56.45.21 needs to send a packet to the host
whose address is 193.56.45.131. By using the subnet mask, the transmitting
host notes that the destination, while on the same network, is on a different
subnet. Thus, the transmitting host will require the use of a router in the same
manner as if the destination host was on a completely separate network.
Figure 7.1 illustrates the internal and external network view of the subnetted
network. Note that from locations exterior to the network, routers forward
routers
367
Exterior view
193.56.45.0
Router
193.56.45.128
193.56.45.0
Internal network
view
Figure 7.1
Using subnet masks to subdivide a common IP network address.
packets to the router connecting the two subnets as if no subnetting existed.
The corporate router is configured via the use of subnet masks to differentiate
hosts on one subnet from those on the other subnet. From an interior view,
packets originating on one subnet must use the resources of the router to reach
hosts on the other subnet as well as hosts on other networks.
Once the transmitting host notes that the destination IP address is either on
a different network or different subnet, it must use the services of a router.
Although each host will be configured with the IP address of the router, the
host will transport packets via the data link layer, which requires knowledge
of the 48-bit MAC address of the router port connected to the segment the
transmitting host resides on.
The translation between IP and MAC addresses is accomplished by the use
of the Address Resolution Protocol (ARP). To obtain the MAC address of the
router’s LAN interface the host will broadcast an ARP request. This request
will be received by all stations on the segment, with the router recognizing its
IP address and responding by transmitting an ARP response.
Because a continuous use of ARP would rapidly consume network bandwidth, hosts normally maintain the results of ARP requests in cache memory.
Thus, once the relationship between an IP address and MAC address is learned,
subsequent requests to transmit additional packets to the same destination
can be accomplished by the host checking its cache memory.
When packets arrive at the router destined for a host on one of the subnets,
a similar process occurs. That is, the router must obtain the MAC addresses
associated with the IP address to enable the packet to be transported by data
link layer frames to its appropriate destination. Thus, in addition to being
able to correctly support the transmission of packets from one interface to
368
chapter seven
another, an IP-compatible router must also support the ARP protocol. Later
in this chapter we will discuss and describe additional protocols routers
can support.
Basic Operation and Use of Routing Tables
To see the basic operation of routers, consider the simple mesh structure
formed by the use of three routers labeled R1, R2, and R3 in Figure 7.2a. In
this illustration, three Ethernet networks are interconnected through the use
of three routers.
The initial construction of three routing tables is shown in Figure 7.2b.
Unlike bridges, which learn MAC addresses, routers are initially configured,
and either routing tables are established at the time of equipment installation
or the configuration process informs the router of addresses associated with
each of its interfaces and the attachment of networks and subnets, enabling
the device to construct its routing table. Thereafter, periodic communication
between routers updates routing tables to take into consideration any changes
in internet topology and traffic.
In examining Figure 7.2b, note that the routing table for router R1 indicates which routers it must communicate with to access each interconnected
Ethernet network. Router R1 would communicate with router R2 to reach
network 2, and with router R3 to reach network 3.
Figure 7.2c illustrates the composition of a packet originated by station S2
on Ethernet 1 that is to be transmitted to station S12 on Ethernet 2. Router R1
first examines the destination network address and notes that it is on another
network. The router searches its routing table and finds that the frame should
be transmitted to router R2 to reach Ethernet network 2. Router R1 forwards the
frame to router R2. Router R2 then places the frame onto Ethernet network 2
for delivery to station S12 on that network.
Since routers use the network addresses instead of MAC addresses for
making their forwarding decisions, it is possible to have duplicate locally
administered MAC addresses on each network interconnected by a router.
The use of bridges, on the other hand, requires you to review and then
eliminate any duplicate locally administered addresses. This process can be
time-consuming when large networks are connected.
Another difference between bridges and routers is that a router can support
the transmission of data on multiple paths between local area networks.
Although a multiport bridge with a filtering capability can perform intelligent
routing decisions, the result of a bridge operation is normally valid for only
one point-to-point link within a wide area network. In comparison, a router
routers
S1
S4
S9
R1
S12
R2
Network 1
S2
369
Network 2
S3
S10
S11
R3
S5
S8
Network 3
S6
S7
(a) Simple mesh structure
R1
1
R2
*
R2
1
2
3
R3
3
2
R3
R1
1
R1
*
R2
2
R2
3
*
(b) Routing tables
Destination
Source
2.S12
1.S2
DATA
(c) Packet composition
Legend:
R
Figure 7.2
= Router
S
= Network station
Basic router operation.
may be able to acquire information about the status of a large number of paths
and select an end-to-end path consisting of a series of point-to-point links. In
addition, most routers can fragment and reassemble data. This permits packets
to flow over different paths and to be reassembled at their final destination.
With this capability, a router can route each packet to its destination over the
370
chapter seven
best possible path at a particular instant in time, and change paths dynamically
to correspond to changes in network link status on traffic activity.
For example, each of the routing tables illustrated in Figure 7.2b can be
expanded to indicate a secondary path to each network. While router R1
would continue to use the entry of R2 as its primary mechanism to reach
network 2, a secondary entry of R3 could be established to provide an
alternative path to network 2 via routers R3 and R2, rather than directly via
router R2.
Networking Capability
For an illustration of the networking capability of routers, see Figure 7.3.
It shows three geographically dispersed locations that have a total of four
Ethernet and three Token-Ring networks, interconnected through the use of
four routers and four wide area network transmission circuits or links. For
simplicity, modems and DSUs on the wide area network are not shown. This
figure will be referred to several times in this chapter to illustrate different
types of router operations.
E1
L1
R1
R3
TR2
L5
L2
E2
R4
L3
L4
E3
R2
E4
TR3
TR1
Figure 7.3 Router operation. Routers enable the transmission of data over
multiple paths, alternate path routing, and the use of a mesh topology that
transparent bridges cannot support.
routers
371
In addition to supporting a mesh structure that is not obtainable from
the use of transparent bridges, routers offer other advantages in the form of
addressing, message processing, link utilization, and priority of service. A
router is known to stations that use its service. Packets can thus be addressed
directly to a router. This eliminates the necessity for the device to examine in
detail every packet flowing on a network, and results in the router having to
process only messages that are addressed to it by other devices.
Assume that a station on E1 transmits to a station on TR3. Depending on
the status and traffic on network links, packets could be routed via L1 and
use TR2 to provide a transport mechanism to R4, from which the packets are
delivered to TR3. Alternatively, links L2 and L4 could be used to provide
a path from R1 to R4. Although link availability and link traffic usually
determine routing, routers can support prioritized traffic, and may store lowpriority traffic for a small period of time to allow higher-priority traffic to gain
access to the wide area transmission facility. Because of these features, which
are essentially unavailable with bridges, the router is a more complex and
more expensive device.
7.2 Communication, Transport, and Routing Protocols
For routers to be able to operate in a network, they must normally be able
to speak the same language at both the data link and network layers. The
key to accomplishing this is the ability to support common communication,
transport, and routing protocols.
Communication Protocol
Communication protocols support the transfer of data from a station on one
network to a station on another network; they occur at the OSI network
layer. In examining Figure 7.4, which illustrates several common protocol
implementations with respect to the OSI Reference Model, you will note that
Novell’s NetWare uses IPX as its network communications protocol, while
IBM LANs use the PC LAN Support Program, and Microsoft’s LAN Manager
uses the Internet Protocol (IP). Also note that when a TCP/IP stack is used,
certain applications are transported by TCP while others are transported using
UDP; however, both TCP and UDP obtain their routing via the use of IP. This
means that a router linking networks based on Novell, IBM, and Microsoft
LAN operating systems must support those three communication protocols.
Thus, router communication protocol support is a very important criterion
372
chapter seven
OSI layer
Common protocol implementation
Application programs
Application protocols
Application
Presentation
Session
Transport
Network
Novell
Network File
Server Protocol
(NFSP)
Xerox Networking N
System
e
(XNS)
t
B
Sequenced
I
Packet Exchange O
(SPX)
S
Internetwork
Packet Exchange
(IPX)
Data link
IBM
Server Message Block
(SMB)
Microsoft
LAN manager
NetBIOS
NetBIOS
advanced peer-topeer communications
PC LAN
support
program
Transmission
Control Protocol (TCP)
Transport Protocol
Class 4 (TCP4)
Internet
Protocol
(IP)
TCP/IP
applications
TCP
UDP
Internet
Protocol
(IP)
Logical link control 802.2
Media access control
Physical
Transmission media:
twisted pair, coax, fiber optic
Figure 7.4 Common protocol implementations. Although Novell, IBM, and
Microsoft LAN operating system software support standardized physical and
data link operations, they differ considerably in their use of communication
and routing protocols.
in determining whether a particular product is capable of supporting your
networking requirements.
Routing Protocol
The routing protocol is the method used by routers to exchange routing
information; it forms the basis for providing a connection across an internet.
In evaluating routers, it is important to determine the efficiency of the routing
protocol, its effect upon the transmission of information, the method used
and memory required to construct routing tables, and the time required to
adjust those tables. Examples of router-to-router protocols include Xerox
Network Systems’ (XNS) Routing Information Protocol (RIP), TCP/IP’s RIP,
Open Shortest Path First (OSPF), and Hello routing protocols.
Handling Nonroutable Protocols
Although many mainframe users consider IBM’s System Network Architecture
(SNA) as a router protocol, in actuality it is nonroutable in the traditional
routers
373
sense of having network addresses. This means that for a router to support
SNA or another nonroutable protocol, such as NetBIOS, the router cannot
compare a destination network address against the current network address
as there are no network addresses to work with. Instead, the router must be
capable of performing one or more special operations to handle nonroutable
protocols. For example, some routers may be configurable such that SNA
addresses in terms of physical units (PUs) and logical units (LUs) can be
associated with pseudonetwork numbers, enabling the router to route an
unroutable protocol. Another common method employed by some routers is
to incorporate a nonroutable protocol within a routable protocol, a technique
referred to as tunneling. A third method, and one considered by many to be
the old reliable mechanism, is to use bridging. Later in this chapter when
we cover protocol-independent routers, we will describe methods that can be
used to route nonroutable protocols, to include SNA traffic.
Transport Protocol
The transport protocol guarantees the delivery of information between two
points. Here, the transport protocol represents the fourth layer illustrated in
Figure 7.4. Examples of transport protocols include SPX, TCP, UDP, X.25, and
Frame Relay.
There is a wide variety of communication and transport protocols in use
today. Some of these protocols, such as Apple Computer’s AppleTalk, were
designed specifically to operate on local area networks. Other protocols, such
as X.25 and Frame Relay, were developed as wide area network protocols.
Fifteen popular communication and transport protocols are listed below.
Many routers support only a subset of these protocols.
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
♦
AppleTalk
Apple Domain
Banyan VINES
CHAOSnet
DECnet Phase IV
DECnet Phase V
DDN X.25
Frame Relay
ISO CLNS
HDLC
Novell IPX
SDLC
374
chapter seven
♦ TCP/IP
♦ Xerox XNS
♦ X.25
7.3 Router Classifications
Depending upon their support of communication and transport protocols,
routers can be classified into two groups: protocol-dependent and protocolindependent.
Protocol-Dependent Routers
To understand the characteristics of a protocol-dependent router, consider the
network illustrated in Figure 7.3. If a station on network E1 wishes to transmit
data to a second station on network E3, router R1 must know that the second
station resides on network E3, and it must know the best path to use to reach
that network. The method used to determine the destination station’s network
also determines the protocol dependency of the router.
If the station on network E1 tells router R1 the destination location, it must
supply a network address in every LAN packet it transmits. This means that
all routers in the intranet must support the protocol used on network E1.
Otherwise, stations on network E1 could not communicate with stations
residing on other networks, and vice versa.
NetWare IPX Example
To illustrate the operation of a protocol-dependent router, let us assume
that networks E1 and E3 use Novell’s NetWare as their LAN operating system. The routing protocol used at the network layer between a station and
server on a Novell network is known as IPX. This protocol can also be used
between servers.
Under NetWare’s IPX, a packet addressed to a router will contain the destination address in the form of network and host addresses, and the origination
address in the form of the source network and source host addresses. Here,
the IPX term host is actually the physical address of a network adapter card.
Figure 7.5a illustrates in simplified format the IPX packet composition
for workstation A on network E1, transmitting data to workstation B on
network E3, under Novell’s NetWare IPX protocol.
After router R1 receives and examines the packet, it notes that the destination
address E3 requires the routing of the packet to router R2. It converts the first
routers
B
E3
A
E1 R1
375
Data
(a) Packet from workstation A, network E1 to router R1
B
E3 R2
A
E1 R1
Data
(b) Router (R1) to router (R2) packet
B
E3
A
E1 R2
Data
(c) Router R2 converts packet for placement on network E3
Figure 7.5
NetWare IPX routing.
packet into a router (R1) to router (R2) packet, as illustrated in Figure 7.5b. At
router R2, the packet is again examined. Router R2 notes that the destination
network address (E3) is connected to that router, so it reconverts the packet
for delivery onto network E3. It does this by converting the destination
router address to a source router address and transmitting the packet onto
network E3. This is illustrated in Figure 7.5c.
Addressing Differences
In the preceding example, note that each router uses the destination workstation and network addresses to transfer packets. If all protocols used the same
format and addressing structure, routers would be protocol-insensitive at the
network layer. Unfortunately, this is not true. For example, TCP/IP addressing
conventions are very different from those used by NetWare. This means that
networks using different operating systems require the use of multiprotocol
routers configured to perform address translation. Each multiprotocol router
must maintain separate routing tables for each supported protocol, requiring
additional memory and processing time.
Other Problems
Two additional problems associated with protocol-dependent routers are
the time required for packet examination and the fact that not all LAN
protocols are routable. If a packet must traverse a large network, the time
required by a series of routers to modify the packet and assure its delivery to the next router can significantly degrade router performance. To
overcome this problem, organizations should consider the use of a frame
relay service.
376
chapter seven
In addition to providing an enhanced data delivery service by eliminating
error detection and correction within the network, the use of a frame relay
service can significantly reduce the cost of routers. Consider, for example,
the network in Figure 7.3, in which four routers are interconnected through
the use of five links. To support transmission on five links, the routers
require ten ports. Normally, each router port is obtained as an adapter card
installed in a high-performance computer. If a frame relay service is used,
the packet network providing that service also provides the routing paths to
interconnect routers, as illustrated in Figure 7.6. This reduces the number
of required router ports to four. This reduction can result in a considerable
hardware savings.
In addition to using a frame relay service provider, another method that can
reduce the cost of router hardware and communications circuits is obtained
from establishing a virtual private network (VPN) through the Internet. In
doing so, you would need to consider encryption of data as the Internet represents an open, public network. However, the physical topology associated
with connecting geographically separated locations would remain similar to
that shown in Figure 7.6, with the frame relay packet network service being
replaced by the Internet.
A second problem associated with protocol-dependent routers is the fact
that some LAN protocols cannot be routed using that type of device. This is
because some LAN protocols, such as NetBIOS and IBM’s LAN Server — and
unlike NetWare, DECnet, and TCP/IP — do not include routing information
R1
R3
Frame relay
packet network
service
R2
R4
Figure 7.6 Using a frame relay service. If a frame relay service is used,
the packet network provides the capability for interconnecting each network
access port to other network access ports. Thus, only one router port is required
to obtain an interconnection capability to numerous routers connected to
the network.
routers
377
within a packet. Instead, those protocols employ a user-friendly devicenaming convention instead of using network and device addresses. This
convention permits such names as ‘‘Gil’s PC’’ and ‘‘Accounting Printer’’ to
be used. For example, IBM’s NetBIOS was designed to facilitate program-toprogram communication by hiding the complexities of network addressing
from the user. Thus, NetBIOS uses names up to 16 alphanumeric characters
long to define clients and servers on a network. Unfortunately, NetBIOS does
not include a facility to distinguish one network from another, since it lacks
a network addressing capability. Such protocols are restricted to using the
physical addresses of adapter cards, such as Ethernet source and destination
addresses. Since a protocol-dependent router must know the network on
which a destination address is located, it cannot in a conventional sense
route such protocols. A logical question, then, is how a router interconnects
networks using an IBM LAN protocol or a similar nonroutable protocol.
The answer to this question will depend upon the method used by the router
manufacturer to support nonroutable protocols. As previously discussed, such
methods can include bridging, tunneling, or the configuration of a router that
enables pseudonetwork addresses to be assigned to each device.
Protocol-Independent Routers
A protocol-independent router functions as a sophisticated transparent bridge.
That is, it addresses the problem of network protocols that do not have
network addresses. It does this by examining the source addresses on connected networks to learn automatically what devices are on each network.
The protocol-independent router assigns network identifiers to each network
whose operating system does not include network addresses in its network
protocol. This activity enables both routable and nonroutable protocols to be
serviced by a protocol-dependent router.
In addition to building address tables automatically like a transparent
bridge, a protocol-independent router exchanges information concerning its
routing directions with other internet routers. This enables each router to
build a map of the interconnected networks. The method used to build the
network map falls under the category of a link state routing protocol, which
is described later in this chapter.
Advantages
There are two key advantages to using protocol-independent routers. Those
advantages are the abilities of routers to learn network topology automatically
and to service nonroutable protocols. The ability to learn network topology
378
chapter seven
automatically can considerably simplify the administration of an internet. For
example, in a TCP/IP network, each workstation has an IP address and must
know the IP addresses of other LAN devices it wants to communicate with.
IP addresses are commonly assigned by a network administrator, and they
must be changed if a station is moved to a different network, or if a network is
segmented because of a high level of traffic or other reason. In such situations,
LAN users must be notified about the new IP address, or they will not
be able to locate the moved station. Obviously, the movement of stations
within a building between different LANs could become a considerable
administrative burden. The ability of a protocol-independent router to learn
addresses automatically removes the administrative burden of notifying users
of changes in network addresses.
An exception to the preceding occurs through the use of the Dynamic
Host Configuration Protocol (DHCP). Through the use of a DHCP server and
appropriate client software, stations are dynamically assigned IP addresses
for a relatively short period of time. Once they complete an application the
server can reassign the address to a new station. Although the use of the
DHCP can ease the administrative burden of configuring and reconfiguring IP
workstations, it requires the use of a server and client software. Thus, there
continues to be no free lunch in networking.
The ability to route nonroutable protocols can be of considerable assistance
in integrating IBM SNAs into an enterprise network. Otherwise, without the
use of protocol-independent routers, organizations may have to maintain
separate transmission facilities for SNA and LAN traffic.
Supporting SNA Traffic
Figure 7.7 illustrates an example of the use of protocol-independent routers
to support both inter-LAN and SNA traffic. In this example, an IBM SNA
network, a 3174 control unit with a Token-Ring adapter (TRA) at a remote
site provides communications connectivity to an IBM 3745 communications
controller at a central site. Routers must be capable of routing both SNA and
LAN traffic to enable the use of a common transmission facility between the
central and remote site.
Methods to Consider
There are essentially three methods by which SNA and LAN traffic can
be combined for routing over a common network: encapsulation, conversion, and protocol-independent routing. Under the encapsulation method,
SNA packets are modified so that another protocol’s header, addressing,
routers
Central site
Remote site
TR1
379
TRA
Host
IBM
3745
3174
R1
R2
E1
E2
TR2
Figure 7.7 Supporting SNA traffic. A protocol-independent router can support SNA traffic and other LAN traffic over a common transmission facility.
and trailer fields surround each SNA packet. For example, a TCP/IP protocol–dependent router would encapsulate SNA into TCP/IP packets for
routing through a TCP/IP network. Since a TCP/IP packet has over 60 bytes
of overhead, and the average SNA packet is 30 bytes in length, encapsulation can reduce performance considerably when transmission occurs over
low-speed WAN links. A second disadvantage of encapsulation is that
it requires the existence of a corporate network using the encapsulation
protocol. Otherwise, you would have to build this network to obtain an
encapsulation capability.
The second method used for integrating SNA traffic with LAN traffic occurs
through the use of protocol conversion. This technique eliminates the need
for adding network headers and enhances the efficiency of the protocol
integration efforts.
The third method by which an SNA network can be integrated with LAN
traffic is through protocol-independent routing. Protocol-independent routers
assign a LAN device address to each SNA control unit and communications
controller. Then, SNA packets are prefixed with source and destination
addresses to permit their routing through the internet. At the destination
router, the addressing information is removed, and the SNA packets are
delivered to their destination in their original form. Since the addition of
source and destination addresses adds a significantly lower number of bytes
than an encapsulation process, overhead is reduced. This, in turn, enables
you to consider using lower-speed circuits to interconnect locations.
380
chapter seven
Another advantage of protocol-independent routing over encapsulation
relates directly to the operation of SNA networks, with several SNA operational
characteristics warranting attention. They will help us appreciate the advantages of protocol-independent routing, and they will explain the rationale for
such routers’ requiring a traffic priority mechanism to support SNA traffic
efficiently.
Need for Priority Mechanism
In its original incarnation, SNA represents a hierarchically structured network,
as shown in Figure 7.8. Here, the communications controller communicates
with control units, which in turn communicate with attached terminal devices.
The communications controller periodically polls each control unit, and the
control unit periodically polls each terminal device. If there is no data
to transmit in response to a poll, each polled device indicates this fact
to the higher-level device by responding negatively to the poll. Thus, the
communications controller expects to receive a response to each poll it
generates. In fact, if it does not receive a response within a predefined
period of time (typically less than five seconds), the communications controller will assume that the control unit has malfunctioned, terminate any
Host computer
Communications
controller
Control
unit
Control
unit
Control
unit
Figure 7.8 SNA network hierarchy. In an SNA network, communications
controllers poll control units, which in turn poll attached devices.
routers
381
active sessions to devices attached to the control unit, and then attempt
to reestablish communications by sending an initialization message to the
control unit.
When SNA and LAN traffic is integrated into a common router-based
network, the operation of routers can adversely affect SNA traffic because
of the time delays associated with routing and encapsulation. For example,
routers may alter the path used to transmit data, depending on internet traffic
and circuit availability. If path switching makes the SNA traffic exceed its
timeout threshold, the communications controller will terminate sessions and
reinitialize control units that fail to respond to its poll within the predefined
time period. Similarly, delays caused by encapsulation can also result in
unintended timeouts.
To prevent unintended timeouts caused by path switching, some vendors
have added a traffic priority mechanism to their protocol-independent routing
capability. This priority mechanism enables users to assign a higher priority to
SNA traffic than to LAN traffic, thus enabling SNA data to be forwarded across
wide area networks ahead of other inter-LAN traffic. This can considerably
reduce or even eliminate the potential for LAN traffic to obstruct the flow of
SNA data; it may also result in unintended timeouts, inadvertently causing
session terminations.
7.4 Routing Protocols
The routing protocol is the key element to transferring information across
an internet in an orderly manner. The protocol is responsible for developing
paths between routers, using a predefined mechanism.
Types of Routing Protocols
There are two types of routing protocols: interior and exterior domain. Here,
we use the term domain to refer to the connection of a group of networks to
form a common entity, such as a corporate or university enterprise network.
Interior Domain Routing Protocols
An interior domain routing protocol is used to control the flow of information
within a series of separate networks that are interconnected to form an internet.
Thus, interior domain routing protocols provide a mechanism for the flow
of information within a domain and are also known as intradomain routing
382
chapter seven
protocols. Such protocols create routing tables for each autonomous system
within the domain, using such metrics as the hop count or time delay to
develop routes from one network to another within the domain. Examples of
interior domain routing protocols include RIP, OSPF, and Hello.
Exterior Domain Routing Protocols
Exterior domain routing protocols are used to connect separate domains
together. Thus, they are also referred to as interdomain routing protocols.
Examples of interdomain routing protocols include the Exterior Gateway
Protocol (EGP), the Border Gateway Protocol (BGP), and the Inter-Domain
Routing Protocol (IDRP). Unlike interior domain routing protocols, which
are focused on the construction of routing tables for data flow within
a domain, interdomain routing protocols specify the method by which
routers exchange information concerning what networks they can reach on
each domain.
Figure 7.9 illustrates the use of interior and exterior domain routing protocols. In this example, OSPF is the intradomain protocol used in Domain A,
while RIP is the intradomain protocol used in Domain B. Routers in Domains
A and B use the interdomain routing protocols EGP and/or BGP to determine
what networks on other domains they can reach.
Exterior Gateway Protocol
There are four basic functions performed by the Exterior Gateway Protocol.
First, the EGP performs an acquisition function, which enables a router in one
domain to request information exchange with a router on another domain.
Since routers also serve as gateways to the domain, they are sometimes
referred to as gateways. A second function performed by the router gateway is
to test periodically whether its EGP neighbors are responding. The third and
most important function performed by the EGP is to enable router gateways
to exchange information concerning the networks in each domain by transmitting routing update messages. The fourth function involves terminating an
established neighbor relationship between gateways on two domains.
To accomplish its basic functions, EGP defines nine message types.
Figure 7.10 illustrates EGP message types associated with each of the three
basic features performed by the protocol.
Under the EGP, once a neighbor is acquired, Hello messages must be
transmitted at a minimum of 30-second intervals. In addition, routing updates
must be exchanged at a minimum of two-minute intervals. This exchange of
information at two-minute intervals can result in the use of a considerable
routers
383
Domain A
R
Intradomain (interior)
routing protocols
OSPF
R
R
Domain B
Interdomain
(exterior)
routing protocols
EGP,
BGP
R
RIP
R
R
Figure 7.9 Interior and exterior routing protocols. An interior routing protocol controls the flow of information within a collection of interconnected
networks known as a domain. An exterior routing protocol provides routers
with the ability to determine what networks on other domains they can reach.
amount of the bandwidth linking domains when the number of networks on
each domain is large, or when the circuits linking domains consist of lowspeed lines. The Border Gateway Protocol was developed to alleviate those
potential problems.
Border Gateway Protocol
The Border Gateway Protocol represents a follow-on to the EGP. Unlike
the EGP, in which all network information is exchanged at two-minute or
shorter intervals, the BGP transmits incremental updates as changes occur.
This can significantly reduce the quantity of data exchanged between router
gateways, thus freeing up a considerable amount of circuit bandwidth for the
transmission of data. Both the EGP and the BGP run over TCP/IP and are
standardized by Internet documents RFC904 and RFC1105, respectively.
384
chapter seven
Function
Message type
Acquisition request
Acquiring neighbors
Acquisition confirm
or
Acquisition refuse
Neighbor reachability
Hello
Routing update
Poll request
I heard you
Routing update
De-acquiring neighbors
Cease request
Cease confirm
Domain A
R
R
R
Domain B
R
R
Figure 7.10
R
Exterior gateway protocol message types.
Types of Interior Domain Routing Protocols
As previously discussed, interior domain routing protocols govern the flow of
information between networks. This is the type of routing protocol that is of
primary interest to most organizations. Interior domain routing protocols can
routers
385
be further subdivided into two broad categories, based on the method they use
for building and updating the contents of their routing tables: vector distance
and link state.
Vector Distance Protocol
A vector distance protocol constructs a routing table in each router, and
periodically broadcasts the contents of the routing table across the internet.
When the routing table is received at another router, that device examines
the set of reported network destinations and the distance to each destination.
The receiving router then determines whether it knows a shorter route to a
network destination, finds a destination it does not have in its routing table,
or finds a route to a destination through the sending router where the distance
to the destination changed. If any one of these situations occurs, the receiving
router will change its routing tables.
The term vector distance relates to the information transmitted by routers.
Each router message contains a list of pairs, known as vector and distance.
The vector identifies a network destination, while the distance is the distance
in hops from the router to that destination.
Figure 7.11 illustrates the initial distance vector routing table for the routers
R1 and R2 shown in Figure 7.3. Each table contains an entry for each directly
connected network and is broadcast periodically throughout the internet.
The distance column indicates the distance in hops to each network from
the router.
At the same time router R1 is constructing its initial distance vector
table, other routers are performing a similar operation. The lower portion
of Figure 7.11 illustrates the composition of the initial distance vector table
for router R2. As previously mentioned, under a distance vector protocol, the
contents of each router’s routing table are periodically broadcast. When routers
R1 and R2 broadcast their initial distance vector routing tables, each router
uses the received routing table to update its initial routing table. Figure 7.12
Figure 7.11
Initial distance vector routing tables.
386
chapter seven
Figure 7.12
Initial routing table update.
illustrates the result of this initial routing table update process for routers R1
and R2.
As additional routing tables are exchanged, the routing table in each router
will converge with respect to the internet topology. However, to ensure
that each router knows the state of all links, routing tables are periodically
broadcast by each router. Although this process has a minimal effect upon
small networks, its use with large networks can significantly reduce available
bandwidth for actual data transfer. This is because the transmission of lengthy
router tables will require additional transmission time in which data cannot
flow between routers.
Popular vector distance routing protocols include the TCP/IP Routing Information Protocol (RIP), the AppleTalk Routing Table Management Protocol
(RTMP), and Cisco’s Interior Gateway Routing Protocol (IGRP).
Routing Information Protocol
Under RIP, participants are either active or passive. Active participants are
normally routers that transmit their routing tables, while passive machines listen and update their routing tables based upon information supplied by other
devices. Normally, host computers are passive participants, while routers are
active participants.
Operation
Under RIP, an active router broadcasts its routing table every 30 seconds.
Each routing table entry contains a network address and the hop count to
the network. To illustrate an example of the operation of RIP, let’s redraw
the network previously shown in Figure 7.3 in terms of its links and nodes,
routers
387
B
1
A
3
5
2
C
4
D
Figure 7.13 Redrawing the network in Figure 7.3 in
terms of its links and nodes.
replacing the four routers by the letters A, B, C, and D for simplicity of
illustration. Figure 7.13 contains the revised network consisting of four nodes
and five links.
When the routers are powered up they only have knowledge of their
local conditions. Thus, each routing table would contain a single entry. For
example, the table of router n would have the following value:
From n to
Link
Hop Count
n
local
0
For the router represented by node A, its table would then become:
From A to
Link
Hop Count
A
local
0
Thirty seconds after being turned on, node A will broadcast its distance
vector (A = 0) to all its neighbors, which, in Figure 7.13 are nodes B and C.
Node B receives on link 1 the assistance vector A = 0. Upon receipt of this
message, it updates its routing table as follows, adding one to the hop count
associated with the distance vector supplied by node 1:
From B to
Link
Hop Count
B
A
Local
1
0
1
388
chapter seven
Node B can now prepare its own distance vector (B = 0, A = 1) and transmit
that information on its connections (links 1, 3, and 5).
During the preceding period node C would have received the initial distance
vector transmission from node A. Thus, node C would have updated its routing
table as follows:
From C to
Link
Hop Count
C
A
Local
2
0
1
Once it updates its routing table, node C will then transmit its distance
vector (C = 0, A = 1) on links 2, 3, and 4.
Assuming the distance vector from node B is now received at nodes A
and C, each will update their routing tables. Thus, their routing tables would
appear as follows:
From A to
Link
Hop Count
A
B
Local
1
0
1
From C to
Link
Hop Count
C
A
B
Local
2
3
0
1
1
At node D, its initial state is first modified when it receives the distance
vector (B = 0, A = 1) from node B. Since D received that information on
link 5, it updates its routing table as follows, adding one to each received
hop count:
From D to
Link
Hop Count
D
B
A
Local
5
5
0
1
2
Now, let’s assume node D receives the update of node C’s recent update
(C = 0, A = 1, B = 1) on link 4. As it does not have an entry for node C, it
will add it to its routing table by entering C = 1 for link 4. When it adds 1 to
routers
389
the hop count for A received on link 4, it notes that the value is equal to the
current hop count for A in its routing table. Thus, it discards the information
about node A received from node C. The exception to this would be if the
router maintained alternate routing entries to use in the event of a link failure.
Next, node D would operate upon the vector B = 1 received on link 4, adding
one to the hop count to obtain B = 2. Since that is more hops than the current
entry, it would discard the received distance vector. Thus, D’s routing table
would appear as follows:
From D to
Link
Hop Count
D
C
B
A
Local
4
5
5
0
1
1
2
The preceding example provides a general indication of how RIP enables
nodes to learn the topology of a network. In addition, if a link should fail, the
condition can be easily compensated for as similar to bridge table entries, those
of routers are also time stamped and the periodic transmission of distance
vector information would result in a new route replacing the previously
computed one.
One key limitation of RIP is the maximum hop distance it supports. This
distance is 16 hops, which means that an alternative protocol must be used
for large networks.
Configuration Example
Because Cisco Systems manufactures over 70 percent of all routers we will
use its Internetworking Operating System (IOS) to illustrate an example of
the router configuration process. In doing so we will assume that during the
router setup process you entered the name of the router as Cisco, an enable
password which governs access to the router’s privileged mode of operation
as ‘‘wxyz,’’ and agreed to configure IP on a serial and Ethernet interface.
Figure 7.14 illustrates the interfaces of the router and the assignment of IP
addresses to each router interface.
Router Modes
Cisco routers support two modes of operation — user and privileged. The user
mode is the default mode when you reboot a router and access it via a direct
390
chapter seven
Network
205.131.175.1
s0
Router
e0
198.78.46.1
Figure 7.14
Configuring a Cisco router to use RIP as its routing protocol.
console connection or access the router via Telnet. When you access a router
and are in its user mode this fact is denoted by the router name followed
by the greater than (>) sign. If your router name is Cisco, the user mode
prompt becomes:
Cisco >
Once in user mode you can display many items concerning the router to
include its configuration; however, you cannot configure the router. To obtain
the ability to configure the router you need to be in its privileged mode. To do
so, you need to type ‘‘enable’’ at the user mode prompt and then press Enter.
You will then be prompted to enter the enable password, after which the
prompt changes to the router name followed by the pound sign (#) — called a
hash sign in the UK. The following example illustrates how you would move
routers
391
from the user mode into the privileged mode when using a Cisco router:
Cisco > enable
Cisco > Password : wxyz
Cisco#
Cisco’s IOS uses a command interpreter to execute commands you enter.
That interpreter is referred to as the Exec and user and privileged modes can
be considered to represent different levels of the Exec. Assuming you are in the
privileged mode, you can obtain the ability to configure a router by entering
the command ‘‘configure,’’ or if you want to eliminate a prompt, enter the
command ‘‘config t’’ to denote you are configuring the router from a terminal
as opposed to configuring it from memory or the network. To configure the
s0 (serial 0) interface with the IP address 205.131.175.1, you would enter the
following IOS statements:
Cisco#config t
Enter configuration commands, one per line, End with CNTL/Z
Cisco(config)#interface s0
Cisco(config-if)#ip address 205.131.175.1 255.255.255.0
Once you define the IP addresses for each interface you need to configure a
routing protocol, so let’s do so. Since RIP is limited to 16 hops (0 through
15), which makes it suitable for small to medium networks, let’s assume
you will use this routing protocol. To configure RIP you would type ‘‘config
t’’ at the privileged prompt. At the configuration prompt you would enter
‘‘router rip’’ to select RIP as the routing protocol. Next you would enter the
keyword ‘‘network’’ followed by what is referred to as the major network
number at the next config prompt. Here Cisco uses the term major network
number to reference the IP network address for a Class A, B or C network
directly connected to the router. From Figure 7.14 you would note that there
is only one network connected to the router. Thus, you would configure RIP
as follows:
Cisco#config t
Enter configuration commands, one per line. End with CNTL/Z.
Cisco(config)#router rip
Cisco(config-router)#network 198.78.46.0
392
chapter seven
If you had more than one network connected to the router you would repeat
using the keyword ‘‘network’’ followed by the IP address for each network connected to the router. Once you finish entering the directly connected networks
you would press CNTL + Z to end the configuration session. Now that we
have an appreciation of how we would configure RIP in a Cisco environment,
let’s continue our examination of a few additional routing protocols.
Routing Table Maintenance Protocol
The Routing Table Maintenance Protocol (RTMP) was developed by Apple
Computer for use with that vendor’s AppleTalk network. Under RTMP,
each router transmits messages to establish routing tables and update them
periodically. The update process, during which information on an internet
is exchanged between routers, also serves as a mechanism for implementing
alternate routing. This is because the absence of update information for a
greater than expected period of time converts the status of an entry in other
router routing tables from ‘‘good’’ to ‘‘suspect’’ and then to ‘‘bad.’’
RTMP is delivered by AppleTalk’s Data Delivery Protocol (DDP), which
is a network layer connectionless service operating between two upperlayer processes referred to as sockets. Four types of packets are specified by
RTMP: data, request, route data request, and response. Routing updates are
transmitted as data packets. End nodes transmit request packets to acquire
information about the identity of the internet routers (IRs) to which they can
transmit nonlocal packets. Internet routers respond to route request packets
with response packets, while end nodes that want to receive an RTMP data
packet indicate this by sending a route data request packet. The latter packet
type is also used by nodes that require routing information from IRs not
directly connected to their network.
Routing Process
In the AppleTalk routing process, the source node first examines the destination network number. If the packet has a local network address, it is passed
to the data link layer for delivery. Otherwise, the packet is passed to any of
the IRs that reside on a network segment. The IR will examine the destination
address of the packet and then check its routing tables to determine the next
hop, routing the packet on a specific port that enables it to flow toward the
next hop. Thus, a packet will travel through the internet on a hop-by-hop
basis. When the packet reaches an IR connected to the destination network,
the data link layer is used to deliver the packet to its local destination.
routers
Network 8−10
Network 1−3
Port 2
R1
393
Port 2
Port 1
Port 1
R3
Port 3
Network 11
Network
7
Network 4−6
Port 1
R2
Port 2
Network 12
Port 2
R4
Port 1
Rounting Table for Router R1
Network Range
Distance
Port
Next IR
1−3
7−7
4−6
8−10
11−11
12−12
0
0
1
1
1
2
2
1
1
1
1
1
0
0
R2
R3
R3
R3
Figure 7.15
Entry State
Good
Good
Good
Good
Good
Good
AppleTalk routing table example.
Figure 7.15 illustrates a sample AppleTalk network and the routing table for
one router. Each AppleTalk routing table has five entries, as indicated. The
network range defines the range of network numbers assigned to a particular
network segment. The distance entry specifies the number of routers that
must be traversed prior to the destination being reached, and port defines the
router port used to provide access to a destination location. The next IR entry
indicates the identifier of the next IR on the internet, and entry state defines
the status of receiving routing updates. An entry state can go from ‘‘good’’
to ‘‘suspect’’ to ‘‘bad’’ if routing updates are not received within predefined
time intervals.
Interior Gateway Routing Protocol
Cisco’s proprietary Interior Gateway Routing Protocol is a distance vector
protocol in which updates are transmitted at 90-second intervals. Each IGRP
gateway operates on a local basis to determine the best route for forwarding
394
chapter seven
data packets. Unlike protocols that require each router to have a complete
routing table, each IGRP router computes only a portion of the routing table,
based on its perspective of the Intranet. This enables the IGRP to support
distributed routing without requiring each router to have a complete view of
the internet.
Other features of the IGRP include the ability to establish routes automatically, perform load balancing dynamically over redundant links, and detect
and remove bad circuits from the internet routing tables it develops. Routes
are declared inaccessible if an update is not received within three update
periods (270 seconds). After five update periods (450 seconds), the route is
removed from the routing table.
IGRP, unlike RIP, supports a wide range of metrics that can be used to
govern routing. Both reliability and load can be considered as well as internetwork delay and the bandwidth on a link. The use of IGRP occurs within
an autonomous system (AS) that represents a group of routers connecting
networks that use a common routing protocol. This makes it possible for a
network administrator to set weighting factors for each routing metric or to
allow the protocol to use default weights. An enhanced IGRP (EIGRP) was
introduced by Cisco that incorporates the capabilities of link-state protocols
into distant vector protocols. Under EIGRP a router stores all of its neighbors’
routing tables so it can rapidly adapt to alternate routes. Other improvements
under EIGRP include support for variable-length subnet masks and the transmission of partial table entries when the metric for a route changes in place
of full periodic updates. As you might expect, in a large AS, this considerably
reduces the overhead and the adverse effect of bandwidth congestion due to
table updates.
Link State Protocols
A link state routing protocol addresses the traffic problem associated with
any large networks that use a vector distance routing protocol. It does this
by transmitting routing information only when there is a change in one
of its links. A second difference between vector difference and link state
protocols concerns the manner in which a route is selected when multiple
routes are available between destinations. For a vector distance protocol,
the best path is the one that has the fewest intermediate routers on hops
between destinations. A link state protocol, however, can use multiple paths
to provide traffic balancing between locations. In addition, a link state protocol
permits routing to occur based on link delay, capacity, and reliability. This
routers
395
provides the network manager with the ability to specify a variety of route
development situations.
SPF Algorithms
Link state routing protocols are implemented by a class of algorithms known
as Shortest Path First (SPF). Unfortunately, this is a misnomer, since routing
is not based on the shortest path.
The use of SPF algorithms requires each participating router to have complete knowledge of the intranet topology. Each router participating in an SPF
algorithm then performs two tasks: status testing of neighboring routers and
periodic transmission of link status information to other routers.
To test neighboring routers, a short message is periodically transmitted.
If the neighbor replies, the link is considered to be up. Otherwise, the
absence of a reply after a predefined period of time indicates that the link
is down.
To provide link status information, each router will periodically broadcast
a message indicating the status of each of its links. Unlike the vector distance
protocol, in which routes are specified, an SPF link status message simply
indicates whether communications are possible between pairs of routers.
Using information in the link status message, routers are able to update their
network maps.
Unlike vector distance protocols, in which tables are required to be
exchanged, link state protocols such as SPF algorithms exchange a much
lower volume of information in the form of link status queries and replies.
Then, SPF participating routers simply broadcast the status of each of their
links, and other routers use this information to update their intranet maps.
This routing technique permits each router to compute routes independently
of other routers, and eliminates the potential for the table flooding that can
occur when a vector state protocol is used to interconnect a large number
of networks.
To illustrate the operation of a link state routing protocol, let us return to
the internet configuration previously illustrated in Figure 7.15. Figure 7.16
indicates the initial network map for router R1. This map lists the destination
of all networks on the intranet from router R1, with their distances and
routes. Note that if multiple routes exist to a destination, each route is listed.
This defines a complete network topology, and allows alternate routes to be
selected if link status information indicates that one or more routes cannot
be used.
396
chapter seven
Figure 7.16
Router R1 initial network map.
Figure 7.17
Link status messages.
Let us suppose that at a particular point in time, the link status messages
generated by the routers in the intranet are as indicated in Figure 7.17.
Note that both routers R2 and R3 have determined that link L3 is down.
Using this information, router R1 would then update the status column
for its network map. Since link L3 is down, all routes that require a data
routers
397
flow between R2 and R3 would have their status changed to ‘‘down.’’ For
example, for destination E3 via route R3, R2 would have its status changed
to ‘‘down.’’ Since the minimum distance to E3 is 1 hop via router R2, the
failure of link L3 would not affect data flow from router R1 to network
E3. Now consider the effect of link L2 becoming inoperative. This would
affect route R2, which has the minimum distance to network E3. It would
still leave route R3 = R4 = R2, although this route would have a distance
of three hops. Of course, when a new link status message indicates that a
previously declared down link is up, each router’s network map is updated
accordingly.
Examples of link state protocols include Open Shortest Path First (OSPF),
OSI Intermediate System to Intermediate System (IS-IS), DECnet Phase V, and
IBM’s Advanced Peer-to-Peer Networking (APPN). Due to space limitations,
we will review briefly the operational features of only the first of these link
state protocols.
Open Shortest Path First Protocol
The OSPF protocol is an interior domain routing protocol that uses the
SPF algorithm. Like the EGP, the OSPF protocol consists of a small core of
different types of messages. Under the OSPF protocol, five message types
are defined.
A Hello message is used to test the reachability of other devices. A
database description message passes the topology. The remaining message
types include a link status request, a link status update, and a link status
acknowledgement message.
Initially, OSPF generates database description messages to initialize its
network topology database. These messages are flooded through the domain.
However, once a topological database is formed, routing table updates are
transmitted at 30-minute intervals unless there is a change in the network, in
which case there is an immediate update.
Like the EGP, OSPF routers use a Hello message to discover their neighbors.
One of the key features of the OSPF protocol is its ability to authenticate messages. This means that you can’t entice routers to transmit packets to a specific computer for examination or interception by generating
low-cost routes. Another key feature of the OSPF protocol is its ability to support multiple active paths of equal weight: it selects each path
in a round-robin manner. Table 7.1 provides a summary of 16 common
routing protocol abbreviations, their meanings, and a short description of
each protocol.
chapter seven
398
TABLE 7.1
Common Routing Protocols
AURP
AppleTalk Update Routing Protocol. This routing protocol is implemented
in Apple networks and sends changes to routing tables via updates.
BGP
Border Gateway Protocol. This is a TCP/IP interdomain routing protocol.
CLNP
Connectionless Network Protocol. This is the OSI version of the IP routing
protocol.
DDP
Datagram Delivery Protocol. This routing protocol is used in Apple’s
AppleTalk network.
EGP
Exterior Gateway Protocol. This TCP/IP protocol is used to locate
networks on another domain.
IDRP
Inter-Domain Routing Protocol. This is the OSI interdomain routing
protocol.
IGP
Interior Gateway Protocol. This TCP/IP protocol is used by routers to
move information within a domain.
IGRP
Interior Gateway Routing Protocol. This is a proprietary routing protocol
developed by Cisco Systems.
IP
Internet Protocol. This is the network layer protocol of the TCP/IP
(Transmission Control Protocol/Internet Protocol) suite of protocols.
IPX
Internet Packet Exchange. This routing protocol is based on Xerox’s XNS,
was developed by Novell, and is implemented in Novell’s NetWare.
IS-IS
Intermediate-Station-to-Intermediate-Station. This is an OSI link-state
routing protocol that routes both OSI and RIP traffic.
NCP
NetWare Core Protocol. This is Novell’s NetWare specific routing protocol.
OSPF
Open Shortest Path First. This is a TCP/IP link-state routing protocol that
provides an alternative to RIP.
RIP
Routing Information Protocol. This routing protocol is used in TCP/IP,
XNS, and IPX. Under RIP, a message is broadcast to find the shortest route
to a destination based on a hop count.
RTMP
Routing Table Maintenance Protocol. This is Apple’s distance-vector
protocol.
SPF
Shortest Path First. This link-state routing protocol uses a set of
user-defined parameters to find the best route between two points.
routers
399
7.5 Filtering
The filtering capability of routers is primarily thought of as a mechanism to
implement security. Although filtering does indeed provide a mechanism to
implement network security, it also provides a mechanism to regulate network
traffic. Through the regulation of network traffic you may be able to predefine
routes for particular types of packets, protocols, and addresses, as well as
different combinations of the preceding.
Another area where filtering has achieved a considerable degree of use is
in providing a mechanism to differentiate the service provided to different
classes of traffic. This is accomplished by associating filters with different
router queues and enabling a specific router queuing method. Figure 7.18
illustrates how you could use an outbound filter to place all UDP traffic
using a predefined destination port number or IP address into queue 1, while
all other traffic is routed into queue 2. If queue 1 is always serviced when
it is occupied, this would prioritize UDP traffic over all other traffic. If
UDP is used to transport digitized voice, in effect this filtering and queuing
scheme provides a Quality of Service (QoS) capability from the router into
the network.
The filtering capability of routers is highly dependent upon this functionality as well as the ingenuity of the router manager or administrator. To illustrate
the latter, we will define a few generic router filtering functions to illustrate
how those functions can be applied to achieve a variety of filtering results
that could satisfy different organizational requirements. Although the router
filtering functions we will define are not applicable to a specific router, most
routers will support the functions we will cover. Thus, the filtering examples
presented in this section represent practical examples that illustrate how you
can control network traffic. In Chapter 9, when we devote an entire chapter to
the issue of security, we will turn our attention to the creation and operation
of Cisco router access lists. Until then, we will conclude this chapter with
generic examples of the use of router filtering.
The key to the functionality of a router’s filtering capability is the router’s
ability to look inside the composition of packets. Most routers, at a minimum,
provide filtering based upon the examination of the contents of the destination
and source addresses transported by a packet. Other routers provide a filtering
capability based upon the Ethernet protocol value carried in the type/length
field, and the DSAP and SSAP values carried in the data field. This additional
capability provides you with the ability, for example, to enable or disable
Novell IPX traffic between certain router ports, enable or disable TCP/IP traffic
400
chapter seven
Wide
area
network
Router
Outbound traffic
queues
UDP
Q1
All other
traffic
Q2
Figure 7.18 Using an outbound filter to prioritize UDP traffic with a predefined destination port.
between the same or different router ports, and regulate other protocols that
may be carried to and from an Ethernet LAN.
Another common filtering capability associated with routers that support
the TCP/IP protocol stack is the ability to filter based upon TCP and UDP
well-known ports. For example, by blocking outbound packets with a TCP
port address of 80 an organization could bar HTTP traffic, in effect prohibiting
employees from surfing the World Wide Web.
Figure 7.19 illustrates the connection of an Ethernet network to a four-port
router. In this example, ports 1, 2, and 3 can represent connections to other
routers via wide area network transmission facilities or to other LANs a short
distance away. Since we are primarily interested in how we can use filtering
as a mechanism to regulate traffic, it is more important to focus upon the
flow of data between router ports than the devices connected to each hub.
However, since we need a point of reference, we will discuss traffic routed to
and from the Ethernet network connected to port 0 in Figure 7.19.
routers
401
Ethernet
1
0
2
Router
Figure 7.19 Through filtering you
can implement security as well as
regulate network traffic.
3
Filtering Expressions
Most routers perform filtering based upon specified patterns linked by logical
operators. Thus, you would first specify one or more patterns to be filtered,
and then link multiple patterns or prefix a single pattern with an appropriate
logical operator. Before actually creating a filter pattern, it is important to note
how the manufacturer of the router implements filtering. Some vendors enable
everything, requiring you to preclude certain data flows by filtering. Other
vendors inhibit everything, only permitting data flow based upon positive
filters. For our examples, we will presume the manufacturer of our router
permits or enables all data flow unless it is specifically precluded.
Filtering Examples
For our first example, let’s assume you have a UNIX server connected to the
Ethernet LAN and do not want IP traffic from port 2 to flow to the server.
Assuming P1 (pattern 1) = IP, originating port is PORT2, and destination port
is PORT0, you would set up the filter as follows:
Originate
Destination
P1 AND PORT2
PORT0
Thus, any IP frames received on port 2 and destined for port 0 would be
filtered or blocked.
Now let’s assume you do not want to transfer Ethernet broadcast packets
beyond that network. To do so you would set the pattern (P1) to recognize
a destination address of FF-FF-FF-FF-FF-FF, which is the Ethernet broadcast
402
chapter seven
address. Then, you would set up the filter as follows:
Originate
Destination
P1 AND PORT0
PORT1 OR PORT2 OR PORT3
For another example, let’s assume router filtering patterns and ports support
the use of the logical NOT operator. Then, you could set up the filter for the
preceding example as follows:
Originate
Destination
P1 AND PORT0
NOT PORT0
Router Access Lists
To illustrate the use of filtering for protecting stations on an internal corporate
network from potential hackers lurking on the Internet, consider Figure 7.20.
In that illustration several workstations and a server are shown connected to
an Ethernet LAN, which in turn is shown connected via a router to an ISP,
which in turn is connected to the Internet. Assuming the corporate Ethernet is
assigned the IP network address 206.172.31.0, let’s assume the workstations
and server have host addresses of .2, .3, and .4, as indicated in the lower
portion of Figure 7.20.
Many routers use what are referred to as access lists to enable filtering
operations on an inbound and outbound basis. To illustrate the potential use
of an access list, let’s assume we are using that list with the TCP/IP protocol
and each list entry has the following format:
Operation direction from IP address/port to IP address/port
Here ‘‘operation’’ is either enable or disable, while ‘‘direction’’ is either
inbound or outbound.
Let’s further assume that you can use the global asterisk character (∗) to
reference all. Then, assuming you want to restrict inbound traffic to the server
to Telnet operations from any IP address, you would code the following access
list entry:
ENABLE INBOUND ∗ /23 206.172.31.3/23
The preceding entry allows Telnet (port 23) from any IP address to the server
running Telnet at address 206.172.31.3 to its Telnet port. The reason a port
routers
403
Internet
Internet
service
provider
Router
Router
Server
206.172.31.2
206.172.31.3
206.172.31.4
Figure 7.20 Connecting a corporate LAN to the Internet exposes each corporate computer to a virtually unlimited number of hackers. Through programming router access lists this exposure can be minimized.
address is required at the destination results from the fact that both workstations and servers can support multiple simultaneous TCP/IP applications. For
example, a Web server can also support FTP as well as ping, finger, and other
TCP/IP applications.
For a second example of the use of an access list, let’s assume you wish to
bar all employees on your local LAN from sending Internet mail. Since SMTP
uses port 25, your access list entry would appear as follows:
DISABLE OUTBOUND 206.172.31 ∗ /25 ∗ /25
In this example we used the global character (∗) to replace the last digit in the
206 network dotted decimal position, in effect barring all stations (1 to 254)
on the network from sending data on port 25 to any IP address.
Although the preceding examples illustrate but a fraction of router filtering
capability, they illustrate several important concepts. First, by filtering on
source and/or destination addresses and protocols, it becomes possible to
404
chapter seven
enable or disable traffic. Secondly, by reaching into the frame router, filtering
permits you to enable or disable the flow of specific protocols. The latter is
an important consideration when connecting a LAN to the Internet, since
many organizations do not wish to provide access to their computer systems
connected to the LAN by anyone that has Internet access.
Through the packet filtering capability of a router you obtain the ability to
control the flow of data between and through router interfaces. As indicated
in this section, this control can be used to expedite traffic or as a security
mechanism. Concerning the latter, readers are referred to Chapter 9 for specific
information concerning security, including an examination of the types of
access lists supported by Cisco routers.
7.6 Performance Considerations
Regardless of the type of router and its protocol support and routing algorithm,
the processing required for its operation is considerably more than that
required for a bridge. This means that you can expect the packet processing
capability of routers to be considerably less than that of bridges.
High-capacity bridges can be expected to provide a forwarding rate between
100,000 and 200,000 packets per second. In comparison, most routers have a
forwarding capacity rated under 100,000 packets per second. Although this
may appear to indicate a poor level of performance in comparison to bridges,
note that only when functioning as a local bridge will a high-capacity bridge
actually use its full capacity. Otherwise, when used to interconnect remote
networks via a wide area transmission facility, a remote bridge will be able
to use only a fraction of its packet-processing capability for the forwarding
of packets over a relatively slow-speed WAN transmission facility. Similarly,
when routers are connected to a T1 or E1 line or to a frame relay service, they
may not be able to use their full packet forwarding capability. To see this,
refer to Table 7.2, which indicates the maximum packets-per-second transfer
capability of a router connected to five types of transmission lines, based
on five packet sizes. Note that a T1 line operating at 1.544 Mbps supports
a maximum transfer of 3015 64-byte packets per second. In actuality, since
the wide area network transmission facility results in the use of a WAN
protocol to wrap LAN packets, the actual PPS rate obtainable is normally 15
to 20 percent less than that indicated in Table 7.2. Thus, the forward rate of
most routers greatly exceeds the capacity of a single WAN connection. This
means that only a requirement for the use of multiple router ports should
make the forwarding rate of the router into a key equipment acquisition issue.
routers
TABLE 7.2
405
Maximum Packet Transfer Rates (Packets per Second)
Wide Area Network Transmission Facility
Packet Size (bytes)
56 Kbps
128 Kbps
256 Kbps
512 Kbps
1.544 Mbps
64
128
500
1000
1518
109
54
14
7
5
250
125
32
16
10
500
250
64
32
20
1000
500
128
64
40
3015
1508
386
193
127
Otherwise, the communication, transport, and routing protocols supported
are more important criteria for determining whether a vendor’s product can
support the requirements of an organization.
One key exception to the preceding involves the use of routers connected
to very high-speed transmission facilities, such as a T3 line operating at
approximately 45 Mbps or a synchronous optical network (SONET) connection operating at 155 Mbps. In such situations, the forwarding rate of the
router would be an extremely important consideration.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter eight
Wireless Ethernet
The purpose of this chapter is to make readers familiar with a relatively new
method of constructing or expanding an Ethernet network using the ‘‘ether’’
as the transmission media. Referred to as wireless Ethernet, the technology
we will examine in this chapter is also commonly known as wireless LANs,
802.11 networks and ‘‘Wi-Fi,’’ with the latter a term used to denote wireless
fidelity, a new term used for fast wireless LAN technology that is a play of
words on a stereo system.
8.1 Overview
The IEEE 802.11 standard for wireless LANs represents a base that has evolved
over the years. The original 802.11 standard dates to 1997 and resulted in
three physical layers being standardized to transport Ethernet frames over
the air. One physical layer is for infrared transmission, while the other two
physical layers provide support for radio frequency (RF) communications
through the use of Frequency Hopping Spread Spectrum (FHSS) and Direct
Sequence Spread Spectrum (DSSS) that were originally developed to resist
jamming in a military environment.
Figure 8.1 illustrates the architecture associated with the original IEEE
802.11 standard in comparison with the lower two layers of the ISO OSI
Reference Model. In examining Figure 8.1 you will note that the media access
control method used by the IEEE 802.11 standard is not the basic CSMA/CD
access protocol used by wired Ethernet. Instead, a protocol referred to as
Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) is used.
Later in this chapter we will turn our attention to how CSMA/CA operates.
The original IEEE 802.11 standard supports operations at either 1 Mbps or
2 Mbps under each of the three physical layers. Recognizing that infrared
is rarely used and data rates of 1 Mbps and 2 Mbps are not sufficient for
many modern applications in a shared over-the-air environment resulted in
407
408
chapter eight
ISO OSI
Reference Model
IEEE 802.11 layers
Logical link control
Data
link
layer
Media access control (CSMA/CA)
Physical
layer
FHSS
DSSS
IR
Legend:
FHSS
DSSS
IR
Frequency Hopping Spread Spectrum
Direct Sequence Spread Spectrum
Infrared
Figure 8.1
IEEE 802.11 architecture.
several additions to the IEEE standard. One addition was the IEEE 802.11b
specification, which extended the operating rate of DSSS to 5.5 Mbps and
11 Mbps and which represented the most popular type of wireless LAN
when this book revision occurred. Both the basic 802.11 and the 802.11b
specifications operate in the 2.4 GHz unlicensed Industrial Scientific and
Medical (ISM) band. While the Federal Communications Commission (FCC)
in the U.S. regulates the maximum power and transmission method, the fact
that the ISM band is unlicensed means that a user does not have to obtain a
license to use equipment in that frequency band.
A second addendum to the IEEE 802.11 standard is the 802.11a specification.
This specification defines the use of a multi-carrier frequency transmission
method in the 5 GHz ISM band. The multi-carrier frequency method is referred
to as orthogonal frequency division multiplexing (OFDM), which results in
a large number of carriers being used, each of which operates at a low data
rate, but cumulatively they support a high data rate up to 54 Mbps. Because
higher frequencies attenuate more rapidly than lower frequencies, the range of
802.11a-compatible devices is significantly less than that of 802.11b devices.
This results in a requirement to install additional access points to obtain the
same area of wireless LAN coverage and increases the cost of a very high speed
wireless LAN. Because many network operators require more speed than that
wireless ethernet
409
provided by the 802.11b specification but a higher range than that supported
by the 802.11a specification, the IEEE has been working on a new standard,
referred to as 802.11g, which doubles the data rate of 802.11b networks to
22 Mbps in the 2.4 GHz frequency band.
Network Topology
The IEEE 802.11 wireless LAN standards support two types of network
topology, referred to as ad hoc and infrastructure. Figure 8.2 illustrates an
example of an ad hoc network. An ad hoc network consists of two or more
wireless nodes or stations that recognize one another and communicate on a
peer-to-peer basis within their area of RF or IR coverage. The term ‘‘ad hoc’’
is assigned as this type of network environment is commonly formed when
two wireless devices come into the range of one another and communicate on
a temporary basis until one or more devices depart the area.
A second type of wireless LAN topology is known as a network infrastructure. In its most basic form a wireless network infrastructure consists of an
access point (AP) connected to a wired LAN and one or more client stations.
Figure 8.3 illustrates an example of a wireless network infrastructure. In this
example an access point is shown connected to a hub on a wired LAN. The
access point can be considered to represent a bridge between the wired and
wireless LANs. However, in addition to providing bridging between the wired
and wireless networks, an access point also interconnects wireless clients.
That is, when an access point is present, client stations communicate with
one another through the AP and not on a peer-to-peer basis.
Station
Station
Station
Figure 8.2 A wireless ad hoc network infrastructure.
410
chapter eight
Wired hub / switch
Client
station
Access
point
Basic service area
Figure 8.3 A wireless network infrastructure contains at least one access
point and one wireless station, referred to as a Basic Service Set.
When two or more mobile nodes come together to communicate or if one
mobile client comes into close proximity to an access point, this action results
in the formation of a Basic Service Set (BSS). Each BSS has an identification
that typically corresponds to the 48-bit MAC address of the wireless network
adapter card. That identification is referred to as a Basic Service Set Identification (BSSID) and the area of coverage within which members of a BSS can
communicate is referred to as a Basic Service Area (BSA).
When wiring an office, college campus or government agency, you will
more than likely need to install multiple access points. When this is done, the
basic service areas of coverage from multiple Basic Service Sets form what is
referred to as an Extended Service Set (ESS). The wired LAN infrastructure
functions as a distribution system, which enables clients to roam and be
serviced by different APs. Figure 8.4 illustrates an Extended Service Set
formed by the use of two access points interconnected by a wired LAN used
as a Distribution System (DS). Each BSS within a DS is said to be operating in
an infrastructure mode.
In examining Figure 8.4 it should be noted that the Basic Service Sets may
or may not overlap. In addition, each station associates itself with a particular
access point based upon selecting the one with the greatest received signal
strength. Each access point in the Extended Service Set will have an ESSID
(Extended Service Set Identifier) programmed into it. The ESSID can be
considered to represent the subnet the access point is connected to. You can
wireless ethernet
411
ESS
Hub
Hub
Distribution
system
Hub
Server
Client
Access
point
BSS-1
Client
Client
Access
point
BSS-2
Legend:
BSS Basic Service Set
ESS Extended Service Set
Figure 8.4 An extended service set consists of one or more basic service sets
connected via a distribution system.
also program the ESSID into a station, which then requires it to connect to a
like programmed access point.
When creating an extended service set it is also important to consider
the frequency of operation of each access point. This is due to the need to
minimize the overlapping of frequency use by adjacent access points. Because
FHSS and DSSS operating access points have different restrictions concerning
frequency overlap, you must also consider the transmission scheme used
when you design a large wireless infrastructure.
Roaming
In examining Figure 8.4 note that the movement of a client from BSS-1 to BSS2 or vice versa represents a roaming action. Although IEEE 802.11 wireless
LANs support roaming, a wireless operational LAN environment is commonly
a fixed-location environment in comparison to cellular telephones, which are
used anywhere from a reception area, to the office, and even in the powder
room. Thus, while 802.11 wireless LANs support roaming, the actual degree
of this activity is limited in comparison to a cellular telephone.
412
chapter eight
As a mobile client moves from one access-point service area to another, a
mechanism is required for one AP to drop the user while the other begins
servicing the user. A mobile client will typically monitor the signal-to-noise
ratio (SNR) as it moves and, if required, scan for available access points
and connect to a desired AP. APs periodically transmit a beacon frame that
enables clients to note the presence of one or more APs and select the
one with the best SNR. However, the actual method used depends upon
a vendor’s implementation method. For example, in a Cisco wireless LAN
roaming environment a client will become associated with a new access point
when three conditions occur. First, the signal strength of the new access
point must be at least 50 percent. Second, the percentage of time the client’s
transmitter is active is less than 20 percent of the present access point. The
third condition requires the number of users on the new access point to be
four fewer than on the present access point. If the first two conditions are not
met, then the client will not change access points regardless of the number of
users associated with the AP.
Physical Layer Operations
As discussed earlier in this chapter, the original IEEE 802.11 wireless LAN
standard supports a choice of three physical layers — infrared and two radiofrequency layers. The infrared physical layer is based upon the use of pulse
position modulation (PPM) at peak data rates of 1 Mbps, with an optional
2 Mbps rate. Because infrared is limited to use within a single room without
barriers, its use is severely limited. In fact, this author is not aware of
any infrared-based 802.11 LANs. Because of this, in this section we will
focus our attention upon the RF physical layers. Both Frequency Hopping
Spread Spectrum (FHSS) and Direct Sequence Spread Spectrum operate in
the 2.4 GHz ISM band, which represents a worldwide-recognized unlicensed
band. However, it should be noted that the actual frequencies for the 2.4 GHz
band can vary from country to country, as noted in Table 8.1.
FHSS
Under Frequency Hopping Spread Spectrum data is transmitted for a short
duration, referred to as dwell time at a single frequency. At the end of that time
duration the transmitter shifts to a new frequency and resumes transmission.
Thus, a FHSS system uses narrow-band data transmission but changes its
frequency periodically to create a wide-band transmission system.
Figure 8.5 illustrates an example of how an FHSS system hops for predefined
time intervals using different center frequencies based upon a predefined
wireless ethernet
TABLE 8.1
413
2.4 GHz ISM Frequency Allocation
Region
Allocated Frequency
United States
2.400–2.4835
Europe (except France/Spain)
2.400–2.4835
Japan
2.4710–2.4970
France
2.4465–2.4835
Spain
2.4450–2.4750
f5
Frequency
f4
f3
f2
f1
f0
Time
Figure 8.5 A frequency hopping spread spectrum system hops at a fixed time
interval, known as the dwell time, around a wide band using different center
frequencies in a predefined sequence.
algorithm. By only dwelling at one frequency for a short time duration,
an FHSS system can alleviate the effect of narrow-band noise occurring in
portions of the transmission band.
Although a military system based upon FHSS keeps the algorithm used
for hopping a secret, in the wonderful world of wireless LANs the hopping
sequence is well known. In fact, both the frequencies at which hopping occurs
as well as the number of hops within an assigned ISM band are commonly
414
chapter eight
regulated to prevent a wireless LAN from interfering with other electronic
equipment. In the United States FHSS uses 79 channels, each 1 MHz wide. In
Japan, the number of channels is reduced to 23. For both locations channels
are selected according to a pseudo-random selection algorithm that requires
a dwell time of 20 ms per channel and all channels to be used prior to being
able to reuse a channel. Under the IEEE 802.11 standard 78 different hopping
sequences are defined. Each hopping sequence is referred to as a channel,
which can cause a degree of confusion if you scan the standard without
noting this relationship. At the physical layer FHSS uses two- or four-level
Gaussian Frequency Shift Keying (GFSK) modulation. Under two-level FSK
modulation each bit is encoded by the transmission of a distinct frequency
from two available frequencies. Thus, the bit rate is the same as the baud
or signaling rate and the 1 MHz bandwidth used for each short transmission
supports a data rate of 1 Mbps. When four-level GFSK is used, each pair (dibit)
of bits is encoded into one of four frequencies. Thus, the bit rate is twice the
baud rate, resulting in a data rate of 2 Mbps. The term ‘‘Gaussian’’ prefixes
FSK because the wave form is Gaussian filtered.
Now that we have an appreciation for FHSS let us turn our attention to how
DSSS operates.
DSSS
Under Direct Sequence Spread Spectrum (DSSS) a spreading code is used
to spread each bit to be transmitted such that a number of bits representing
each bit are actually transmitted. The spreading code used under the 802.11
standard is referred to as a Barker code and its use results in each bit being
replaced by 11 bits. At 1 Mbps Differential Binary Phase Shift Keying (DBPSK)
is used for modulation, resulting in each bit being represented by one of two
possible phase changes. Because 11 bits replace each data bit, the resulting
signal is spread over 11 MHz. At 2 Mbps Differential Quadrature Phase Shift
Keying (DQPSK) is employed as the modulation method, which results in
two bits being encoded into one phase change. When this modulation method
is used, the bit rate becomes twice the baud rate, which results in a 2 Mbps
data rate.
Table 8.2 gives an example of DSSS coding using a five-bit sequences from
a pseudo-random bit generator. Note that data for transmission is simply
logically modulo-2 added to obtain the data stream to be modulated.
Upon demodulation the same pseudo-random bit sequence is modulo-2
subtracted to obtain the original setting of the bit that was spread. If a
transmission error occurs, the receiver simply selects the most popular bit
wireless ethernet
415
TABLE 8.2 DSSS Bit Spreading Example using a Five-bit
Spreading Code
Data bits
1
0
Five-bit spreading code
10110
01001
Modulo-2 addition (data to be modulated)
01001
01001
Demodulated data
01001
01001
Five-bit spreading code
10110
01001
Modulo-2 subtraction
11111
00000
setting. That is, a if a five-bit spreading code was used and as a result of the
modulo-2 subtraction process at the receiver, three bits were set to a value of
1 while two were set to a value of 0, the bit setting would be assumed to be 1.
Similar to FHSS, the use of DSSS can vary by location. Under the IEEE
802.11 standard the use of 13 DSSS channels is defined for transporting
an 11-bit Barker-coded 22 MHz signal. For operation in the United States
the 802.11 standard defines the use of 11 independent channels. Although
Europe and many Asian countries permit the use of 13 channels, in Japan the
small amount of available bandwidth (see Table 8.1) results in the support of
a single channel. Table 8.3 lists the carrier-frequency channel assignments.
As previously noted, depending upon the physical location of a DSSS system
a subset of available channels may be required to be used.
In the United States and Europe DSSS channel definitions permit three
frequency-isolated channels available for co-location. An example of channel co-location is illustrated in Figure 8.6. This frequency isolation enables
organizations to operate up to three DSSS functioning access points within
close proximity to one another without one access point interfering with
another.
High-Speed Wireless LANs
There are two extensions to the basic IEEE 802.11 standard for which equipment had reached the market when this book revision was performed. Those
extensions are the IEEE 802.11b specification, for which equipment conforming to that standard dominates the market, and the IEEE 802.11a specification.
Although the modulation methods differ for each method, they use the same
416
chapter eight
TABLE 8.3
Channels
Channel
2.4 GHz
2.4 GHz DSSS
Frequency (MHz)
1
2412
2
2416
3
2422
4
2427
5
2432
6
2437
7
2442
8
2447
9
2452
10
2457
11
2462
12
2467
13
2473
Frequency
2.4835 GHz
Figure 8.6 DSSS supports up to three non-overlapping channels in the
2.4 GHz band.
access protocol, a topic we will focus our attention upon once we obtain an
appreciation of the two extensions to the basic IEEE 802.11 standard.
802.11b
Under the 802.11b extension to the IEEE 802.11 standard the data rate was
increased to 5.5 Mbps and 11 Mbps under DSSS operations. At the higher data
wireless ethernet
417
rates of 5.5 Mbps and 11 Mbps DSSS transmitters and receivers use different
pseudo-random codes. Collectively, the higher modulation rates are referred
to as Complementary Code Keying (CCK).
802.11a
Power
Under the 802.11a extension to the IEEE 802.11 standard orthogonal frequency
division modulation (OFDM) is employed in the 5 GHz frequency band. Under
OFDM multiple-modulated carriers are used instead of a single carrier, as
illustrated in Figure 8.7. Here each modulated signal is orthogonal to the
other modulated signals.
The term orthogonal describes the axis of the signals and the fact that they do
not interfere with one another. Because multiple signals are transmitted by a
single user, the carriers can be said to be multiplexed. Thus, the transmission
of multiple carriers at 90 degree angles to one another was given the term
OFDM. However, if you are familiar with the operation of DSL modems or
one of the first 9600 BPS analog dial modems, you are also probably aware of
the term ‘‘multitone’’ used to denote the use of multiple carriers. Thus, OFDM
can be considered to represent a multitone transmission scheme.
Under the 802.11a standard 48 data and four pilot carriers or a total of
52 carriers are transmitted within a 20 MHz channel. This action makes use
of the three blocks or bands of frequency allocated by the FCC for unlicensed
operations in the 5 GHz band. A 200 MHz band from 5.15 GHz to 5.35 MHz
has two sub-bands. The first 100 MHz in the lower section is restricted to a
maximum power output of 50 mW, while the second 100 MHz has a more
generous 250 mW maximum power output. A third band at 5.725 MHz to
5.825 MHz is designed for outdoor applications and supports a maximum of
1 W of power output.
Because the 5 GHz band has almost four times the bandwidth of the ISM
band, the developers of the 802.11a specification turned to OFDM to make
...
Frequency
Figure 8.7 Orthogonal frequency division multiplexing results in the transmission of multiple carriers, each modulating a small amount of data.
418
chapter eight
better use of available bandwidth. As previously mentioned, each 20 MHz
channel consists of 48 data subchannels and four used for pilot tones and
error correction, with each subchannel approximately 300 kHz wide.
Several different modulation methods are supported under the 802.11a
standard. Binary Phase Shift Keying (BPSK) is used to encode 125 kbps of
data per channel, resulting in a 6 Mbps data rate. When Quadrature Phase Shift
Keying (PSK) is used, the amount of data encoded increases to 250 kbps per
channel, which results in a 12 Mbps data rate. A 16-level quadrature amplitude
modulation method that encodes four bits per signal change permits a data
rate of 24 Mbps. At the ‘‘top end of the line’’ a 64-level QAM modulation
method is supported. 64 QAM can operate encoding either 8 or 10 bits per
signal change, permitting a maximum data rate of 1.125 Mbps per 300 Hz
channel. Because 48 data subchannels are supported per channel, this results
in a maximum data rate of 54 Mbps.
Although the 802.11a specification supports a much higher data rate than
the 802.11b specification, it is important to remember that higher frequencies
attenuate much more rapidly than lower frequencies. As a result of this, the
range of 802.11a equipment is probably half that of 802.11b products, which
means the radius of coverage of an 802.11a access point will be one-fourth
that of an 802.11b access point.
Access Method
Unlike wired Ethernet, which uses the CSMA/CD access protocol, wireless
Ethernet LANs use what is referred to as a distributed coordination function (DCF). DCF represents a modification of the Carrier Sense Multiple
Access/Collision Avoidance (CSMA/CA) protocol. Under the CSMA/CA protocol each station listens to the air for activity of other users. If the channel
it is tuned to is idle, the station can transmit. However, if the channel has
activity, the station will wait until transmission ceases and then enter a random back-off procedure. This action is designed to prevent multiple stations
from seizing the channel immediately after the completion of an in-progress
transmission. Under the distribution coordination function access method a
period of time referred to as the DCF interframe space (DIFS) determines if
a packet can be transmitted. That is, if the medium is sensed to be available
for a duration of time that exceeds the DIFS, a packet can be immediately
transmitted.
A second time interval that comes into play under the DCF access method is
the short interframe space (SIFS). Under the IEEE 802.11 standard a receiver
must transmit a positive acknowledgement (ACK) to the transmitter when a
wireless ethernet
419
packet is received error free. An ACK will be transmitted after the SIFS, which
is of less duration than the DIFS. This ensures that an ACK is transmitted
prior to any new frame being transmitted. If an ACK is not received within a
period of time, the transmitter will assume the frame was corrupted and will
re-transmit the frame at the first opportunity to do so.
Figure 8.8 illustrates the relationship of the DIFS and SIFS to the transmission of data. At the top of the illustration the transmitting device is assumed
to listen to the channel and observe no activity for at least one DCF Interframe
Space (DIFS) prior to transmitting a frame. The receiving device must then
wait one Short Interframe Space (SIFS) prior to acknowledging the frame.
A second device requiring the ability to transmit is shown in the lower
portion of Figure 8.8. This device is assumed to need to transmit a frame,
but listens to the channel and hears the transmission of the first device or
the acknowledgement of the receiver. The time from the frame being placed
onto the channel through the DIFS following the receiver’s ACK represents a
deferred access time. Because a transmission was sensed to be in progress, the
second device must wait a random period after the deferred access time. The
second transmitter sets an internal time to an integer number of slot times and
observes when the DIFS time expires. Upon the expiration of the DIFS time
the timer of the second transmitter decrements towards zero. If the channel
is still available when the timer decrements to zero, the second station can
Transmitting
station A
DIFS
Data
SIFS
ACK
Receiving
station B
Transmitting
station C
DIFS
Contention
window
Next
data
Deferred access
Time
Slot times
Back off
After defer
Legend:
DIFS DFC Interframe Space
SIFS Short Interframe Space
Figure 8.8 The CSMA/CA access protocol is based upon two key timers and
a back-off algorithm.
420
chapter eight
commence transmission. Otherwise, if the channel is used by another station
prior to the timer reaching zero, its setting is retained at its current value for
future use.
The Hidden Node Problem
Because radio-frequency communications can be easily blocked by obstructions, it becomes possible for one node to be placed in a situation where it
doesn’t hear another. When this situation occurs, another node would listen
to a channel and an obstruction hiding the transmission of another station
would make the node think it is available for use when it is actually occupied.
The result of this action would be a propagation of two radio waves that at
a distant point collide, preventing other nodes from using the channel. To
reduce the probability of collisions, a derivative of the CSMA/CA protocol
referred to as Virtual Carrier Sense (VSC) is used by the 802.11 standard.
Under VCS a station that needs to transmit information will first transmit a
Request to Send (RTS) frame. The RTS frame represents a relatively short control frame that contains the source and destination address and the duration
of the following transmission. The duration is specified in terms of the time
for the transmission of a frame carrying data and the acknowledgement of the
frame by the receiver. The receiver responds to the RTS frame with a Clear To
Send (CTS) control frame that indicates the same time duration information
as contained in the RTS control frame.
A station that receives either an RTS or CTS control frame will set its virtual
carrier sense indicator for the duration of the transmission. The VSC indicator
is referred to as the Network Allocation Vector (NAV) by the 802.11 standard
and serves as a mechanism to alert all other stations on the air to back off or
defer their transmission.
If a station transmitting an RTS frame does not receive a corresponding
CTS frame within a predefined period of time, the originator will assume a
collision has occurred. Then, the originator will listen to the channel and,
upon noting it is free, transmit another RTS frame. Once a CTS frame is
received, the originator will send a data frame. The receiver will then return
an ACK frame to acknowledge a successful transmission.
The use of RTS and CTS frames, while reducing the probability of collisions
occurring at a receiver from a station ‘‘hidden’’ from the transmitter, adds
overhead to the media access operation. Due to this, most manufacturers
disable this option by default, requiring network managers to enable it on
both client stations and access points.
wireless ethernet
421
8.2 Frame Formats
Similar to wired Ethernet, where there is one basic frame format, wireless
LANs also have a basic data frame format. However, wireless LANs also
support two additional types of frames. One type, referred to as control
frames, was briefly mentioned when we discussed the hidden node. The third
type of frame supported by wireless LANs is management frames, which are
used to exchange management information between stations at layer 2 but
which are not forwarded to upper layers in the protocol suite.
Data Frame
Figure 8.9 illustrates the format of the MAC data frame which is used to
transmit information between stations. This basic data frame contains nine
fields, with two fields subdivided into additional fields. As we will note later
in this section, several fields from this frame are used in other types of frames.
In examining Figure 8.9, you will note that the 802.11 frame permits a
body that can be up to 2312 bytes in length. Because the maximum length
Ethernet frame has a 1500-byte Information field, the wireless LAN frame
can transport a maximum wired Ethernet frame. However, because the bit
error rate on a radio link can considerably exceed that of a wired LAN, this
2
Frame
Bytes control
2
Duration/
ID
6
Address
1
6
Address
2
6
Address
3
2
6
0−2312 6
Sequence Address Frame
CRC
control
4
body
Fragment Sequence
number
number
2
Protocol
version
2
Type
4
1
1
1
To From More
Subtype
Ds Ds
frag
Figure 8.9
1
Retry
1
1
1
1
Pwr More
WEP Rsvd
Mgt data
The basic 802.11 MAC data frame format.
422
chapter eight
means that the probability of a bit error increases as the length of the wireless
frame increases. To compensate for this higher wireless bit error probability,
a simple fragmentation and re-assembly mechanism is included in the 802.11
standard and we will shortly examine this. To obtain an appreciation of the
manner by which the MAC data frame conveys information, let us turn our
attention to the use of the fields and subfields in the frame.
Control Field
The 16-bit control field consists of 11 subfields, with eight representing onebit fields whose setting indicates whether a specific feature or function is
enabled or disabled. In this section we will examine the use of each subfield
in the order they appear in the control field.
Protocol Version Subfield
The two-bit Protocol Version subfield provides a mechanism to identify the
version of the IEEE 802.11 standard. In the initial version of the standard the
value of the Protocol Version subfield is set to 0.
Type and Subtype Subfields
The Type and Subtype subfields consist of six bits that identify the type of
frame and its function or subtype. Bits 2 and 3 denote the type of frame.
Although the use of two bits permits four types of frames to be defined,
at the present time only three types are defined — management, control,
and data. The Subtype subfield consists of bits 4 through 7 and defines the
function of a specific type of frame. Table 8.4 lists the Type and Subtype
subfield values to include a description of what the values of the y-bit
positions indicate.
In examining the entries in Table 8.4 note that the previously mentioned
RTS, CTS and ACK functions represent the frames we briefly described earlier
and the format of which we will investigate later in this section. The Beacon
frame represents the frame an access point periodically generates to indicate
its presence to stations while probe frames are used to query the status of
a device.
ToDS
This 1-bit field is set to a value of 1 when the frame is addressed to an access
point for forwarding to the distribution system. Otherwise, the bit is set to a
value of 0.
wireless ethernet
TABLE 8.4
Type and Subtype Values
Type Value
b3 b2
Type Description
00
Management
0000
Association Request
00
Management
0001
Association Response
00
Management
0010
Association Request
00
Management
0011
Association Response
00
Management
0100
Probe Request
00
Management
0101
Probe Response
00
Management
0110–0111
00
Management
1000
Beacon
00
Management
1001
ATM
00
Management
1010
Disassociation
00
Management
1011
Authentication
00
Management
1100
De-authentication
00
Management
1101–1111
Reserved
01
Control
0000–0001
Reserved
01
Control
1010
PS-Poll
01
Control
1011
RTS
01
Control
1100
CTS
01
Control
1101
ACK
01
Control
1110
CF End
01
Control
1111
CF End + CF-ACK
10
Data
0000
Data
10
Data
0001
Data + CF-ACK
10
Data
0010
Data + CF-Poll
10
Data
0011
Data + CF-ACK + CF-Poll
10
Data
0100
Null Function (no data)
10
Data
0101
CF-ACK (no data)
10
Data
0111
CF-Poll (no data)
Subtype Value
b7 b6 b5 b4
Subtype Description
Reserved
423
424
chapter eight
TABLE 8.4
(Continued)
Type Value
b3 b2
Type Description
10
Subtype Value
b7 b6 b5 b4
Subtype Description
Data
0111
CF-ACK + Cf + Poll (no data)
10
Data
1000–1111
Reserved
10
Data
0000–1111
Reserved
11
Reserved
0000–1111
Reserved
FromDS
This field is used to indicate whether or not a frame was received from the
distribution system. If the frame was received from the distribution system
this 1-bit field is set to 1. Otherwise, this field is set to 0.
More Fragments Subfield
This subfield is one bit in length and denotes if more fragments follow the
current fragment. If the value of this field is set to 1, then one or more fragments
follow. If the value of this field is set to 0, then no fragments follow. Thus,
this field permits the originator to note whether or not a frame represents a
fragment and enables a receiver to reconstruct a series of fragments into a
complete frame.
To illustrate the frame fragmentation process, consider Figure 8.10. This
example shows a frame consisting of four fragments. To identify that the
frame was fragmented as well as to let the receiver reconstruct the fragmented
Physical data unit (PDU)
MAC Frame
HDR body CRC
MAC Frame
HDR body CRC
MAC Frame
HDR body CRC
MAC Frame
HDR body CRC
Fragment 0
Fragment 1
Fragment 2
Fragment 3
Figure 8.10
An example of frame fragmentation.
wireless ethernet
425
frame, fragments 0, 1 and 2 would have their More Fragments subfield values
set to 1 in the MAC header in each frame.
Under the IEEE 802.11 standard the fragmentation process is based upon a
simple send-and-wait algorithm. Under this algorithm the transmitting station
cannot send a new fragment until it either receives an ACK for the prior
segment or decides that the fragment was retransmitted a predefined number
of times and drops the entire frame.
Retry Subfield
The value of this one-bit subfield is set to 1 to indicate that the frame is a
fragment representing the retransmission of a previously transmitted fragment.
The receiving station uses this field to recognize duplicate transmissions that
can occur if an ACK frame is lost.
Power Management Subfield
The IEEE 802.11 standard defines two power modes that a station can be
in — Power Save or Active. A station that is Active when transmitting a frame
can change its power status from Active to Power Save.
The Power Management setting is used by access points, which continuously maintain a record of stations working in the Power Saving mode. The
access point will buffer frames addressed to those stations until either they
specifically request them via the transmission of a polling request or they
change their power status.
A second technique employed to transmit buffered frames to a station in its
Power Save mode of operation is obtained through the use of Beacon frames.
An access point periodically broadcasts frames that includes information
concerning which stations operating in a Power Saving mode have frames
buffered by the access point. The station uses the Beacon of information to
wake up and remains in an Active power mode while it transmits a polling
message to the AC to retrieve those buffered frames.
More Data Subfield
The purpose of the More Data subfield is to indicate if there are more frames
following the current frame. This one-bit field is set by an access point
to indicate that there are more frames buffered to a particular station. The
destination station will use this bit setting to decide if it should continue
polling or if it should change its power management state.
426
chapter eight
WEP Subfield
The Wired Equivalent Privacy (WEP) subfield indicates whether or not the
body of the frame is encrypted. WEP uses the RC4 encryption algorithm, which
is a stream cipher. As a reminder, a stream cipher operates by expanding a
short key into an infinite pseudo-random key stream. The transmitter XORs
(modulo-2 adds) the key stream to the plain text, resulting in the generation
of encrypted ciphertext. The receiver uses the same key to generate the same
sequence of pseudo-random bits, which are then modulo-2 subtracted from
the received ciphertext to reconstruct the plain text.
As we will note later in this chapter when we examine some wireless
equipment configurations, the WEP algorithm uses a pseudo-random number
generator that is initialized by a 40-bit key. Through the use of a 40-bit key
and a 24-bit initialization vector a 64-bit key is generated that, according to
many reports, is relatively easy to break. Although some products support
128-bit WEP keys, papers have been published that appear to indicate that the
extended key is also susceptible to being broken. Because only one bit is used
in the field to indicate whether WEP is enabled or disabled, all stations within
a BSS must be configured similarly with respect to WEP. That is, either all
stations and the access point within a BSS must have WEP disabled or they
must be configured to use the same key.
Order Subfield
The last position in the Control field is the one-bit Order subfield. The setting
of this bit is used to indicate that the frame is being transmitted using the
Strictly Ordered service class. This bit position was added to accommodate the
DEC LAT protocol, which cannot accept change of ordering between unicast
and multicast frames. Because the DEC LAT protocol is not exactly a popular
one for the vast majority of wireless applications, this subfield is ignored.
Now that we have an appreciation of the subfields within the control field,
let us continue our tour of the MAC data frame.
Duration/ID Field
This two-byte field indicates either the station identification (ID) or the
duration in microseconds requested to transmit a frame and its interval to the
next frame. The actual interpretation of the value stored in this field depends
upon the type of the frame. In a Power-Save Poll message this field indicates
the station ID. In all other types of frames the value in this field indicates the
duration in milliseconds requested to transmit a frame and its interval to the
next frame.
wireless ethernet
427
Address Fields
If you examine Figure 8.9 you will note the presence of four address fields,
labeled Address 1 through Address 4. This enables a frame to transport four
addresses, with the address carried in each address field based upon the
settings of the ToDS and From DS bits in the Control field.
Table 8.5 summarizes the type of address transported in each address field
based upon the values of the ToDS and From DS bits in the Control field.
In examining Table 8.5 note that Address 1 always indicates the recipient,
which can be the destination address (DA), Basic Service Set ID (BSSID), or
the Recipient Address (RA). If the ToDS bit is set, Address 1 contains the
AP address. When the ToDS bit is not set, the value of the Address 1 field
contains the station address. All stations filter on the Address 1 field as it
always indicates the recipient address.
Address 2 is always used to identify the station transmitting the frame. If
the From DS bit is set, the value contained in the Address 2 field is the AP
address. Otherwise the address represents the station address.
Moving on to the Address 3 field, you will note from Table 8.5 that it also
depends upon the ToDS and From DS bit settings. When the FromDS bit is set
to a value of 1, the Address 3 field contains the Source Address (SA). If the
frame has the ToDS bit set, then the Address 3 field contains the Destination
Address (DA).
The fourth and last address field, which is Address 4, is used for the special
situation where a wireless distribution system is employed and a frame is
being transmitted from one access point to another. In this situation both the
ToDS and FromDS bits are set. Thus, neither the original destination address
TABLE 8.5 The Settings of the ToDS and From DS Bits in the Control
Field Govern the Use of the Address Fields
ToDS
FromDs
Address 1
Address 2
Address 3
Address 4
0
0
DA
SA
BSSID
N/A
0
1
DA
BSSID
SA
N/A
1
0
BSSID
SA
DA
N/A
1
1
RA
TA
DA
SA
Legend:
TA = Transmitter Address
RA = Receiver Address
BSSID = Basic Service Set
428
chapter eight
nor the original source address is applicable and Address 4 is then limited to
identifying the source of the wireless DS frame.
Sequence Control Field
The two-byte Sequence Control field provides a mechanism to represent the
order of different fragments that are part of a frame. As previously illustrated
in Figure 8.9, the Sequence Control field consists of two subfields — Fragment
Number and Sequence Number. Those subfields are used to define the frame
and the number of the fragment that is part of a frame.
Frame Body Field
The Frame Body field is the field that transports information between stations.
As indicated in Figure 8.9, this field can vary in length up to 2312 bytes.
CRC Field
The last field in the MAC data frame is the CRC field. This field is four bytes
in length and is used to contain a 32-bit CRC.
Now that we have an appreciation of the composition of the MAC data
frame, let us turn our attention to the composition of several control frames.
Control Frames
As previously noted in this chapter, the IEEE 802.11 standard defines the
use of several types of control frames that govern access to the media as
well as provide acknowledgement of a received frame. In this section we will
examine the format and utilization of three control frames — RTS, CTS and
ACK. Figure 8.11 indicates the format of each frame.
RTS Frame
The RTS and CTS frames have a similar format, with the MAC header
contained in the Frame Control field for each frame. Concerning the RTS frame,
the Receiver Address represents the address of the wireless network station
that is the intended immediate recipient of the next data or management
frame. The transmitted address (TA) represents the address of the station
transmitting the RTS frame, while the Duration field contains the time in
microseconds required to transmit the next data or management frame plus
one CTS frame, one ACK frame, and three interval periods between frames.
wireless ethernet
429
RTS and CTS
Bytes
2
Frame
control
2
6
Duration
Receiver
address
6
Transmitter
address
4
CRC
ACK
Bytes
2
Frame
control
2
6
4
Duration
Receiver
address
CRC
Figure 8.11
Common control frames.
Because the RTS frame is generated by a transmitter requesting access to the
medium, it will be responded to by a CTS frame.
CTS Frame
The CTS frame has the same format as the RTS frame and the entry of data
in the fields of the frame forms a relationship between the two. That is, the
Receiver Address (RA) of a CTS frame is copied from the Transmitter Address
(TA) field of the received RTS frame. The value of the duration field is obtained
from the duration field of the previously received RTS frame less the time, in
microseconds, required to transmit the frame and the Short Interframe Space
(SIFS) interval. The Receiver Address and Transmitter Address for both RTS
and CTS frames are 48 bits in length and represent the address length used by
IEEE 802.3 wired LANs.
ACK Frame
A third commonly used control frame is the ACK frame, the format of which
is shown in the lower portion of Figure 8.11.
Similar to the CTS frame, several fields in the ACK frame contain values
based upon a previously received frame. For example, the Receiver Address
field value of the ACK frame is copied from the Address 2 field of the
previously received frame that the ACK acknowledges. A second example of
field relationships between frames concerns the setting of the More Fragment
bit in the Frame Control field of the previous frame. If that bit was set to 0,
the Duration field in the ACK frame is set to 0. Otherwise, the Duration field
value is obtained from the Duration field of the previous frame minus the time
in microseconds required to transmit the ACK frame and its SIFS interval.
430
chapter eight
Management Frames
As noted in Table 8.4, there are 10 defined management frames. Two of the
more popular types of management frames are Beacon and Probe frames, both
of which we will examine in this section.
The Beacon Frame
Figure 8.12 illustrates the basic format of the body of a Beacon and Probe
frame as well as the Capability field included in each frame.
When a client comes in range of an access point it will hear the periodic
broadcast of Beacon frames transmitted by the access point to indicate its
presence. In addition to notifying stations of the presence of the access point,
Beacon frames provide all stations within a BSS with synchronization information and power management support. Concerning the latter, as previously
noted clients can be in a Power Save or Awake mode. In the Awake mode
stations are fully powered on and can receive frames at any time. If a node
goes into a Power Save mode it must first inform the access point. Once in
the Power Save mode a station will periodically wake up to listen for beacons
that indicate that the AP has queued messages for it.
In examining the Parameter Set shown in Figure 8.12, note that a particular
parameter, such as FH, is only present if a station is using the applicable
physical layer. The IBSS parameter set is only present within Beacon frames
generated by stations within an IBSS, while TIM information is only present
within Beacon frames generated by an access point. Here the term IBSS
references an independent basic service set, which is a single BSS that
operates independently within an area.
Probe Response Frame
The Beacon can be considered to represent an advertisement that tells stations
an access point is alive. If a station notes the presence of a Beacon frame and
wants to join an existing cell it will transmit a Probe Request frame to an
access point.
The response to a Probe Request is conveyed by a Probe Response frame,
whose body is shown in the middle portion of Figure 8.12. Note that the body
is similar to the Beacon frame body; however, the TIM information element is
not present.
Capability Information Field
Within both Beacon and Probe frames is a capability information field. This
field consists of two bytes, with the first used to define eight one-bit subfields
Timestamp
Timestamp
B1
IBSS
B0
ESS
CF poll
request
CF
pollable
B4
Privacy
Beacon
interval
Beacon
interval
Figure 8.12
B3
B2
Capability information field
Information
Probe response
Information
Beacon frame
PBCC
B6
SSID
SSID
Channel
agility
B7
Supported
rates
Supported
rates
DS
EF
B15
FH
DS
EF
Parameter set
FH
Parameter set
Reserved
Beacon and probe frame bodies.
Short
preamble
B5
Capability
information
Capability
TIM
IBSS
parameter
set
IBSS
wireless ethernet
431
432
chapter eight
as indicated in the lower portion of Figure 8.12. The function of the capability
information field is to indicate requested or advertised capabilities. Under
the current Draft 8 version of changes to the 802.11 standard the second byte
remains to be defined.
Physical Protocol Data Units
The transfer of information in an IEEE 802.11 environment occurs using Physical Protocol Data Units (PPDUs). The composition and format of the PPDU
varies based upon the physical layer used. Thus, because the 802.11 standard
supports three physical layers, as you might expect there are three PPDU frame
formats. Because practical wireless LANs are restricted to RF communications,
we will focus our attention upon the protocol frames for FHSS and DSSS.
FHSS
Figure 8.13 illustrates the frame format for the FHSS physical layer. This
frame consists of an 80-bit preamble of synchronization bits in the repeating
80 bits
SFD
16 bits
PLCP preamble
96 bits
Length
12 bits
Signaling
4 bits
PLCP header
32 bits
CRC
16 bits
Payload
PPDU
Legend:
PLCP
PPDU
SFD
Physical Layer Convergence Protocol
Physical Protocol Data Unit
Start of Frame Delimiter
Figure 8.13
The FHSS frame format.
wireless ethernet
433
pattern of 0101. . .01. This pattern is used for signal detection and is followed
by a 16-bit Start of Frame Delimiter (SFD). The SFD field is followed by a
Physical Layer Convergence Procedure (PLCP) header, which includes three
subfields. The length subfield denotes the length of the payload in bytes, a
four-bit Signaling field that denotes the operating rate and a 16-bit CRC that is
used to detect errors in the header. Under FHSS, initial transmission occurs
at a 1 Mbps data rate until the signaling field value is read. As a refresher, a
1 Mbps data rate is obtained by using two-level GFSK while a 2 Mbps data
rate occurs through the use of four-level GFSK.
DSSS
In comparison to the FHSS frame format, the format for the DSSS frame is
slightly more complex. That frame format, which is illustrated in Figure 8.14,
uses a 144-bit Physical Layer Convergence Procedure (PLCP) preamble divided
into a 128-bit sequence used for signal detection and a Start of Frame Delimiter (SFD).
The PLCP header consists of four fields. Those fields include an eight-bit
Signal field, which indicates the data rate. Currently this field supports four
128 bits
SFD
16 bits
PLCP preamble
144 bits
Signal
8 bits
SERVICE
8 bits
PLCP header
48 bits
Length
16 bits
Payload
PPDU
Legend:
PLCP
PPDU
Physical Layer Convergence Protocol
Physical Protocol Data Unit
Figure 8.14
DSSS frame format.
CRC
16 bits
434
chapter eight
values that correspond to operating rates of 1, 2, 5.5, and 11 Mbps. Three
bits in the Service field are used to support a high-rate extension, indicating
the modulation method, if transmit frequency and symbol clocks are derived
from the same oscillator and if an extension to the Length field is in effect.
Concerning the Length field, that 16-bit field indicates the number of bytes
in the MAC layer Protocol data Unit (PDU). The fourth field is the CRC field,
which protects the Signal, Service and Length fields.
8.3 Deployment
When considering the deployment of wireless LAN technology, there are four
key hardware products you need to consider. Those products include wireless
PC network adapter cards, access points, a combined router–access point, and
a bridge.
Wireless PC Network Adapter Cards
The purpose of the wireless PC network adapter card is to turn a computer
into a participant on a wireless LAN. There are three common form factors by
which wireless LAN network adapter cards are fabricated. Those form factors
include manufacture as a Type II PCMCIA card (now referred to as a PC card),
as a PCI adapter card designed for insertion into the system unit of a desktop
computer, and as a self-contained unit. When fabricated as a self-contained
unit, the network adapter includes a USB connector, which facilitates its use
with the growing number of computers manufactured with such ports instead
of legacy parallel and serial ports.
Figure 8.15 shows a picture of the SMC Networks EZ Wireless PC Card. The
left portion of the card slides into a PC Card slot in a laptop or notebook.
The dark area on the right of the card represents a self-contained antenna
that protrudes from the card slot. This PC card is designed for use in an IEEE
802.11b ad hoc or infrastructure network environment and supports DSSS
radio frequency communications at 1, 2, 5.5, and 11 Mbps. As we will note
later in this chapter, this adapter supports wired equivalent privacy (WEP).
The installation of a wireless LAN adapter card or the cabling of a selfcontained network adapter via a USB bus cable turns a computer into a wireless
LAN client or station. Once software drivers are installed, you can normally
tailor the operation of the adapter card. Depending upon the manufacturer of
the adapter card, you may be able to select an ad hoc or infrastructure mode
of operation, enable or disable a power-saving mode of operation, select one
of 13 RF channels for DSSS operation, and enable or disable WEP. Typically,
wireless ethernet
435
Figure 8.15 The SMC networks EZ wireless PC card is designed for insertion
into a Type II PC slot in a notebook or laptop computer.
default values are selected for each option that minimize or eliminate the
need for user configuration; however, accepting default settings can result in
certain problems. For example, the default setting for WEP is disabled, which
means that all transmission is in the clear, a topic we will examine later in
this chapter.
Access Point
A second network component associated with wireless LANs is designed to
interconnect wired and wireless LANs. That network component is the access
point, which functions as a bridge between wired and wireless LANs.
Figure 8.16 illustrates a dual-antenna access point manufactured by SMC
Networks. The use of dual antennas permits the device to select a stronger
received signal at a particular point in time. This can be important, because in
a wireless LAN environment transmitted signals will hit different objects that
result in RF signals being reflected in different directions. Such reflections
result in a spread of signals being received over many paths, which is referred
to as multipath transmission. Through the use of dual antennas it becomes
possible to discriminate among reflected signals and select the most applicable
signal at a point in time.
The SMC Networks EZ Connect wireless access point shown in Figure 8.16
has an operating range up to approximately 1800 feet. However, the exact
range that can be obtained, as well as the data rate, depends upon the number
of obstructions between a client and an access point. In an office environment
436
chapter eight
Figure 8.16 The SMC networks 11 Mbps wireless access point supports up
to 64 users at a maximum range of 1800 feet.
where cubicles, doors and corridors may be fabricated out of metal it may be
difficult to achieve the stated maximum operating range.
The use of an access point resembles a two-port Ethernet wired bridge.
However, instead of two wired ports, the access point has one. For the SMC
Networks EZ Connect access point shown in Figure 8.16 a built-in RJ-45 port
provides cabling for a connection to an IEEE 802.3 10 Mbps network or a
10/100 Mbps auto-negotiation port on a hub or switch. The other port on the
access point is its dual antenna, which provides an IEEE 802.11b network
connection. According to the vendor, the access point can support up to
64 wireless users and, like its wired cousin, the access point is a ‘‘plugand-play’’ device that automatically learns MAC addresses on the wired and
wireless sides.
wireless ethernet
437
Combined Router/Access Point
Recognition of the growth in the use of DS Land Cable modems as a mechanism
to access the Internet resulted in many hardware developers combining a
router and access point into a common housing. Through the use of this
combined housing it becomes possible to extend a single Internet connection
for use by multiple wireless clients.
Figure 8.17 illustrates the SMC Networks Barricade broadband router, which
combines an access point, a router, and a three-port 10/100 Mbps Ethernet
switch into one housing. A fourth Ethernet port is used to provide a connection
to a DSL or cable modem. The use of this device or similar products from
other vendors provides a high degree of networking flexibility. For example,
you could cable this device to a DSL or cable modem and use it to provide
clients with wireless access to the Internet. As an alternative, you could cable
one of the built-in switch ports to your existing wired hub or switch and
obtain the capability for wireless clients to access the Internet or your wired
infrastructure. In addition to supporting both wired and wireless LANs, the
Barricade broadband router includes an asynchronous interface that enables
an ISDN dial connection to the Internet to be shared. Because wireless clients
commonly do not have their own printer, the Barricade also functions as a
print server and includes a printer interface on the device.
Network Address Translation
Because Internet Service Providers (ISPs) only issue one IP address per DSL
or cable modem connection, a mechanism is required to share that address
Figure 8.17 The SMC Networks Barricade broadband router consists of an
access point, a router and a three-port 10/100 Mbps Ethernet switch in a
common housing.
438
chapter eight
among many wireless clients as well as any devices wired to the router. The
mechanism used is referred to as Network Address Translation (NAT). The
SMC Networks Barricade router is similar to other products used by this
author in that it uses a Class C network address defined in RFC 1918 behind
the router as a mechanism for translating to and from the single IP address
assigned to the DSL or cable modem connections.
RFC 1918 defines Class A, B, and C address blocks that are reserved for private network use. Because two or more organizations can use those addresses,
they are not unique and cannot be directly used on the Internet. However,
they can be used internally to an organization and only require a translation
mechanism to provide clients with the ability to access the Internet. That
translation mechanism is performed by the router by mapping each RFC 1918
Class C IP address into a high-value TCP or UDP source port number but not
altering the single IP address allocated to the DSL or cable modem connection.
When a response occurs, the source port value is returned as a destination
port value that enables the router performing NAT to change the IP address to
an applicable RFC 1918 Class C address.
Figure 8.18 illustrates an example of the use of a wireless router/access point
in an office environment. In this example the wireless router/access point is
shown connected to a DSL modem that provides access to the Internet.
The wireless router/access point is cabled to the wired network. Thus,
the wireless router/access point provides NAT to wireless clients as well as
Internet
DSL
modem
Wireless router
access point
Wired LAN
hub
ISP assigned
IP address
RFC 1918
translated
addresses
Client
Figure 8.18
Wired
stations
Client
Using a wireless router/access point.
wireless ethernet
439
wired clients on the existing wired LAN. Because most wireless router/access
point devices perform NAT using a single RFC 1918 network address, you are
limited to supporting a total of 253 stations for translation purposes. This is
because host addresses of 0 and 255 are not usable and one Class C address is
used by the router/access point.
Wireless Bridge
While an access point represents a bridge between wired and wireless Ethernet
networks, a wireless bridge represents a facility used to interconnect two wired
LANs via wireless transmission at line-of-sight distances up to approximately
10 miles. A wireless bridge consists of two components — an access point that
is cabled to the wired LAN and a directional antenna, which is commonly
mounted on the roof of a building and which is cabled to the access point
located in the building.
Now that we have an appreciation of the four common types of wireless
LAN devices, we will conclude this chapter by focusing our attention upon
the configuration of a wireless router/access point and wireless client station.
Router/Access Point Configuration
To illustrate the configuration possibilities when using a wireless router/access
point, this author used the SMC Networks Barricade broadband router.
Figure 8.19 illustrates the system status menu of the Barricade when viewed
through the Microsoft Internet Explorer browser. If you look at the address
entered in the browser, you will note that it is an RFC 1918 Class C address.
That IP address is 192.168.123.254 and represents one of 254 total Class C
addresses that will be mapped to the IP address assigned by the Internet
Service Provider (ISP) to the Internet connection.
If you focus your attention on the System Status portion of the display in
Figure 8.19, you will note under the column labeled Item the first row labeled
Remaining Lease Time and an associated value. This indicates that the IP
address on the Internet connection represents leased addresses provided by the
ISP using the Dynamic Host Configuration Protocol (DHCP). The other entries
in the system status area indicate the IP address that is leased, the subnet mask,
the gateway for the ISP and two DNS addresses. This initial screen provides
system status information; however, until you enter a correct password and
log into the device you cannot perform any configuration operations. Thus,
let us log into the router/access point by entering a system password, which
after the initial installation should be changed from the default of ‘‘admin.’’
440
chapter eight
Figure 8.19
tus screen.
The SMC Networks Barricade broadband router system sta-
Figure 8.20 The primary setup entry permits a user to define how the SMC
Networks Barricade broadband router will be connected to the Internet.
Figure 8.20 illustrates one of several Barricade screens you can use to configure the router/access point once you have logged into the system. After you
successfully log into the system, the left panel changes from a Status entry and
the password request shown in Figure 8.19 to a series of selection items. In
Figure 8.20 the Primary Setup entry was selected, which results in the display
wireless ethernet
441
of a screen that enables you to select the type of Wide Area Network (WAN)
connection. Note that you can select a static or dynamic IP address, PPP over
Ethernet or dial-up, the latter enabling wireless clients to share a common
modem or ISDN connection to the Internet.
In concluding our brief examination of the Barricade’s configuration let
us turn our attention to its access list capability. Figure 8.21 illustrates the
Router Access Control menu. This graphic user interface provides you with
the ability to enable or disable (block or permit) three groups and a default
group for all hosts not defined in a numbered group.
Blocking or permitting is based upon the entry of one or more port numbers,
such as 80 for HTTP and group members. The latter is defined by entering
the last digit or range of digits for the host portion of the RFC 1918 address
assigned to a wireless client. For example, to block Telnet for all hosts whose
IP address is in the range 192.168.123.100 to 192.168.123.199 you would enter
23 as the port value and 100–199 as the member value for a group number.
Client Configuration
In concluding this chapter we will examine several wireless client configuration screens, primarily focusing our attention upon security issues. Figure 8.22
illustrates the advanced display during the installation of software drivers for
an SMC Networks Wireless PC Card. Although there are seven entries you
Figure 8.21 The SMC Networks Barricade broadband router access control
configuration screen.
442
chapter eight
Figure 8.22 The SMC networks advanced client setup screen provides users
with the ability to modify the settings for seven properties.
can configure, the default settings are designed to get an organization up and
running without having to perform any configuration changes. However, from
a security viewpoint this can be dangerous, as we will shortly note.
The first entry in the property box in Figure 8.22 is Authentication Algorithm
and its setting defines how stations authenticate themselves. If you select the
default of Shared Key, this means that users authenticate one another by
using the same WEP encryption key. Unfortunately, by default the Use WEP
property value is set to ‘‘disabled,’’ which means that if you select all defaults
you have no authentication and no transmission security.
If you accept all of the default values during the setup of the adapter card,
you can use a utility program provided by the vendor to reconfigure one or
more settings at a later date. The main display of this utility program is shown
in Figure 8.23 and warrants a discussion as its Link Info tab provides both
valuable information for site selection of an access point and a mechanism to
enhance security.
If you examine the lower portion of Figure 8.23, you will note that the utility
program displays both Link Quality and Signal Strength information. If you
use a laptop or battery power to roam over an office area, you can determine
wireless ethernet
443
Figure 8.23 The utility program
provided with SMC networks
wireless PC cards provides a
mechanism to monitor link quality
and signal strength.
if the location of an access point is sufficient to provide a good-quality signal
to various station locations or if the access point should be moved. Because
wireless LAN transmission can easily ‘‘leak’’ out of a building, you can take
your laptop outside to determine if a snooper setting in the parking lot can
obtain a good signal. If so, you could consider shielding a portion of the walls
facing the parking lot or move the access point to a location where radiated
signals into the parking lot do not represent a problem.
In concluding our examination of the SMC Networks utility program let
us focus our attention on the WEP key. By default WEP is disabled, which
explains the reason why two men in a van were able to travel from one Silicon
Valley parking lot to another and ‘‘read’’ wireless LAN traffic. Figure 8.24
illustrates the default setting of WEP, which as indicated results in no encryption. SMC Networks, like other vendors, provides users with the ability to
define up to four shared keys. Thus, you might use one key when in one
office, another key when you travel to a remote office, a third key for home use
and perhaps a fourth key when you stay overnight at a hotel with a wireless
Internet portal.
Assuming you check the box to the left of Enable Data Security, your screen
display will appear similar to that shown in Figure 8.25. In this example the
SMC Networks utility program permits users to enter a pass phrase to generate
up to four WEP key settings. Other vendors require users to enter alphabetic
or hex values for key settings.
444
chapter eight
Figure 8.24 The default setting for
WEP is disabled.
Figure 8.25 The SMC networks
wireless adapter card utility program
supports the entry of a pass phrase
to create a WEP key.
There are two WEP settings that warrant discussion. A 64-bit key is generated
by entering 10 hex digits, which are extended by a 24-bit initialization vector
(IV) into the 64-bit key. Several papers have noted that this key is relatively
insecure as the IV is transmitted in the clear and capturing when it repeats
permits a frequency analysis to decipher the encoded text. A second key,
which is 128 bits in length, is proprietary to vendors. Thus, extended security
could lock your organization to a single vendor. Because many organizations
wireless ethernet
445
cannot wait for new security solutions, a bit of common sense and the
appropriate location of clients and access points can reduce the risk of
interception of data. First, enabling WEP makes it harder for interceptions to
read your traffic. Second, placing access points and clients so as to minimize
radiation into the parking lot can make it difficult, if not impossible, for
an unauthorized person to sit in a van during the day and monitor an
organization’s wireless traffic. Third, if you notice a van in the parking lot
with a directional antenna pointed towards your building you should realize
something is amiss.
Ethernet Networks: Design, Implementation, Operation, Management.
Gilbert Held
Copyright  2003 John Wiley & Sons, Ltd.
ISBN: 0-470-84476-0
chapter nine
Security
As indicated previously in this book, Ethernet primarily represents a layer 2
data link protocol used to transport information in a local environment. As
such, the design of Ethernet is focused upon the lower two layers of the
OSI Reference Model. This means that the basic structure of Ethernet is not
secure and an Ethernet network has a variety of vulnerabilities. Thus, it is
incumbent upon the higher layers in the protocol stack to provide security to
an Ethernet network.
In this chapter we will turn our attention to the general topic of security
and techniques that can be used to protect an Ethernet network. Because we
briefly described and discussed wireless LAN security in the form of Wired
Equivalency Privacy (WEP) in Chapter 8, our primary focus in this chapter is
upon wired Ethernet. However, because wireless Ethernet LANs are commonly
used to provide a connection to a wired infrastructure, our discussion of
security will be applicable to protecting wireless devices accessing a corporate
network or the Internet. In this chapter we will first examine the role of the
router, as it represents the first line of defense of a network. This will be
followed by discussing the operation of a firewall and virus scanners.
9.1 The Security Role of the Router
Because a router functions as a gateway between networks, it also represents
your first line of network defense. In this section we will turn our attention
to the security role of this vital communications device. In the first portion
of this section we will focus our attention upon router access control. The
rationale for this coverage is based upon the fact that most routers are
remotely administered. Thus, they can be remotely accessed by persons with
evil intentions as well as by authorized employees.
In the second portion of this section we will turn our attention to router
access lists, examining how they are created, their use, and limitations.
447
448
chapter nine
Because Cisco Systems has approximately 70 percent of the market for routers,
we will primarily describe and discuss access control and access lists with
respect to Cisco routers.
Access Control
Cisco’s Internetworking Operating System (IOS) controls the operation of
that vendor’s communications devices to include routers and switches. IOS
support a command line interface that can be accessed via the router console
port, a modem connection, or a Telnet connection. Of the three methods only
the first can offer the possibility of being truly secure as it requires a directly
connected terminal device, which can be co-located with the router in a
secure, locked area. Because the other two methods support remote access,
they theoretically permit a virtually unlimited number of persons to attempt
to access a router’s operating system and represent potential vulnerabilities
that you need to consider. However, prior to giving an explanation of the
three methods, a brief discussion of one of the reasons it is relatively simple
to attack a router is in order.
Address and Message Identity
Many organizations employ an IP addressing scheme which results in the
assignment of a .1 (dot one) network address to a router’s interface. Thus, if
you can use ping or nslookup to determine the IP address of a host, it is then
relatively easy to replace the host address with a ‘‘dot 1’’ address to determine
if the device at that address is a router.
A second technique that enables the location of a router is to ping a host for
which pings onto the host network are blocked. This action results in a router
returning a ‘‘destination net unreachable’’ message. If the previous methods
do not yield results, another popular method is to develop a Telnet scanner to
search for the prompt ‘‘User Access Verification,’’ which is returned by Cisco
routers when you Telnet to a router’s IP address that supports remote access.
You can note this by examining Figure 9.1, which indicates an attempted
access to a router whose IP address is 205.131.176.1. Note the prompt ‘‘User
Access Verification.’’
Cisco routers provide you with three tries to enter a password. If you do not
enter a correct password the router displays the message ‘‘% bad passwords’’
and terminates the Telnet session. This results in the display of the dialog box
shown in the middle of the display that informs the user that the ‘‘connection
security
Figure 9.1
449
Attempting to use Telnet to gain access to a Cisco router.
to host lost.’’ At this time you can manually reconnect and try another series
of passwords.
Although we did not break into the router in Figure 4.1, we had three tries
to do so. If we write a program to continuously try different combinations of
characters over a long weekend, it may be possible to gain access to the router.
Thus, let’s examine router access in detail.
Cisco EXEC Sessions
Access to a Cisco router results in a command line session that is referred to
as an EXEC session. For those familiar with UNIX, an EXEC session is similar
to a UNIX shell process.
There are two different types of EXEC sessions supported by Cisco
routers — user EXEC and privileged EXEC. A user EXEC session allows
a person to perform a limited number of functions, such as displaying
interface information, but does not allow the user to change the router’s
configuration. In a privileged EXEC session you obtain the ability to perform
router configuration operations as well as all of the functions available to the
user EXEC session mode of operation.
All logins to a Cisco router first result in an entry into the router’s user EXEC
mode. When this occurs you will encounter a greater-than sign (>) prompt.
That prompt follows the name configured for the router or, if no name was
assigned, the default ‘‘Router’’ is displayed. To enter the privilege EXEC mode
a person first enters the command ‘‘enable.’’ This action results in the prompt
‘‘Password:’’ which, when a password is successfully entered, places the user
in the privileged EXEC mode. When this occurs, the prompt changes to a
450
chapter nine
pound (#) sign — called a hash sign in the UK — as illustrated by the following
sequence for a router named Macon.
Macon > enable
Password:
Macon #
Password Protection
If you have initial access to a Cisco router you can assign a password for
initial access to the router with the password line command. The format of
this commands is:
Password text
Where ‘‘text’’ can contain any string of alphanumeric characters to include
spaces up to 80 characters in length. The alphabetic characters are case
sensitive, which provides a router administrator with the ability to create
passwords that would be extremely difficult and time-consuming to guess via
a dictionary or brute-force attack method. Unfortunately, in a haste to configure
a router for remote access and to keep things simple, many administrators use
such classic passwords as ‘‘bingo,’’ ‘‘tiger,’’ ‘‘router,’’ and even ‘‘Cisco.’’
To assign a password for the privileged command level you would use
a router’s enable password global configuration command. This means you
must first enter the privileged command level via the enable command. If
a password for the privileged mode was not previously set, you will not
be prompted for one. Unfortunately, this is a common omission when some
organizations receive their first few routers and rush to establish an Internet
connection. The absence of router passwords can be equated to walking
barefoot. While you might reach your destination unharmed, there is also a
distinct chance of a serious reversal of fortune! Once you enter the privileged
mode, you could use the enable password command as shown below to set
the password to supersecret.
enable-password supersecret
Telnet Access
Each Telnet port on a router is referred to as a virtual terminal (vty). Cisco
routers support a maximum of five vty ports that are numbered 0 through 4.
You can configure nonprivileged passwords for Telnet access via the use of
security
451
the password command. The following example illustrates the assignment of
the password 4hard2guess to vty ports 0 through 4.
line vty 0 4
login
password 4hard2guess
When a user Telnets to the router’s IP address, the router will prompt the
user to enter a password. The user–router interaction will be similar to that
shown below:
%telnet 198.78.46.1
Trying. . . .
Connected to 198.78.46.1
Escape character is ‘∧ ]’
User Access Verification
Password:
Once the user enters the correct nonprivileged password, the following prompt
is displayed:
Router>
At this point the remote user can use the enable command in an attempt to
access the router’s EXEC privileged mode.
Password Encryption
By default anyone who has privileged access to a router can view all of
the passwords set on the router. For example, suppose you assigned usernames techman and investor the passwords good4me and notgood4u as
shown below:
username techman good4me
username investor notgood4u
Without encrypting passwords anyone with privileged access to the router can
view each user’s password as well as the general user and enable passwords.
This means that it becomes possible for one user to access the router as another
user. If you configured your router to log activity to a server, you might be
452
chapter nine
tempted to blame the wrong party for the mess he or she creates, either on
purpose or inadvertently.
Prior to discussing password encryption, a few words about usernames
are in order. The use of the username command allows you to assign separate, unique, personalized login names for each administrator. By assigning
usernames to virtual terminal (vty) liens you can require persons attempting
to gain remote access to a router to know both a username and associated
password, which makes it more difficult for an uninvited user to gain access
to a router.
To encrypt router passwords you would use the service passwordencryption global configuration command. The format of that command is:
service password-encryption
It should be noted that Cisco’s password encryption scheme produces what
is referred to as a type 7 password because this digit is used as a prefix
to the encrypted password scheme. It should also be noted that a search
of the Web using ‘‘type 7 passwords’’ or ‘‘Cisco router password encryption’’ will provide several references to the fact that the encryption scheme
used is more accurately a simple scrambling method and that the resulting
scrambled passwords can be decoded quite easily. Due to this it is suggested that the enable secret command should be used to better protect the
enable password.
Using the Enable Secret Command
The use of the enable secret command uses Message Digest 5 (MD5), a one-way
cryptographic hash function to encrypt the password. This means that simple
re-scrambling will not suffice to obtain the source of the password. Instead,
someone must use a password generation program against the algorithm and
compare the result against the MD5 result to determine the clear value of the
password. This is obviously much more difficult than simply viewing stored
passwords. The following example illustrates the use of the enable secret
global configuration command:
Macon (config) #enable secret bingo7
Macon (config) #exit
Macon#show running-config
Building configuration. . .
enable secret 5 $e$K3ghM$ 4X. . .
security
453
Note that a prefix of ‘‘5’’ is used to identify the use of MD5. Because of
this, the resulting encrypted password is also referred to as a type 5 password. You should also note that, because MD5 is a one-way hash algorithm,
it cannot be used to encrypt all passwords, such as those used by the Network Time Protocol (NTP) and similar protocols that require access to clear
text strings.
You can considerably preclude dictionary and brute-force attacks by selecting applicable length passwords with an appropriate mix of characters. By
encrypting passwords you make it more difficult for one person to use the
access privileges of another. While each of these techniques is recommended
to provide a barrier to unauthorized persons gaining access to your organizational routers, you can also limit access via the use of an applicable
access list.
Access List Restrictions
Although we will discuss the use of Cisco access lists in considerable detail
in the second portion of this section, we can note that you can limit access or
even block all Telnet access. To do so you would use several IOS commands.
For example, you can use the transport command in conjunction with a line
command to disable the Telnet listener on your router. The following example
turns off the Telnet listener on all vty lines:
line vty 0 4
transport input none
You can also use the access-list and access-class commands to limit access
from specific IP addresses. For example, assume you want to restrict vty access
to your router to users on the 198.78.46.0 network. To do so you would use
the following IOS statements:
access-list 1 permit 198.78.46.0 0.0.0.255
!
line vty 0 4
access-class 1 in
In this example, the access-list statement permits packets with a source
address on the 198.78.46.0 network. As we will note when we examine access
lists in detail in the second portion of this section, 0.0.0.255 represents a
454
chapter nine
wildcard mask that functions in a reverse manner to a subnet mask, requiring
matches for ‘‘0’’ bit positions and a don’t care for ‘‘1’’ bit positions.
The access-class statement provides you with the capability to define the
access checks made in a particular direction. The format of this command is
shown below:
access-class list number {in|out}
where the list number is an integer between 1 and 99 and references the
direction an access list is applied. Because the list number is between 1 and
99, this also means that the access-class statement is only applicable to a
standard access list. The keyword ‘‘in’’ applies to incoming connections, such
as virtual terminals. In comparison, the keyword ‘‘out’’ applies to outgoing
Telnet connections, such as a person who gains access to your router’s EXEC
mode and attempts to initiate a Telnet connection to another host. Thus,
the preceding series of three IOS commands restricts the ability to connect to
the virtual terminals on the router (terminal lines 0 through 4) to hosts on the
198.78.46.0 network.
Protecting the Hardwired Connection
Because a connection to the console is accomplished by directly cabling
a terminal device to the port, it is assumed that the person operating the
terminal is authorized. Due to this, this type of connection is by default
provided EXEC mode access without a password. Because it is quite possible
for an unauthorized person to experiment with gaining full router access, it is
a good idea to set a user EXEC password on those ports. To do so you would
enter the following statements:
line console 0
login
password hard2guess4
exec-timeout 5 0
When you log onto the router its login prompt would be as follows:
User Access Verification
Password:
At this point you would have to enter the password to gain access to the
router. Because the default timeout for an unattended console is 10 minutes,
security
455
you may wish to lower this value. To do so you would use the exec-timeout
command whose format is shown below:
exec-timeout m s
where m is the time in minutes and n is the time in seconds of no activity
that forces a session timeout. In the above example we lowered the timeout to
5 minutes.
Table 9.1 provides a summary of Cisco security management related commands that we previously examined. While the exec-timeout command can
technically be applicable to supporting a policy to prevent persons from
connecting and hogging access to the router throughout the day, this author
also considers the command to enhance security management. Thus, this
command is included in the referenced table.
Considering SNMP
The Simple Network Management Protocol (SNMP) represents a potential
vulnerability because it uses community names that are test strings for access
control. All too often the default community name of ‘‘public’’ is used and
provides a mechanism for persons to attempt to exploit an opening in your
router’s defense.
TABLE 9.1
Cisco Security Management Commands
Command
Operational Effect
line console 0
Establish a password on the console terminal.
line vty 0 4
Establish a password for Telnet connections.
enable-password
Establish a password for access to the privileged
EXEC mode.
enable secret
Establish an enable secret password using MD5
encryption.
exec-timeout m s
Establish an idle timeout period in minutes and
seconds, which when reached terminates the
connection.
password text
Establish a password for access to user
EXEC mode.
service password-encryption
Scramble the display of passwords to protect
them from view by the use of the show
running-config command.
456
chapter nine
You can use the snmp-server community command to assign a community
name as well as restrict SNMP operations to read only. The format of this
command is shown below:
snmp-server community name {ro|rw} list number
where ‘‘name’’ represents a community name, ‘‘ro’’ represents read only, and
‘‘rw’’ represents read/write capability, so certain Management Information
Base (MIB) objects can be changed. ‘‘List number’’ represents an integer
between 1 and 99 that references a standard access list. Readers are referred
to Chapter 10 for detailed information concerning SNMP.
To illustrate an example of the use of the snmp-server community statement,
let’s assume you want to restrict access to SNMP reads to one host on the
198.78.46.8. To do so you would code the following IOS commands:
snmp-sever community yo4mo52be ro 1
access-list 1 permit host 198.78.46.8
Note that in the first statement we changed the default community name
to you4mo52be. Because you can configure the management station with
the changed community name, you should always use a hard to guess
name. In addition, note that we used the ‘‘ro’’ option to restrict access to
the MIB to read-only and used a list number of 1 to associate the first
command with the access list that only permits access from the hosts we
previously defined.
A second SNMP-related area you need to consider is the transmission of
traps. Although most SNMP commands are initiated by a management station
to an agent, an agent can be configured to transmit unsolicited packets to
the management station where certain predefined conditions occur. These
unsolicited packets represent SNMP traps.
There are three SNMP-related commands you can use to control traps. The
first is snmp-server trap-authentication, which enables a router to transmit
a trap message when it receives a packet with an incorrect community
string. This is a highly recommended command to use as it will alert the
management station to mistaken access attempts as well as brute force and
dictionary based attacks.
The second command is snmp-server trap-source, which indicates the
interface where traps originate. The third command, snmp-server host, permits you to associate an IP address or host name with a community name.
For example, assume you want to send traps to the host at 198.78.46.8 from
security
457
interface Ethernet1 using the community name yo4mo52be. You would then
code the following IOS statements:
snmp-server trap-authentication
snmp-server trap-source Ethernet1
snmp-server host 198.78.46.8 yo4mo52be
Now that we have an appreciation of router access control methods including
SNMP, let’s turn our attention to a more formal and detailed examination of
access lists.
Access Lists
An access list can be defined as ‘‘an ordered set of statements that permit or
deny the flow of packets across an interface.’’ As such, an access list represents
a filtering mechanism since the list parameters are matched against applicable
information in each packet.
Interface Considerations
The creation of an access list represents the first in a two-part process. The
second step is to apply the access list to a specific direction on an interface.
Concerning the direction, it is important to note that the flow of packets from
a router is considered to represent an outbound direction while the flow of
packets toward a router represents the inbound direction. This means you
must carefully consider the direction of packet flow in conjunction with the
devices whose packets you wish to filter. To understand this concept better,
consider Figure 9.2, which illustrates the direction of inbound and outbound
packet flow with respect to a router that has one serial and two Ethernet ports.
In examining Figure 9.2 note that if you want to block certain types of
packets arriving from the Internet you would apply an access list to the
s0 (serial 0) port in the inbound direction. In comparison, let’s assume you
want to block certain users on the Ethernet1 (e1) network from accessing the
Internet. You could apply an access list to either the e1 or s0 interface. If you
apply the access list to the e1 interface you would apply it in the inbound
direction. In comparison, if you wanted to use the s0 interface you would
apply the access list in the outbound direction.
Also note that if you want to block traffic from the 198.78.46.0 network
from flowing to the 205.131.175.0 network, you could either filter packets
inbound to interface e1 or outbound from interface e0. In comparison, if you
458
chapter nine
Internet
Router
E1
S0
E0
198. 78. 46. 0
Server
198 . 78. 46. 8
205. 131. 175 . 0
Legend:
Outbound packet flow
Inbound packet flow
Figure 9.2
Inbound and outbound packet flow.
wanted to block traffic on the 198.78.46.0 and 205.131.175.0 networks from
flowing to the Internet, you could apply an appropriate access list to interface
s0. This illustrates an important concept concerning access lists. That is, you
must carefully consider both the interface and the direction an access list is
applied to an interface, topics which we will examine in additional detail as
we expand our coverage of access lists.
Types of Access Lists
Cisco routers support two types of access lists — standard and extended. In
the first portion of this section we briefly looked at the use of a standard
access list to control hosts that could gain vty access to a router. Now we will
examine the format and use of access lists in much more detail. In doing so
security
459
we will primarily focus our attention upon IP access lists since IP is the only
protocol supported for use on the Internet.
Standard IP Access Lists
The format of a standard IP access list is shown below:
access-list [list #] [permit|deny] [source address] [wild-card-mask] [log]
In examining the format of a standard IP access list a few items require a
degree of elaboration. First, the term ‘‘access-list’’ requires a hyphen between
the two words. Second, the list number is a numeric between 1 and 99 for the
IP protocol and is used to identify the access list statements belonging to a
common access list. Because a standard IP access list number must be within
the range 1 to 99, the list number also defines the protocol that the access list
operates upon.
Keywords
The permit or deny keyword identifies whether filtering permits or prohibits
packets that match the source address and wildcard mask from flowing
through an interface. The keyword permit allows packets that match the
address criteria to flow through an interface, while the keyword deny sends
packets matching the specified source address into the bit bucket.
Source Address
For a standard IP access list the source address represents the IP address of
a host or group of hosts specified using dotted decimal notation. In actuality,
the specification of a group of hosts is accomplished through the use of a
wildcard-mask. Thus, let’s turn our attention to this topic prior to examining
several examples of the use of standard IP access list statements.
The Wildcard Mask
As previously noted in this section, the wildcard mask functions in a reverse
manner to a subnet mask. That is, this mask uses a binary 0 to represent a
match and a binary 1 to represent a don’t care condition.
To illustrate the use of a wildcard mask, assume your organization has
the Class C network address of 205.131.175.0. If no subnets were configured
for that network, each station would have a subnet mask of 255.255.255.0.
460
chapter nine
Here each binary 1 (255 is a byte of all 1’s) represents a match, while a 0 bit
represents a don’t care condition. Thus, specifying 255 in each of the first
three positions of the four position subnet mask field results in the TCP/IP
protocol stack matching the network address in each packet and not caring
about the host address. Because the Cisco wildcard mask functions in an
opposite manner with respect to binary 1s and 0s, the mask required to allow
all packets with a source network address of 205.131.175.0 would be 0.0.0.255.
Then, a standard access-list statement that would enable all packets with a
source network address of 205.131.175.0 would be coded as follows:
access-list 1 permit 205.131.175.0 0.0.0.255
Because the wildcard-mask represents the complement of the subnet mask,
you can determine the subnet mask and take its inverse. For example, a subnet
mask of 255.255.0.0 results in a wildcard mask of 0.0.255.255.
As another example of a standard access list, suppose you wish to permit
packets from hosts on the previously mentioned network as well as the
198.78.46.0 network but preclude hosts from the 198.78.22.0 network. Your
access list would be as follows:
access-list 1 permit 205.131.175.0 0.0.0.255
access-list 1 permit 198.78.46.0 0.0.0.255
Wait! Where is the deny statement, you ask? While you could add the
appropriate statement, each access list has an implicit ‘‘deny everything’’ at
the end of the list. Because access lists are processed top down, we only
need to permit what we want to permit as all else is denied. For those of
us who get tired of typing network addresses and wildcard masks, Cisco
access lists support two keywords that can simplify the address specification
process. A third keyword provides a logging function. Although most accesslist keywords are only applicable to extended access lists, these three are also
applicable to standard access lists; so let’s examine them now.
The Host Keyword
Host is used to denote an exact match and represents a shorthand notation
for a mask of 0.0.0.0. For example, assume you want to permit packets from
the host address 205.131.175.1. To do so you could code the following access
list statement:
access-list 1 permit 205.131.175.1 0.0.0.0
security
461
Because the keyword host signifies an exact match you can re-code the
previous access list statement as follows:
access-list 1 permit host 205.131.175.1
The Any Keyword
A second keyword supported by a standard access list is any. This keyword
is used as an abbreviation for a source address and wildcard mask of 0.0.0.0
255.255.255.255. To illustrate the use of host and any keywords, let’s assume
you want to deny packets from source address 205.131.175.1 and permit all
other addresses. The standard access-list statements to accomplish this would
be as follows:
access-list 1 deny host 205.131.175.1
access-list 1 permit any
If you observe the order of the two statements, we first explicitly deny
packets from the specified host. This is because processing occurs top down.
If we reversed the order, the effect would be to permit all packets, a condition
we would not want to occur in this situation.
The Log Keyword
The keyword log represents an addition to access lists that occurred in IOS
Version 11.3. When you include this keyword in an access list it results in the
logging of packets that match permit and deny statements in the access list.
After you apply an access list with the log keyword to an interface the
first packet that meets a condition in the list causes an immediate logging
message to occur. Thereafter, packets matching the access-list conditions are
either displayed on the console or logged to memory every five minutes.
Logged messages include the access-list number, whether the packet was
permitted or denied, the source IP address and the number of packets that
were matched during the five-minute interval succeeding the first match. For
example, assume the following standard IP access-list statement:
access-list 1 permit host 205.131.175.1 log
access-list 1 deny host 205.131.175.2 log
Assume that over a five-minute period host 205.131.175.2 issued 10 packets.
When the first match occurred, the following would be displayed on the router
462
chapter nine
console or logged to memory, with the actual logging controlled by the ‘‘logging
console’’ command:
list 1 deny 205.131.175.2 1 packet
Then, five minutes later the following display or logged message to memory
would occur:
list 1 deny 205.131.175.2 9 packets
The ability to log access-list matches provides you with both a testing and
alert capability. You can use logging to test the development of different
access lists by observing the resulting match of packets as different activities
are attempted. When used as an alert facility, you would scan the display
to locate repeated attempts to perform an activity you designed an access
list to deny. In this situation, repeated attempts to perform an activity the
access-list statement was developed to deny would more than likely indicate
a potential attack.
In examining the standard access-list format, you will note it only provides
a mechanism to filter based upon source address. In comparison, an extended
access list, as its name implies, extends the packet filtering capability.
Extended IP Access Lists
Through the use of an extended IP access list you obtain the ability to filter
based upon source and destination address, protocol, source and destination
port, as well as other options that significantly enhance your ability to permit
or deny packets flowing across an interface. The general format of an extended
IP access list is shown below:
access-list [list#] [permit|deny] [protocol] [source address source-wildcard]
[source port] [destination address destination-wildcard] [destination port]
[log] [options]
To obtain an appreciation of the operation of an extended IP access list, let’s
examine the use of each of the fields in the above format to include some of
the options supported by this statement.
The List Number Field
The list number field identifies statements that make up an access list as well
as the type of access list. An extended IP access list uses numbers 100 through
199, which enables 100 unique extended IP access lists to be defined.
security
463
The Protocol Field
The protocol field entry defines the protocol to be filtered. You can filter on
IP, TCP, and UDP, as well as on different routing protocols. When creating an
extended IP access-list statement, it is important to remember the relationship
of protocols within the TCP/IP protocol suite that form IP datagrams. That
is, an IP header is used to transport ICMP, TCP, UDP and various routing
protocols. This means that if you specify IP as the protocol to be filtered, all
matches against other fields will cause the packet to be either permitted or
denied regardless of whether the packet represents an application transported
by TCP or UDP, an ICMP message, or a routing protocol. This also means that
if you intend to filter based upon a specific protocol you need to specify that
protocol. Thus, you need to specify more specific entries prior to less specific
entries. For example, assume the following statements:
access-list 101 permit ip any any
access-list 101 deny tcp any host 198.78.46.8
Because the first statement permits IP from any host to any host, this means
you could not block TCP from any host to the host whose IP address is
198.78.46.8 as the second statement would never take effect. You would need
to reverse the order of statements, which is why more specific entries should
be placed ahead of less specific entries.
Source Address and Wildcard Mask Fields
The source address and wildcard mask for an extended IP access list function
in the same manner as for a standard IP access list. This means that you can
use the keywords ‘‘host’’ and ‘‘any’’ and avoid specifying a mask.
Source Port Number Field
You can use either an integer or a mnemonic to specify a port number. For
example, to specify the Web’s HyperText Transmission Protocol (HTTP) you
can either use 80 or http. In addition, for TCP and UDP you can use the keyword
operators lt (less than), gt (greater than), eq (equal), and neq (not equal).
Destination Address and Wildcard Mask Fields
The destination address and wildcard mask have the same structure as the
source address and wildcard mask. Thus, you can use the keywords ‘‘host’’
464
chapter nine
and ‘‘any’’ to specify a specific address or any destination address without
having to specify a mask.
Destination Port Number
Similar to the source port you can specify the destination port as a numeric or
a mnemonic. You can also use an operator to specify a range. The following
example illustrates the use of both a numeric and a mnemonic to block Web
surfing from any source to the 198.78.46.0 network:
access-list 101 deny tcp any 198.78.46.0 0.0.0.255 eq 80
access-list 101 deny tcp any 198.78.46.0 0.0.0.255 eq http
Options
Extended IP access lists support a wide range of options. One commonly
used option is ‘‘log,’’ which was described when we examined standard IP
access lists. Another commonly used option is ‘‘established.’’ This option is
only applicable to the TCP protocol and is employed to restrict TCP traffic in
one direction as a response to sessions initiated in the opposite direction. To
accomplish this, an access-list statement that contains the word ‘‘established’’
results in each packet being examined to determine if its ACK or RST bit
is set. If so, this condition indicates that the packet is part of a previously
established TCP connection.
To illustrate the use of the keyword ‘‘established’’ in an extended IP accesslist statement, let’s assume that the Ethernet network shown in Figure 9.2
has the IP address 205.131.175.0. To restrict TCP packets flowing from the
Internet to those in response to packets originating on the Ethernet network,
you would use the following access list statement:
access-list 101 permit tcp any 205.131.175.0 0.0.0.255 established
In examining the use of the keyword ‘‘established’’ there are several points
to note. First, although its use provides a mechanism to control access from a
router’s untrusted side based upon traffic originated from the trusted side, the
method used is rather primitive. That is, packets are considered to be kosher if
their ACK or RST bit is set, a condition easy for a hacker to overcome. Second,
‘‘established’’ is only applicable to TCP as there is no field in a UDP header
to indicate that a UDP Segment is a response packet. As we will note later in
this section, reflexive access lists and Context Based Access Control (CBAC)
provide more powerful mechanisms for controlling access via the untrusted
security
465
side of a router, based upon the flow of packets from the trusted side of
the router.
To facilitate the use of Cisco extended IP access lists, Table 9.2 provides a
list of commonly used keywords and a brief description of their use. As we
continue our examination of the use of IP access lists in this section, we will
examine the use of most of the keywords listed in this table. However, one
keyword deserves a brief mention now as it provides a mechanism to add
comments to an access list. That keyword is remark, which is placed after an
TABLE 9.2
Extended IP Access List Keywords
Keyword
Utilization
any
An abbreviation for an address and wildcard-mask
value of 0.0.0.0 255.255.255.255. This keyword is
applicable to both source and destination
address fields.
established
Causes a match if the ACK or RST bits are set in the
TCP header.
host
An abbreviation for a wildcard-mask of 0.0.0.0. This
keyword is applicable to both source and destination
address fields.
icmp-type
Provides a mechanism for filtering ICMP messages by
their message type. You can also specify the ICMP
message code (0 to 255).
port
Provides a mechanism to define the decimal number or
name of a TCP or UDP port.
protocol
Provides a mechanism to define a specific protocol for
filtering. The specified protocol can include one of the
keywords eigrp, gre, icmp, igmp, igrp, ip, ipinip, nos,
ospf, tcp or udp, or an integer between 0 and 255,
which represents an IP protocol.
precedence precedence
Provides a mechanism for filtering by the precedence
level name or number (0 to 7) in the IP Type of
Service field.
remark
Provides a mechanism for adding text comments for an
access list.
TOS tos
Provides a mechanism for filtering based upon the
name or service level in the IP Type of Service field.
466
chapter nine
access-list number and allows you to place a comment in an access list. The
following example illustrates its use:
access-list 101 remark allow TCP to our network
access-list 101 permit tcp any any
Rules and Guidelines
There are several general ‘‘rules’’ and guidelines you should consider when
creating and applying Cisco access lists. The rules we will examine represent
factual information, while the guidelines represent informative suggestions.
The factual information is based upon the manner by which access-list
processing occurs, while the information suggestions are based upon the
experience of this author.
Top Down Processing The evaluation of an access list results in a sequential
examination of statements, commencing at the top of the list. This means
you must carefully consider the order in which you place statements in an
access list.
Access List Additions New entries are automatically placed at the bottom of
an access list. This means it may be difficult or even impossible to change the
functionality of an access list. If so, you will then have to consider creating a
new list, deleting the existing list, and applying the new list to an interface.
Access List Placement You should consider placing extended access lists
as close as possible to the source being filtered. This minimizes the potential
effect of filtering on the dataflow on other interfaces. In comparison, consider
placing standard access lists as close as possible to the destination. This
results from the fact that a standard access list is limited to filtering on source
addresses. Thus, if you place the list too close to the source, it could block the
flow of packets to other ports.
Statement Placement Because an IP datagram is used to transport ICMP,
TCP, UDP, and various routing protocols, you should place more specific
entries in an access list before less specific entries. This ensures that the
placement of one statement before another does not negate the effect of a
statement appearing later in the list.
Access List Application An access list does not go into effect until it is
applied to an interface via an access-group command. Remember, until the
security
467
access list is applied to an interface, the list has no effect upon the flow
of packets.
Filtering Direction Use the filtering direction to define whether inbound
or outbound packets are filtered. Remember that packets flowing away from
a router’s interface are outbound, while packets flowing towards a router
are inbound.
Router-Generated Packets Packets generated by a router, such as a ‘‘destination net unreachable’’ message or a routing table update, cannot be filtered
by an access list applied to outbound traffic. However, you can control such
packets, either by filtering their arrival via an access list applied in an inbound
direction or by turning their generation off via an applicable IOS command.
Creating and Applying an Access List
Cisco routers support the creation of an access list via several methods. You
can create an access list directly from the console, via a vty connection, or via
a word processor or text editor. If you use a word processor or text editor you
would store the resulting file as ASCII text and use the computer on which
you stored the file as a server. To accomplish this you need to install on your
computer a trivial file transfer program (TFTP) that operates as a server. You
would then use the router as a client to upload or retrieve the previously
created file. Figure 9.3 illustrates the Cisco TFTP Server screen with its option
dialog box displayed. Note that this is a relatively simple program that allows
you to specify the TFTP server root directory and the name and location of
a log file, as well as whether or not the progress of file transfers is displayed
and logging is enabled.
To apply an access list to a router requires a three-step process. First, you
need to create an access list. Once it is created, you need to apply it to an
interface. Finally, you need a method to define the direction that the access
list will be applied to an interface.
Specifying an Interface You specify a router interface through the use of
the ‘‘interface’’ command. For example, to apply an access list to serial port 0
(previously shown in Figure 9.2), you would define the interface with the
following command:
interface serial 0
Similarly, to apply an access list to a router port connected to a LAN, you
would use the interface command with the name and port identifier of the
468
chapter nine
Figure 9.3 Cisco’s TFTP server program permits you to create access lists
that can be uploaded from a PC by a router.
connection. For example, assuming the router port was e0 shown in Figure 9.2,
you could enter the following:
interface ethernet0
Because you can abbreviate most keywords in a command, you could also use
e0 in the command as shown below.
interface e0
Using the IP Access-group Command The third step in the three-step
process is to define the direction of the interface that the access list works
on. To do so you would use the ip access-group command whose format is
shown below:
ip access-group [list number] [in|out]
Similar to other commands, the list number identifies a particular access
list. The keywords ‘‘in’’ and ‘‘out’’ identify the direction that the access list
will be applied. That is, if packets are flowing towards the router you would
use the keyword ‘‘in,’’ while you would use the keyword ‘‘out’’ if they are
flowing out of a particular router interface.
To illustrate an example of the three-step process, let’s create an access
list. In doing so, let’s assume our network configuration is that shown in
security
469
Figure 9.2. Let’s further assume that we want to restrict data originating from
the Internet to Web traffic bound for the host whose IP address is 198.78.46.8.
In addition, we want to allow users on the 205 and 198 networks shown in
Figure 9.2 to initiate Telnet and Web-based queries to hosts on the Internet
and allow the results of those queries to return via the Internet. To do so our
access list would be as follows:
interface serial0
ip access-group 110 in
access-list 110 remark allow TCP with ACK or RST bit set
access-list 110 permit TCP any any established
access-list 110 remark allow Web access to server
access-list 110 permit TCP any host 198.78.46.8
access-list 110 remark block everything else
access-list 110 deny ip any any
In this example the ‘‘interface’’ command is first used to define serial
port 0. Next, the ‘‘ip access-group’’ command is used to apply the access list
statements we will enter as access list number 110 in the inbound direction on
the serial interface. This is followed by the entry of six access list statements
that form the actual access list. Note that for internal documentation three
statements include remarks. Thus, let’s focus our attention upon the other
three statements in the access list.
The first non-remark statement permits TCP traffic that responds to queries
initiated from the internal Ethernet networks. This is due to the keyword
‘‘established’’ in the statement. The second non-remark statement permits
Web traffic (port 80) from any host on the Internet to the server whose IP
address is 198.78.46.8. The third non-remark statement is not really necessary
as an access list has a ‘‘deny all’’ implicitly placed at the end of the list.
In examining this access list it is important to note that it could create a
problem for Web surfers and other persons using the Internet. This is because,
although TCP traffic is permitted, all other traffic is blocked. This means your
internal network users cannot use DNS as it is carried via UDP, nor can they
use ping as ICMP messages are not permitted.
To allow UDP we could add the following statement before the
‘‘deny’’ statement:
access-list 110 permit UDP any any eq 53
470
chapter nine
In this statement port 53 represents DNS. Thus, instead of opening all UDP
access we will just allow DNS.
To allow pings we can permit echo replies to echo requests. Thus, we could
add either of the following statements to our access list:
access-list 110 permit icmp any any echo-reply
or access-list 110 permit icmp any any eq 0
Here the value 0 represents the ICMP type number for echo reply.
Limiting ICMP To make your TCP connection more secure you should
consider limiting the types of ICMP messages allowed to flow through your
router. As previously indicated in our short extended IP access list example
addendum, you can include a permit statement to allow a specific type of ICMP
message. If you do not allow any ICMP and do not include a permit for the IP
protocol, then all ICMP messages will be blocked. Before you block all ICMP
messages there are certain ICMP messages you should consider permitting
and others you may wish to block. Thus, let’s focus our attention upon those
messages. In doing so we will again assume the access list will be applied to
the serial interface in Figure 9.2 in the inbound direction to the 205.131.175.0
network. You will obviously add an additional permit statement if you wish
to allow applicable traffic to the other Ethernet network.
Permitting Echo-reply If you have no outbound restrictions, then you are
only filtering an inbound packets. Thus, echo requests transmitted from either
Ethernet result in an echo-reply. To allow responses to those pings you could
code the following statement into your access list:
access-list 101 permit icmp any 198.78.46.0 0.0.0.255 echo-reply
Permitting Pings Although ping can be used as a discovery weapon to probe
your network to discover information about your organization’s resources, it
can also be helpful for determining the status of certain devices. For example,
if your organization administers a remote Web server, you might wish to
permit pings to that server from your IP address. Assuming your IP address is
192.36.25.11, to permit pings to the server at 198.78.46.8 you would enter the
following statement:
access-list 101 permit icmp host 192.36.25.11 host 198.78.46.8 echo
Considering Destination Unreachable By default, when an access list sends
a packet to the great bit bucket in the sky the router will return a type 3 ICMP
security
471
message that has a code value of 13. This message, which is displayed as
‘‘Destination net unreachable,’’ informs the person issuing the packet that the
router on the path to that target address is performing access list filtering.
There is a special router statement you can use to prevent this information
from being determined by a potential hacker. After all, why make their job
easier? That statement is:
no ip unreachables
which you would enter for each router interface providing an Internet connection. The reason you cannot use an access-list statement is due to the fact
that router-generated packets are not checked by an access list applied in an
outbound direction. Thus, to turn off the previously mentioned message, you
need to enter the no ip unreachables statement.
Anti-Spoofing Statements
Regardless of the type of access list employed, one of the first series of
statements in your list should be anti-address spoofing statements. Because
hackers like to use RFC 1918 addresses, you should deny all packets with a
source address in three address blocks in RFC 1918. In addition, you should
block all packets with an IP address of all zeros, all ones, and the IP loopback
address of 127.0.0.0. If your organization does not support multicast (Class D)
nor any experimental access via Class E addresses, there is also no reason to
allow packets with Class D or class E source addresses. Finally, because some
hackers think it is funny to use a source address on the network they are
attacking, you should block all packets with a source address associated with
your network. If we assume that the internal network to be protected has the
network address 198.78.46.0, then the initial anti-spoofing statements at the
beginning of your access list would be as follows:
!Anti-spoofing statements
!
!Deny RFC 1918 addresses
access-list 101 deny 10.0.0.0 0.255.255.255 any
access-list 101 deny 172.16.0.0 0.31.255.255 any
access-list 101 deny 192.168.0.0 0.0.255.255 any
!
!Deny address all zeros, all ones, loopback
access-list 101 deny 0.0.0.0 0.255.255.255 any
access-list 101 deny host 255.255.255.255 any
access-list 101 deny 127.0.0.0 0.0.255.255 any
472
chapter nine
!
!Deny Class D and Class E addresses
access-list 101 deny 224.0.0.0 15.255.255.255 any
access-list 101 deny 240.0.0.0 7.255.255.255 any
!
!Deny source address of your network
access-list 101 deny 198.78.46.0 0.0.0.255 any
Now that we have an appreciation of the operational capabilities of standard
and extended IP access lists, let’s turn our attention to new features that were
added over the past few years that significantly enhance the capability of
router packet filtering technology.
New Capabilities in Access Lists
In tandem with several relatively recent updates to the Cisco Internetwork
Operating System (IOS) were improvements to the functionality and capability
of access lists. Six additions to access lists that warrant attention include
named access lists, dynamic access lists, reflexive access lists, time-based
access lists, TCP intercept and Context Based Access Control (CBAC). In
actuality, these additions represent new capabilities added to access lists and
do not literally represent new types of access lists.
Named Access Lists
Because standard and extended access lists have a limited range of acceptable
numbers, it is theoretically possible, although highly unlikely, that you could
run out of numbers when configuring an enterprise router. Perhaps a more
important reason for the use of named access lists is the fact that a name can
be more meaningful than a number. In addition, as we will shortly note, you
can delete statements in a named access list while a numbered list does not
provide this capability.
Overview
Named access lists were introduced in IOS Version 11.2. As its name implies,
a named access list is referred to by a name instead of a number.
Standard Named IP Access List
Named access lists are applicable to both standard and extended lists. The
format of a standard named IP access list is shown below.
ip access-list standard name
security
473
where ‘‘name’’ represents the name you would assign to the standard named IP
access list. The preceding statement is then followed by one or more permit or
deny statements that define the filtering to take place. The following example
illustrates the creation of a standard named access list to which we assigned
the name ‘‘inbound’’ to denote that it will be applied in the inbound direction.
In this access list we only permit traffic from two defined network addresses:
ip access-list standard inbound
permit 205.131.175.0 0.0.0.255
permit 198.78.46.0 0.0.0.255
To apply a named access list we use a modified version of the previously
described ip access-group command. That modified version includes the name
of a named access list and has the following format:
ip access-group name [in|out]
where ‘‘name’’ represents the name of the named access list. Thus, to apply the
previously created named access list to the serial0 interface to filter inbound
packets, our IOS statements would appear as follows:
interface serial0
ip access-group inbound in
!
ip access-list standard inbound
permit 205.131.175.0 0.0.0.255
permit 198.78.46.0 0.0.0.255
Now that we have an appreciation for the creation of standard named IP access
lists, let’s turn our attention to extended named IP access lists.
Extended Named IP Access Lists
An extended named IP access list is similar to a standard named IP access list.
The format of the extended named IP access list command is shown below:
ip access-list extended <name>
where ‘‘name’’ represents the name assigned to the access list.
474
chapter nine
You can use an extended named IP access list in the same manner as a
standard named IP access list. However, because extended access lists provide
considerably greater filtering capability, you can perform more functions with
this access list. To illustrate an example of the use of an extended named IP
access list, assume you only want to allow http access to the server shown
in Figure 9.2, whose IP address is 198.78.46.8. Let’s further assume you will
name the access list ‘‘security.’’ Because you want to filter packets flowing
from port 31, you would apply the access list in the outbound direction. Based
upon the preceding, the extended named IP access list statements would be
as follows:
interface ethernet1
ip access-group security out
!
ip access-list extended security
ip permit tcp any host 198.78.46.8 eq 80
In examining this extended named access list, you may be puzzled as to
the selection of the ethernet1 interface instead of the serial0 interface. The
reason we apply the access list to the ethernet1 interface instead of the serial0
interface is that selecting the latter would block all Internet traffic flowing
into the router other than Web traffic flowing to the specified server. Also note
that we specify the direction of the list as being applied to outbound (out)
traffic. This is because packets leave the ethernet1 interface to flow to the
specified network.
Editing Capability
We previously mentioned that one advantage of a named access list is the
fact you can remove previously entered list statements. To do so you would
reenter the configuration mode and enter a ‘‘no’’ prefix for the statement you
previously entered. It should be noted that you cannot delete specific entries
in a numbered access list. One other point that deserves mentioning is the fact
that you cannot selectively add statements to any type of access list other than
to the bottom of the list. To add statements to a numbered or a named access
list you must delete an existing list and reapply a new or modified list with
appropriate entries. Now that we have an appreciation for the operation and
utilization of named access lists, let’s continue our exploration of additional
list features and turn our attention to dynamic access lists.
security
475
Dynamic Access Lists
As the name of this access list implies, dynamic access lists create dynamic
entries in a list. Those dynamic entries represent temporary openings in an
access list that occur in response to a user authentication process.
Rationale for Use
The primary use of a dynamic access list is to obtain the ability to authenticate
users attempting to access your network. To accomplish this you would first set
up user accounts on your router. Next, you would create and apply a dynamic
access list to the serial port of your router in the inbound direction, assuming
you wish to authenticate users transmitting in that direction. Users then open
a Telnet session to your router and authenticate themselves, normally by
providing a user ID and password. Once the user is authenticated, the router
will close the Telnet session and place a dynamic entry in an access list,
which permits packets with a source IP address of the authenticated user’s
workstation.
One of the key advantages associated with the use of a dynamic access list
is that it can be used for certain situations where it is not possible to use a
static access list. For example, many Internet Service Providers (ISPs) assign
dynamic IP addresses to subscribers. This means that there is no way to create
a static access list entry to allow authorized users who have dynamically
assigned addresses to access your network via an ISP connection other than
to allow all ISP network addresses. Because this would result in a potentially
large security gap and is not recommended, you would more than likely turn
to the use of a dynamic access list for this situation.
Utilization
The dynamic access list is very similar to an extended IP access list, with
a key difference being the inclusion of the keyword ‘‘dynamic’’ in an
extended access-list statement. The format of a dynamic access-list entry
is shown below:
access-list <list number> dynamic <name> [timeout n] [permit|deny]
<protocol> any <destination address> <destination mask>
The first variable field, ‘‘list number,’’ follows the same format as a traditional extended access list and represents a number between 100 and 199. The
second variable field, ‘‘name,’’ represents the designated name of the dynamic
access list. The optional timeout variable can be used to specify an absolute
476
chapter nine
timeout for a particular dynamic entry. The ‘‘protocol’’ parameters represent
any one of the TCP/IP protocols, such as IP, TCP, UDP, and ICMP. Because the
source IP address is always replaced by the IP address of the authenticating
host, the keyword ‘‘any’’ should be used for the source IP address field. The
last two variable fields, ‘‘destination address’’ and ‘‘destination mask’’ are
used in the same manner as in an extended IP access list.
Prior to examining an example of the use of a dynamic access list, a few
additional details warrant discussion. First, you cannot specify more than
one dynamic access-list statement per access list. This means you need to
consider carefully which protocols you want to create dynamic openings
for in an access list. Second, you need to permit users to Telnet to your
router or they will not be able to authenticate themselves and proceed to
create dynamic openings in the access list. Third, to allow dynamic entries
to be created, you must use the ‘‘autocommand’’ parameter under the vty
line configuration. An example of the use of this command parameter is
shown below:
line vty 0 3
login local
autocommand access-enable host timeout 5
In this example the ‘‘host’’ parameter enables the source IP address of
the authenticating host to be substituted into the dynamic entries that
will be created. The ‘‘timeout’’ parameter is optional and when included
specifies an idle timeout. If you use both absolute and idle timers, the
idle timer value should be set to a value less than the absolute timer.
As a minimum, at least one timer value should be configured. This is
because without a timer the dynamic entries will remain until the router
is reinitialized.
One additional point deserves mention prior to illustrating the utilization
of a dynamic access list. This point is of key importance because its omission
can become a network administrator’s nightmare if he or she is remotely
administrating a router. Because a Telnet session is immediately closed after
authentication, this action will prevent you from managing a router via Telnet.
If you configure your virtual terminal lines as previously described you would
be literally up the creek without a paddle. The way around this problem
is to specify the ‘‘rotary 1’’ command beneath one or more vty ports as this
command enables normal Telnet access to a router on port 3001. For example,
to enable normal Telnet access on port 3001 for vty 4, you would enter the
following commands:
security
477
line vty 4
login local
rotary 1
Once the preceding occurs, you would set your Telnet application to use
port 3001. For example, if your router’s IP address is 205.131.176.1, you
would enter the following command:
telnet 205.131.176.1 3001
Now that we have an appreciation of the details concerning a dynamic access
list, let’s focus our attention upon an example. In doing so, let’s assume your
router is connected to the Internet as shown in Figure 9.4. The serial port
of your router has the IP address 205.131.175.1 and you only want to allow
Internet
Router
s0
205 . 131 . 175 . 1
e0
198. 78. 46. 0
Figure 9.4 Configuration to reference for creating a dynamic access list that
permits authenticated users access to the 198.78.46.0 network.
478
chapter nine
persons with predefined user IDs on the router to access your private network.
In addition, we will assume you wish to permit remote administration of the
router via Telnet. To accomplish the preceding you would enter the following
IOS commands. Note that only the relevant portions of the configuration
commands are listed:
username gxheld password gizmo87
!
interface serial0
ip address 205.131.175.1 255.255.255.0
ip access-group 101 in
!
access-list 101 permit tcp any host 205.131.175.1 eq 23
access-list 101 dynamic gxheld 10 permit ip any 198.78.46.0 0.0.0.255
!
line vty 0 3
login local
autocommand access-enable host timeout 5
line vty 4
login local
rotary 1
In this example note that we create the user name gxheld and reference it
in the second access-list statement. This makes it harder for an attack since
a person must specify both a user name and password to gain access. Also
note that the first access-list statement restricts access to the router to Telnet
(port 23). The second access-list statement contains the keyword ‘‘dynamic.’’
This statement creates dynamic openings in the access list, substituting
the source address of the authenticated Telnet user for all IP traffic with
a destination address on the 198.78.46.0 network. Now that we have an
understanding of the operation and utilization of dynamic access lists, let’s
turn our attention to reflexive access lists.
Reflexive Access Lists
When we previously discussed the use of the keyword ‘‘established’’ in an
extended IP access list, we noted that it is only applicable to TCP. If you wish
to control other upper-layer protocols, such as UDP and ICMP, you would have
to either permit all incoming traffic or define a large number of permissible
source/destination host/port addresses. In addition to representing a timeconsuming and tedious task, the resulting access list could conceivably require
security
479
more memory than available on your router. Perhaps in recognition of this
problem, Cisco introduced reflexive access lists in IOS Version 11.3.
Overview
A reflexive access list creates a dynamic, temporary opening in an access list,
based upon a mirror image of an IP traffic session originated from inside your
network to an external network. The temporary opening is always a permit
entry and specifies the same protocol as the original outbound packet. This
opening also swaps source and destination IP addresses and upper-layer port
numbers and remains in existence until either the session initiated on the
trusted network is closed or an idle timeout value is reached.
Rationale for Use
The rationale behind the use of a reflexive access list is to enable employees on
the trusted internal side of the router to control openings in the access list that
occur from the untrusted side of the network. An example of the operation of
a reflexive access list is shown in Figure 9.5. In examining Figure 9.5 note that
the inbound opening reverses source and destination IP addresses and port
numbers. Also note that the initial Telnet session uses a destination port of 23
and a random source port number greater than 1023. Thus, the opening results
in a source port value of 23 and a destination port value greater than 1023.
Creation
There are four general tasks associated with the creation of a reflexive access
list. The first task is to create an extended named access list. In an IP
environment you would use the following command format:
ip access-list extended name
where ‘‘name’’ represents the name of the access list.
The second task is to create one or more permit entries to establish reflected
openings. Because you normally apply a reflexive access list to outbound
traffic, it will result in an inbound access list. When defining permit statements
for your outbound access list, you use a slightly modified format of the permit
statement. This modified format is shown below:
permit protocol any any reflect name [timeout seconds]
Here the protocol represents the protocol for which you want reflexive openings created. Because you normally want any user behind the router to create
480
chapter nine
Internet
Router
s0
e0
Opening created from
205 . 131 . 122 . 6 to
198 . 78 . 46 . 8 from port
23 to port 1023
Telnet to 205 . 131 . 122 . 6
User
198. 78. 46. 8
Figure 9.5
Examining the operation of a reflexive access list.
dynamic openings, the keyword ‘‘any’’ is used for the source address. Similarly, because reflexive results can represent any swapped addresses and port
numbers, you would also use the keyword ‘‘any’’ for the destination address.
The keyword ‘‘reflect’’ identifies the statement as a reflexive statement, while
‘‘name’’ represents the name of the access list. The optional keyword ‘‘timeout’’ is used to assign a timeout period to each reflexive entry created in
the inbound direction. If this option is omitted, a default timeout value of
300 seconds is used.
You can also place a global timeout value that will be applicable to all
reflexive statements. To do so you would use the ip reflexive-list timeout
global command, whose format is shown below:
ip reflexive-list timeout value
where ‘‘value’’ represents the global timeout value in seconds.
security
481
The third task associated with the creation of a reflexive access list is to
create an access list for inbound filtering. This is required as dynamic reflexive
entries will be added to this access list.
The last task in the reflexive list creation process is to conclude your list
with the evaluate command whose format is shown below:
evaluate name
where the variable ‘‘name’’ represents the name of the access list and determines which packets are to be evaluated by reflexive entries.
To illustrate the use of a reflexive access list, let’s assume we want to limit
reflected openings to 240 seconds of idle time. Let’s also assume your inbound
access list will initially be coded to perform anti-spoofing operations by
sending RFC 1918 addresses as well as ‘‘all 0’s,’’ loopback and ‘‘all 1’s’’ source
addresses to the bit bucket. The following example illustrates the creation of
the applicable statements for the previously described reflexive access list. In
examining the statements, note that the six deny statements in the extended
access list named ‘‘inbound’’ are conventional statements that are not reflected
but indicate where reflexive statements will be dynamically added.
ip reflexive-list timeout 240
!
ip access-list extended outbound
permit tcp any any reflect my-session
permit udp any any reflect my-session
permit icmp any any reflect my-session
!
ip access list extended inbound
deny ip 127.0.0.0 0.255.255.255 any
deny ip host 255.255.255.255 any
deny ip 0.0.0.0 255.255.255.255 any
deny ip host 255.255.255.255 any
deny ip 10.0.0.0 0.255.255.255 any
deny ip 172.16.0.0 0.31.255.255 any
deny ip 192.168.0.0 0.0.255.255 any
evaluate my− session
Limitations
Although the use of reflexive access lists considerably enhances the ability to secure the flow of packets from an untrusted side of a router, their
key limitation is the fact that they are limited to supporting single-channel
482
chapter nine
connections. This means that applications such as file transfer protocol (ftp)
that use multiple port numbers or channels cannot be supported by reflexive
access lists. To obtain support for multi-channel applications, Cisco initially
provided a special release of IOS referred to as the Firewall Feature Set (FFS),
which was limited with respect to the platforms it operated upon. During 1999
FFS was incorporated into IOS Release 12.0 under the name Context Based
Access Control (CBAC). CBAC not only supports multi-channel applications
but, in addition, adds a Java blocking capability as well as denial-of-service
prevention and detection, real-time alerts, and audit trails. Because CBAC
represents the most sophisticated variation of access lists, we will defer an
examination of this access list to the end of this section. Thus, we will continue our examination of access lists by turning our attention to time-based
access lists.
Time-Based Access Lists
Until the release of IOS Version 12.0 there was no easy method to establish
different security policies based upon the time of day or date. Although you
could create multiple access lists and apply them at different times of the
day, doing so could entail a work effort that might be less than desirable.
For example, to implement a new security policy to enable Web surfing after
6:00 p.m., you would either have to stay in the office until that time or Telnet
from home and revise your access list. If your organization decides that the
policy should also revert back to blocking Web surfing at 5:00 a.m., it might
not be very appealing to get up before your pet to go to work. With the
introduction of IOS Version 12.0 you can now use time-based access lists to
obtain the ability to implement different security policies based upon the time
of day.
Creation
The creation of a time-based access list is a relatively straightforward twostep process. First, you define a time range. Once this is accomplished you
reference the time range in an access-list entry.
The specification of a time range is accomplished through the use of a
time-range statement whose format is shown below:
time-range time-range-name
where the ‘‘time-range-name’’ represents the name you assign to the time
range. Once this task is accomplished you can specify a time range in one of
security
483
two ways. You can use an ‘‘absolute’’ statement or you can use a ‘‘periodic’’
statement. The format of each statement is shown below:
absolute [start time date] [end time date]
periodic days-of-the-week hh:mm to [days-of-the-week] hh:mm
The ‘‘time’’ parameter in an absolute statement is entered in the format
hh:mm, where hours (hh) are expressed in a 24-hour format. For the periodic
statement you can list the days of the week separated by spaces or use the
keywords ‘‘daily’’ or ‘‘weekend.’’ Once you create a time range, you can
reference it through the optional keyword ‘‘time-range’’ in a conventional
access-list entry.
Example
Since the old adage ‘‘the proof of the pudding is in the eating’’ holds true
today, let’s focus our attention upon an example. The following example
illustrates the creation of a time-based access list that restricts Web access to
Saturday and Sunday from 9:00 a.m. until 5:00 p.m.:
time range allow-http
!
periodic weekend 09:00 to 17:00
!
access-list 101 permit tcp any any eq 80 time-range allow-http
TCP Intercept
TCP intercept represents a feature that was added in IOS Version 11.3 as a
mechanism to minimize the effect of a special type of denial-of-service attack
referred to as SYN flooding and which is also known as a Smurf attack, after
the cartoon character. With the release of IOS Version 12.0, TCP intercept was
incorporated into Context Based Access Control. Thus, we can literally cover
two topics at the same time by deferring a discussion of TCP intercept and
covering it as we examine context based access control.
Context Based Access Control
Context Based Access Control (CBAC) represents the latest addition to Cisco
router packet filtering capabilities. Introduced during 1999, CBAC provides a
484
chapter nine
router administrator with the ability to create dynamic entries in a router’s
access list for both single- and multi-channel applications. In addition, CBAC
provides the ability to control the flow of Java applets, as well as the ability to
minimize the effect of certain types of denial-of-service attacks.
Overview
Until 1999 CBAC was a special feature referred to as the firewall feature set
(FFS), which was only available for use on the Cisco series 1600 and 2500
router platforms. During 1999 the release of Cisco’s Internetwork Operating
System (IOS) Release 12.0T expanded the availability of FFS, which is now
referred to as CBAC, to Cisco 800, uBR900, 1600, 1700, 2500, 2600, 3600,
7100, and 7200 series platforms.
CBAC operates similarly to reflexive access lists, which were previously
covered in this chapter. As we noted earlier, a reflexive access list is used to
create dynamic openings in an inbound access list in response to an outbound
data transmission. A key limit of a reflexive access list is its inability to support
multi-channel applications, such as FTP, CU-SeeMe, H. 323, and similar
transmissions that use two or more TCP or UDP ports. CBAC overcomes this
limitation, providing you with the ability to extend the capability of reflexive
access lists to multi-channel applications.
Table 9.3 provides a list of representative applications that CBAC can
securely support. Concerning that support, CBAC functions by examining
packets that enter or exit a specified interface. Key information contained in
the packet, such as IP addresses and layer 4 port numbers, is placed in a state
table. The contents of that table are then used by CBAC to create temporary
openings in an access list for returned traffic. Although the operation is similar
to that of a reflexive access list, CBAC does this for both single and multichannel applications. In addition, CBAC tracks the sequence numbers used in
a TCP conversation to ensure that they are within the expected range, which
can prevent a sophisticated attack by someone monitoring a conversation and
attempting to piggyback onto it to break into a network resource.
In addition to extending the ability of reflexive access lists, CBAC adds
several new features that result in much more than an access-list enhancement.
These additional features include Java blocking, denial-of-service prevention
and detection, and the ability to provide real-time alerts and audit trails.
Thus, CBAC can be considered to represent a comprehensive set of security
tools, even though it is enabled in a manner similar to other types of Cisco
access lists.
security
485
TABLE 9.3 Examples of Applications
Supported by CBAC
Single-channel TCP (i.e., Telnet)
Single-channel UDP (i.e., DNS)
CU-SeeMe (White Pine Software version)
FTP
H. 323 (NetMeeting, ProShare)
IP fragments
Java (applets embedded in HTTP)
Unix r commands (rlogin, rexec, etc.)
RealAudio
RPC (Sun version)
SMTP
SQL*Net
TFTP
Operation
As previously noted, CBAC extends the capability of reflexive access lists
to multi-channel applications in addition to adding several new features.
Because CBAC monitors outbound traffic to create applicable inbound accesslist entries, two access lists are required. One CBAC access list will define the
packets that will be inspected by CBAC. The second access list will include
the entries that CBAC dynamically creates as it compares outbound packets
against the statements you will code into an access list that will be applied to
a particular interface in the outbound direction.
The actual CBAC configuration process is similar to the manner by which
a reflexive access list is created. That is, you first select an interface and then
configure the applicable access lists on the interface. In doing so you use one
or more ‘‘ip inspect’’ statements in the access list you are configuring, which
informs IOS that it should perform CBAC operations.
CBAC Example
To illustrate the creation of an access list with CBAC capability, let’s assume
your organization has a branch office with a connection to the Internet as
486
chapter nine
Internet
S0
Router
E0
Ethernet
LAN
Legend:
E1
Ethernet
LAN
Workstations
Figure 9.6 A sample network to be used for illustrating the configuration
and operation of Context-Based Access Control (CBAC).
illustrated in Figure 9.6. In this example we will assume the branch office
operates two Ethernet LANs, connected to ports #0 and E1 on the router, while
port S0 provides a serial interface to the Internet.
Interface Because the first step in creating any type of access list is to select
an appropriate interface, let’s do so. Assuming we are attempting to protect
internal users on the two Ethernet LANs from the virtually unlimited Internet
user population, we configure CBAC on the external interface, S0. This will
result in any traffic attempting to reach either internal Ethernet networks
being inspected by CBAC. Now that we have selected an interface, a few
words about the type and direction of the access lists to be applied to the
interface are in order.
Since we want to protect internal users from traffic flowing inbound from
the Internet, we create an outbound access list that specifies the traffic to be
inspected by CBAC. This access list can be either a standard or extended IP
access list. As CBAC examines outbound packets it will dynamically create
openings in an inbound access list that governs the flow of traffic from the
Internet that can reach either Ethernet network located behind the router. This
access list must be an extended access list.
security
487
One aspect of CBAC that requires elaboration is the fact that it is important
to remember that its use creates dynamic openings in an inbound access
list based upon the IP address and layer 4 ports used in packets flowing
in the outbound direction that match your defined criteria. This means
that unless you supplement your CBAC-related statements with permissions
for different types of data originated from the Internet, your organization’s
Internet connection will block all inbound traffic originated on the distrusted
side of the router. While this may represent your intention, suppose one of
the Ethernet networks shown in Figure 9.6 is connected to a Web server you
wish to allow the general Internet population to access. To do so you must
include an applicable permission in the inbound access list to enable traffic
to flow through the router to the Web server. Another aspect of CBAC that
warrants attention is the fact that it does not protect your organization from
certain types of HTTP application-layer attacks, such as the exploitation of
cgi scripts. Thus, it is important to note that CBAC is not a total barrier to
security problems. Instead, it represents a sophisticated tool that adds to the
capability of your organization’s total security policy. That said, let’s turn our
attention to the statements that add CBAC capability to your access list.
The Inspect Statement
The ip inspect statement represents the key IOS command associated with
configuring a monitored connection for inspection by CBAC. The format of
this command for all protocols other than Java and RPC (remote procedure
call) is shown below:
ip inspect name inspection-name protocol [alert {on|off }] [audit-trail]
[on|off ] [timeout seconds]
In this format, the ip inspect command’s optional keyword ‘‘alert’’ causes
CBAC to generate messages to a syslog server or the router’s buffer whenever it
detects a violation in the monitored application. For example, if you configure
CBAC to monitor TCP and an illegal TCP operation is detected when audit is
enabled, the router will send an alert to the syslog server.
The keyword ‘‘audit-trail’’ enables CBAC to track the connections for a protected application. When this keyword is used, the router will log information
for each permitted application to include source and destination IP addresses,
ports used, and the number of bytes transferred. Although the audit trail was
probably developed to enable network managers to obtain information on
the traffic characteristics of network applications, it also enables managers
to determine the locations employees are accessing. Thus, it also provides a
488
chapter nine
database that could be used to determine if employees are viewing the stock
market, accessing Web auction locations, or performing other activities that
management may not particularly support.
The keyword ‘‘timeout’’ provides you with the ability to specify the duration
of inactive sessions. In addition to being able to specify a general inactive
timeout for specific protocols, CBAC also supports the capability to prevent
denial-of-service (DOS) attacks by monitoring the number and frequency of
half-open connections. For TCP a half-open connection represents a session that has not completed the initial three-way (syn–syn–ack) handshake.
Because UDP has no handshake, a UDP half-open session is one for which
CBAC has not detected return traffic.
You can control the number of TCP and UDP half-open connections through
the use of ten ip inspect related statements that assign a timeout or threshold value. Table 9.4 lists those commands, their default values, and a brief
description of each statement.
To obtain an appreciation of the entries in Table 9.4, a brief review of
the operation of denial-of-service (DOS) is warranted. Under a DOS attack
a hacker floods a network with either TCP or UDP requests and does not
answer responses to those requests. This action rapidly consumes all available
resources on the target host, which is then forced to deny service to legitimate
users. While using a router to intercept and control DOS attacks ensures that
computer resources on the network are available for legitimate use, the router
could easily have its resources consumed. To prevent this situation from
occurring there are timeout and threshold defaults associated with CBAC as
indicated in Table 9.4. You can either elect to accept those defaults and do
nothing, or modify one or more of those defaults through the use of the ip
inspect commands listed in Table 9.4.
To illustrate the use of the basic ip inspect command, let’s assume you want
to create dynamic openings in the inbound direction from the Internet for
TCP and UDP applications originating on the trusted side of the router. Let’s
further assume you want to assign a 60-second timeout for TCP connections
and a 30-second timeout for UDP connections. To do so you would use the
following ip inspect commands:
ip inspect name firewall tcp 60
ip inspect name firewall udp 30
To complete our discussion of ip inspect command formats, let’s turn our
attention to the two previously mentioned variations in the command format
security
489
TABLE 9.4 IP Inspect Commands Used to Configure Time-out and Threshold
Values Used by CBAC
Command
Default
Value
Description
ip inspect tcp synwait-time
seconds TCP
30 s
Length of time to wait for session to
be established.
ip inspect tcp finwait-time
seconds
5 s
Length of time TCP is managed
after FIN exchange.
ip inspect tcp idle-time
seconds
3600 s
TCP idle time-out
ip inspect udp idle-time
seconds
30 s
UDP idle time-out
ip inspect dns-timeout
seconds
5 s
DNS lookup idle timer
ip inspect max-incomplete
high number
500 sessions
Maximum number of half-open
connections permitted before
CBAC begins closing connections.
ip inspect max-incomplete
low number
400 sessions
Number of half-open connections
causing CBAC to stop closing
connections.
ip inspect one-minute high
number
500 sessions
Rate of half-open sessions per
minute before CBAC begins closing
connections.
ip inspect one-minute low
number
400 sessions
Rate of half-open sessions per
minute causing CBAC to stop
deleting connections.
ip inspect tcp
max-incomplete host
Number block-time
seconds
50 sessions
Number of existing half-open
sessions with the same destination
address before CBAC begins
closing sessions.
for RPC and Java. For RPC the format of the command is slightly different as
illustrated below:
ip inspect name inspection-name rpc program-number number [wait-time
minutes] [alert {on|off ]] [audit-trail {on|off }] [timeout seconds]
490
chapter nine
As an example of a CBAC inspect statement for RPC, assume you want to
allow RPC program number 14000 and effect a 60-second idle timeout. To do
so you would use the following inspect command:
ip inspect name firewall rpc program-number 14000 timeout 60
To use CBAC for Java blocking you need to associate a list of permitted IP
addresses via the use of a standard IP access list to a slightly modified inspect
command. This modified inspect command format is shown below:
ip inspect name inspection-name http[java-list access-list#] [alert{on|off ]]
[audit-trail {on|off }] [timeout seconds]
In this format the access-list# represents the standard IP access-list number
you wish to associate with Java blocking. If you should reference an undefined
access list in the java-list definition, the default behavior will result in all
Java applets being denied since there are no permitted IP addresses associated
with the ip inspect command.
As an example of the use of CBAC to block Java applets, let’s assume the
network address of a parent Ethernet network located on the other side of the
Internet in Figure 9.6 is 198.78.46.0. Then, to block all Java applets other than
those from the parent network, you would code the following commands:
access-list 1 permit 198.78.46.0 0.0.0.255
ip inspect name firewall http java-list 1
Although these commands block all Java applets other than those originating
from the 198.78.46.0 network, it should be noted that in its current incarnation
CBAC does not block ActiveX. At the present time Cisco recommends the use
of a dedicated content-filtering product if you need to perform an extensive
amount of content filtering or if you need to filter one or more applications
not presently supported by Cisco access list features to include CBAC.
Applying the Inspection Rules
Similar to other types of access lists, once you create your CBAC inspection
rules you need to apply those rules to an interface. To do so you would use
an ip inspect command that indicates the direction of inspection. The format
used to apply an inspection rule is shown below:
ip inspect inspection-name {in|out}
security
491
Note that when applying the inspection you should apply them to the
direction of outbound traffic to ensure that CBAC checks inbound traffic.
This means that if you are configuring inspection on an internal interface,
the outbound traffic from the network is entering the interface. Thus, in
this situation the inspection rules should be applied inbound. If you are
configuring inspection on an external router interface, the outbound traffic is
leaving the interface. In this situation the inspection rules should be applied
in the outbound direction. While the preceding may appear a bit confusing,
you can avoid potential confusion by remembering that you would apply the
inspection rule to packets leaving the interface to be guarded to insure that
return packets are checked.
Using CBAC
Since the proof of the pudding is in the eating, let’s illustrate the manner
in which we can use CBAC. For our example, let’s assume our organization
has a two-port router with the serial port providing a connection to an
Internet Service Provider, while the Ethernet interface provides a connection
to the internal Ethernet network. Let’s further assume you want to allow
the internal users on your Ethernet LAN to have access to the Internet for
Web browsing, FTP and electronic mail. Let’s also further assume that your
parent organization, whose network address is 205.131.175.0, should be the
only network from which Java applets will be allowed to flow through your
router onto your network. In addition, we will assume that you would like to
use the alerting capability of CBAC to provide information concerning when
Java applets are blocked or permitted and the auditing capability of CBAC to
provide you with information concerning FTP traffic. Last but not least, you
want to enable your internal Ethernet users to perform ping and traceroute
operations to hosts on the Internet.
The specific coding we would use to implement the previously stated
network requirements are listed below:
interface ethernet0
ip address 198.78.46.1 255.255.255.0
ip access-group 101 in
!
!
interface serial0
ip address 198.78.40.5 255.255.255.0
ip inspect firewall out
ip access-group 102 in
492
chapter nine
!
!
ip inspect alert-off
ip inspect name firewall http java-list 1 alert on
ip inspect name firewall ftp audit-trail on
ip inspect name firewall smtp
ip inspect name firewall tcp
ip inspect name firewall udp
!
ip access-list 1 permit 205.131.175.0 0.0.0.255
!
ip access-list 101 permit ip any any
!
ip access-list 102 permit icmp any any echo-reply
ip access-list 102 permit icmp any any time-exceeded
ip access-list 102 permit icmp any any packet-too-big
ip access-list 102 permit icmp any any unreachable
ip access-list 102 permit icmp any any administratively-prohibited
In examining the coding, note that the IOS commands are grouped into
several areas. First, we configure the Ethernet interface by assigning an IP
address to the interface and then associating access list number 101 in the
inbound direction with the interface through the use of an ip access-group
statement. This is followed by the configuration of the serial interface.
Because we will apply CBAC inspection to the serial interface we include
an ip inspect command under the serial interface. Note that the ip inspect
command is set up to inspect packets flowing in the outbound direction as we
want CBAC to create temporary openings in the inbound direction. Because
we use the ip access-group command under the serial interface, CBAC will
create temporary openings at the bottom of access list 102 whose statements
we will shortly review.
After the two interfaces have been configured, you will note a block of six
ip inspect commands. The first ip inspect command disables global alerting,
permitting us to selectively enable this feature for http in the second ip
inspect command. Note that the second ip inspect command specifies that
CBAC should use access list number 1 to selectively permit Java applets. Also
note that after the six ip inspect commands we define a standard access list
number 1 that only allows traffic from the 205.131.175.0 network. Because the
second ip inspect statement is linked to access list number 1, this means that
only Java applets from the 205.131.175.0 network will be permitted inbound
security
493
through our router. Also note that, because we previously mentioned that we
wanted to be alerted concerning the permission or blocking of Java applets,
the second ip inspect command includes an ‘‘alert on’’ option.
The third ip inspect command configures inspection for ftp, while the fourth
command provides inspection for sending email. Because we mentioned a
requirement to audit ftp, the third ip insert command contains an ‘‘audittrack on’’ option. The fifth and sixth ip inspect commands provide generic
TCP and UDP inspection to allow return traffic for queries using different
protocols from your internal network. For example, by specifying generic
TCP permissions, your employees could use call control protocols required
for voice over IP (VoIP) session setup. TCP also allows the use of the post
office protocol (POP), which enables your internal users to retrieve their email
from an ISP’s mail server. In comparison, we explicitly specify smtp in an ip
inspect command since it is used for sending email and, if we did not specify
a generic tcp, our employees would be limited to sending, but not receiving,
email. This obviously would not be particularly useful. The use of a generic
UDP permits the actual transfer of VoIP packets that applications transfer
using the connectionless operation of UDP. In addition, the generic UDP
command permits employees to use DNS and SNMP as both are transported
as UDP datagrams.
After the block of six ip inspect commands we include two access-list
statements. The first is a standard access-list statement which, as previously
noted, is linked with our Java applet blocking. The second access list, list 101,
permits all IP traffic to flow in the inbound direction to the router from the
Ethernet network. This access list is not required, but has been included as a
reference point in the event that we later expand the router and wish to block
certain types of traffic from the Ethernet network into the router. Because the
access list permits all IP traffic from any source address to any destination
address, its current effect is the same as if we did not associate an access list
with the Ethernet port in the inbound direction.
The last four statements in our router configuration listing consist of four
access-list statements. These statements make up an IP extended access
list numbered 102, which is applied in the inbound direction on the serial
interface. Because CBAC only inspects TCP and UDP traffic, we must explicitly
permit other types of IP traffic through the serial port in the inbound direction.
Since part of our previously defined requirement was to permit employees to
perform ping and traceroute operations, we need to explicitly enable the IP
traffic required to allow these operations. Thus, we have coded several ICMP
permissions. The permission of echo-reply allows internal users to ping hosts
on the Internet and receive a reply. The permission of the time-exceeded
494
chapter nine
and unreachable ICMP statements allows traceroute to function. Although
not part of our requirement, we have added permissions for packet-too-big
and administratively prohibited ICMP messages. These messages enable MTU
(maximum transmission unit) discovery and messages indicating possible
access lists on Internet sites our users may attempt to reach. If we had not
included these ICMP permissions, it is possible that our internal users might
experience significant delays when attempting to access some locations and
might be prevented from reaching other locations. Because CBAC will ensure
that TCP and UDP entries are automatically created for return traffic, once
traffic from the internal Ethernet network flows toward the Internet we do not
have to explicitly enable other protocols. Thus, in addition to creating dynamic
openings, CBAC permits us to simplify our access-list creation process.
9.2 The Role of the Firewall
In the first section in this chapter we noted that the router represents the
initial line of defense of an organization’s private network when that network
is connected to the Internet. The key method of defense is the creation of
applicable access lists that perform packet filtering. While there are several
types of access lists and access-list features you can employ by themselves,
they are not sufficient to prevent many types of undesirable operations
against hosts residing behind a router. One solution to this security gap
is to use another communications device that provides additional capability
beyond packet filtering. That device is the firewall, which is the focus of
this section.
Access-List Limitations
Although Cisco routers support many new access-list features, including
dynamic lists that provide an authentication capability and reflexive and
context-based access control that create dynamic openings in an inbound
access list based upon outbound activity from the trusted side of a network, they all have one serious limitation. That limitation is the fact that
all access lists are relatively blind with respect to the operation being performed. This results from the inability of router access lists to look further
into the contents of a packet as a mechanism to determine whether or not
an apparently harmful operation is occurring, and if so, to either stop the
operation or generate an appropriate alert message to one or more persons in
the form of an audio signal, email message, or pager alert, or a combination of
such mechanics.
security
495
Repeated Logon Attempts
Assume you permit employees to use FTP, email, and Web surfing and to
perform other Internet activities. Thus, it is possible for a hacker to attempt to
gain access to your hosts, have some fun, and lock out legitimate users from
accessing one or more servers. The hacker could also perform a combination of
activities harmful to the health of your employees’ ability to use organizational
computational equipment. For example, by transmitting repeated logins, a
person may be able to either break into a host or lock out a legitimate user.
For either situation the hacker repeats a sequence of login actions. Because
an access list does not check the contents of packets, it is possible that the
repeated actions continue until a break in occurs or the lock-out value for a
particular valid user ID on a server is reached. One solution to this problem
is to examine the contents of each packet and note repeating patterns, which
are then blocked. Because this action requires a considerable amount of
processing power, it is normally performed by a firewall. In performing this
function the firewall maintains a state table of operations between a particular
source and destination address, with the examination of packets occurring in
an attempt to both determine and prohibit certain activities, with the process
referred to as stateful inspection.
Application Harm
A second example of an activity that illustrates a limitation of access lists can
be obtained by discussing certain operations associated with the file transfer
protocol (FTP). Although we will focus our attention upon the manner by
which the FTP application can be hazardous to the health of a remote
computer, it should be noted that other TCP and UDP applications can at
times also be harmful when used in certain manners.
When using a router’s access list, you can enable or deny ftp sessions based
upon the source IP address and/or the destination IP address contained in
each packet transporting FTP information. Suppose that your organization
operates an FTP server supporting anonymous access, allowing any person
connected to the Internet to access and retrieve information from the FTP
server, a relatively common occurrence on the Internet. Let’s further assume
that your organization has a large number of files on the server available
for downloading. This means that a person could either intentionally or
unintentionally use the FTP mget (multiple get) command to retrieve a large
number of files with one ftp command line entry. In fact, if the person
accessing your organization’s ftp server issued the mget command using the
wildcard operator of an asterisk (∗ ) in the filename and file extension position
496
chapter nine
to form the command line entry mget ∗ .∗ then this command would result
in your organization’s FTP server downloading every file in the directory,
one after another, to the remote user. If your organization has a large number
of files whose aggregate data storage represents several gigabytes of data
and a low-speed connection to the Internet, such as a 56 kbps, 64 kbps or
fractional T1 connection, the use of an mget ∗ .∗ command could tie up
the outbound use of the Internet connection for many hours and possibly
days. If your organization operates a World Wide Web (WWW) server as
well as an FTP server and provides Internet access to employees over a
common access line, the use of mget on an intentional basis can be considered
to represent an unsophisticated but effective denial-of-service (DOS) attack
method. This type of attack is perfectly legal as the person employing the
mget command is performing a perfectly valid operation, even though the
result of the operation could tie up your organization’s connection to the
Internet for hours or even days. Similarly, letting a person have the ability to
download data to your organization’s FTP server means they could consider
using the reverse of mget, which is the mput command. Through the use of
mput with wildcards they could set up an antiquated 286 processor-based
machine and pump gigabytes of data to your FTP server, clogging the inbound
portion of your organization’s Internet access line. Recognizing the need
to examine application-layer operations and provide organizations with the
ability to control applications resulted in the development of a proxy services
capability, which is included in many firewalls.
Proxy Services
Proxy services represents a generic term associated with the use of a proxy
server. The proxy server is normally implemented as a software coding module
on a firewall and supports one or more applications, for which the server acts
as an intermediary or proxy between the requestor and the actual server that
provides the requested service. When implemented in this manner all requests
for a specific application are first examined by the proxy service operating on
the proxy server. If the proxy service has previously been configured to enable
or disable one or more application features for a specific TCP/IP application,
then the proxy service examines the contents of each packet and possibly
a sequence of packets and compares the contents against the proxy service
configuration. If the contents of the packet or sequence of packets that denote a
specific operation are permitted by the configuration of the proxy service, then
the service permits the packet to flow to the appropriate server. Otherwise
the packet is either directly sent to the great bit bucket in the sky or possibly
security
497
permitted with the server generating a warning message and an alert or alarm
message to the firewall administrator or other designated personnel.
Operation
To illustrate the use of a proxy service let’s return to our ftp server access
example. A common FTP proxy service permits a firewall administrator to
enable or disable different FTP commands. Using this feature, the firewall
administrator can control the ability of FTP users to issue different types of
FTP commands, such as mget and mput.
In a Microsoft Windows environment you can use mget in either a streaming
or an interactive mode. Concerning the latter, FTP will prompt you through the
use of a question mark (?) whether or not the next file should be transferred.
An example of the use of mget is illustrated in Figure 9.7. Note that by
simply entering a carriage return in response to the ? prompt the next file is
transferred. Thus, it is relatively easy for a hacker to write a script to stream
files when using mget under Windows’ interactive mode and a no-brainer
under its streaming mode.
If you are familiar with the manner in which an FTP server is configured,
you probably realize that the FTP server administrator is limited to assigning
read and/or write permissions to directories and possibly, depending upon
Figure 9.7 Using mget under Windows NT requires a response to each file
prompt, which can be simply a carriage return.
498
chapter nine
the operating system used, to files within a directory for either anonymous or
non-anonymous users, with the latter a term used to denote persons who have
an account on the server. However, there is no mechanism that this author is
aware of that enables an FTP server administrator or a router administrator to
selectively enable or disable individual FTP commands. Thus, an FTP proxy
service provides the FTP server administrator with a significantly enhanced
capability, which can be used to configure the capability and features of ftp
services that other users can access.
Firewall Location
The capability to employ proxy services is based on the use of a firewall
located between a router and network servers connected to a LAN behind
the router.
To illustrate the common placement of a firewall, as well as a term associated
with its use, let’s assume we wish to add protection to an Ethernet LAN
connected to the Internet. One common method used to protect an internal
private network from packets flowing from the Internet is to place a firewall
between the router and the network to be protected. In doing so you would
install an essentially non-populated hub, whose only connections would
be to the router and firewall as illustrated in Figure 9.8. Because there are
no workstations, servers or any other device except the router and firewall
connections on this hub, it is referred to as a DMZ LAN. Here the term DMZ
is an acronym for ‘‘demilitarized’’ and originated from a strip of land where
no military activity occurred. If you examine Figure 9.8, you will note that
this network configuration insures that the flow of packets to and from the
Internet has to pass through the firewall before the packets can effect a host
located either on the public Internet or on the private network.
The firewall illustrated in Figure 9.8 represents a generic device, the functionality of which is highly dependent upon the product selected. Just about
all firewalls support packet filtering similar to the filtering performed by router
access lists as a basic capability. Thereafter, the functionality of firewalls can
vary based upon functions and features incorporated into different vendor
products. These functions and features can include proxy services for different applications that allow administrators to control different application
commands, limiting the rate of pending connections to counter different types
of denial-of-service attacks. They also provide network address translation
to hide internal host addresses from direct attack via the Internet, perform
authentication, virus scanning, and even encryption of data, which is a necessity when creating a virtual private network (VPN). It should again be noted
security
499
Internet
Router
s0
e0
Non-populated
hub
DMZ
LAN
Firewall
198 . 78 . 46 . 0
Figure 9.8
Private network
Using a firewall to protect a private network.
that the specific features and functions performed by a firewall considerably
vary between vendor products, a situation we will note when we turn our
attention to two vendor products later in this section.
For organizations that support public access to a Web server, a second
common network configuration method is to separate the public and private
networks and limit firewall protection to the private network. An example of
this network configuration is illustrated in Figure 9.9. In this example a public
Web server is located on the 205.131.175.0 network while the firewall is used
to protect organizational hosts on the 198.78.46.0 network. If the Web server is
the only network device on the 205.131.175.0 network you would more than
likely create a router access list to limit inbound traffic to the HTTP protocol
500
chapter nine
Internet
Router
e0
s0
e1
Public
access
Hub
DMZ
LAN
Server
Firewall
205 . 131 . 175 . 0
Network
Private
network
198 . 78 . 46 . 0
Network
Figure 9.9 Protecting a private network with a firewall while protecting a
public access-list capability.
to the address of the Web server. You should also consider programming TCP
intercept to minimize the effect of SYN flooding to the Web server and block
commonly used spoofed addresses in your access list.
Another popular alternative configuration used with firewalls is to connect
both public- and private-access organizational networks behind the firewall as
illustrated in Figure 9.10. In this configuration the firewall is used to protect
both network segments. This action permits you to reduce or eliminate the
security
501
Internet
Router
s0
e0
Hub
Firewall
DMZ
LAN
Public
access
Server
Private
network
198 . 78 . 46 . 0
Network
Figure 9.10 Protecting both public- and private-access network segments
with a firewall.
access-list processing activity on the router as well as a router port. Because
many firewalls use PC-based platforms, it is normally much less expensive to
add an additional firewall in the form of a LAN adapter card than to add a
port to a router.
While it is possible to protect the public-access server by connecting the
network it is located on to another firewall port, the decision to do so is usually
based upon the operations supported by the public-access host or hosts. If the
502
chapter nine
server only provides HTTP support, usually protection via a router’s access
list should be sufficient. If your organization supports public access for HTTP,
FTP, Telnet and other applications, you would then probably elect to connect
the public-access network to the firewall as shown in Figure 9.10.
Types of Proxy Services
The type of proxy services that can be provided is only limited by the
requirements of an organization and the programming capability of firewall
programmers. Some of the more popular types of proxy services include
remote terminal Telnet and TN3720 proxy services, Simple Mail Transport
Protocol (SMTP) proxy service, HyperText Transport Protocol (HTTP) proxy
service, the previously discussed FTP proxy service and ICMP proxy service.
The last represents a special type of proxy service, which deserves a degree of
elaboration due to the enhanced security capability it provides against certain
types of hacker attacks.
The Internet Control Message Protocol (ICMP) represents a layer 3 protocol
within the TCP/IP protocol suite. ICMP is used to transmit error messages as
well as status queries and responses to those queries. ICMP packets are formed
by the use of an Internet Protocol (IP) header that contains an appropriate
numeric in its Type field, which identifies the packet as an ICMP packet.
Although the use of ICMP is primarily oriented towards transporting error
messages between devices operating a TCP/IP protocol stack and is transparent
to users of a network, the protocol is also popularly used by many individuals
who are probably unaware that they are using transmitting ICMP packets. This
is because two of the more popular types of ICMP packets are Echo Request
and Echo Reply, which are better known to most persons as a ping operation
or application.
The use of ping is primarily designed as a mechanism to allow a user to
determine if a remote host is operational and using a TCP/IP protocol stack.
Pinging a distant host with an ICMP Echo Request packet results in the distant
host returning an ICMP Echo Reply packet if the distant host is reachable, if it
is operational and if its TCP/IP stack is functioning. While the use of ping can
indicate if a distant host is reachable, a ping timeout does not necessarily mean
that the distant host is not operational, because one or more communications
devices in the path to the distant host could be down while the distant host is
operational. However, in most cases ping represents the first troubleshooting
method to use when it appears that a host is not responding to a query.
In addition to indicating that a host is potentially reachable and operational,
the use of ping provides information concerning the round-trip delay to
security
503
a remote host. This information results from the ping application on the
originator setting a clock and noting the time until a response is received
or a timeout period occurs and no response is received. The time between
the transmission of the ping and the receipt of a response represents the
packet round-trip delay and can provide valuable information about why
time-dependent operations, such as Voice over IP (VoIP), produce reproduce
a voice that sounds like a famous mouse instead of a person.
Although continuous pinging may appear innocent, in actuality it represents
a method for a hacker to initiate a denial-of-service attack. This is because the
pinged host must stop what it is doing, even if only for a few milliseconds, and
respond to the ping with an Echo Reply ICMP packet. If the person that sets
the ping application to continuous pinging also sets the packet size beyond its
default size of 32 or 64 bytes, depending upon implementation, that person
forces the destination to respond with responses of increased length, which
requires the use of additional network resources. Thus, although the use of the
ping application may not bring the destination host completely to its knees,
ping can be configured to operate in a manner that can significantly interfere
with the ability of the destination to perform its intended operations.
Another problem associated with the unrestricted use of ping is that it can
be used as a mechanism to discover hosts on a distant network as a prelude
for attacking those hosts. For example, a hacker could write a script to cycle
though all 254 possible addresses on a Class C IP network as a mechanism to
discover which addresses are currently operational.
Based upon the preceding, many organizations may wish to control the
operation of ping and other types of ICMP messages. While many router access
lists provide administrators with the ability to filter ICMP packets based upon
source and/or destination IP address and the type of ICMP message, such
access-list filtering is an all or nothing operation. That is, a router access list
cannot selectively examine and note that a sequence of ICMP Echo Requests
from the same source address occurred after a predefined number of requests
flowed through the router and determine that subsequent requests should be
prohibited. In comparison, on some firewalls an ICMP proxy service feature
can be configured to differentiate between a single sequence of Echo Request
packets and the intentional or unintentional setting of a ping application to
continuously ping a host. Similarly, an ICMP proxy service capability can be
employed to distinguish between a person who may have difficulty accessing
a server and another person who is using the ping application in an attempt to
discover all hosts on your organization’s network. Thus, ICMP proxy service
represents an important type of proxy service, the use of which can enhance
the security of a network.
504
chapter nine
Limitations
Although proxy services can provide a considerable degree of security
enhancement to networks, there are certain limitations associated with their
use that warrant discussion. First and foremost, a proxy service requires a
detailed examination of the contents of individual and sequences of individual but related packets, forcing the application to look deeper into each packet.
This results in an additional degree of processing occurring on each packet,
which introduces a degree of delay. Second, a sequence of packets may have
to be examined to determine if it is acceptable to enable those packets to flow
to their destination. This means that one or more packets in each sequence
may have to be buffered or temporarily stored until the proxy service can
determine if the packets can proceed to their destination or should be sent to
the great bit bucket in the sky. This also means that additional buffer storage
in the proxy server or firewall will be required and the temporary storage of
packets adds to the latency of remote requests flowing to servers operated by
your organization. In fact, according to tests performed by several communications testing laboratories, as a result of proxy delay, using proxy services
from different vendor firewalls results in between 20 and 40 percent of the
bandwidth of an Internet connection at the proxy server side being unused
during the full utilization on the other side. This also results in a packet loss
of between 20 and 40 percent, resulting in additional remote transmissions to
the desired site. Thus, you must consider the effect of proxy service delays
and balance the potential need to upgrade your Internet access line against
the potential enhancements to the security of your organization’s network.
Operational Examples
Now that we have an appreciation of the capabilities of a proxy firewall,
we will turn our attention to an examination of the capabilities of two
firewalls. First, we will examine several configuration screens generated by
the Interceptor firewall, a product of Technologic of Atlanta, GA. Once this is
accomplished, we will turn our attention to FireWall-1, a product of Check
Point Software Technologies Ltd of Ramat Gan, Israel and Redwood City, CA.
The Technologic Interceptor
Similar to most modern firewalls, the Technologic Interceptor supports its
configuration via a Web browser. Figure 9.11 illustrates the Interceptor’s
Advanced Policy Options screen, in which the cursor is shown pointed to the
toggled check associated with the FTP Put command to block FTP uploads. In
security
505
Figure 9.11 Using the Technologic Interceptor firewall configuration screen
to block all FTP Put commands.
examining Figure 9.11 and subsequent Interceptor screen displays, you will
note they represent HTML screens displayed using a Netscape browser. The
Technologic Interceptor firewall generates HTML forms to enable network
managers to view, add, and modify firewall configuration data. To secure such
operations the firewall uses encryption and authentication by supporting the
Secure Sockets Layer (SSL) protocol for encrypting all traffic between the
firewall and a Web browser used to configure the firewall while passwords are
used for authentication. This means network managers can safely configure
the firewall via the World Wide Web.
Using Classes
The Technologic Interceptor firewall includes a class definition facility, which
provides users with a mechanism to replace address patterns, times of day, or
URLs by symbolic names. Classes are initiated by selecting the Classes button
506
chapter nine
on the left portion of the configuration screen. By using an equal sign (=) as a
prefix, they are distinguished from literal patterns.
Through the use of classes you can considerably facilitate the configuration
of the Technologic Interceptor firewall. For example, suppose you want to
control access from users behind the firewall to Internet services. To do so you
would first enter the IP addresses of computers that will be given permission
to access common services you wish them to use. Then you would define a
class name that would be associated with the group of IP addresses and create
a policy that defines the services the members of the class are authorized
to use.
Figure 9.12 illustrates the use of the Technologic Interceptor Edit Policy
configuration screen to enable inbound traffic for FTP, HTTP, Telnet, and
SNMP. Note that this policy uses the classname ‘‘= ALL-Internal-Hosts’’ in
Figure 9.12 Using the Technologic Interceptor firewall to create a policy
allowing outbound FTP, HTTP, Telnet, and SMTP traffic from all users in the
previously defined class ‘‘All− Internal-Hosts.’’.
security
507
the box labeled ‘‘From.’’ Although not shown, you would have first used the
class configuration to enter that class name and the IP addresses you want
associated with that class. Then, this new edit policy would allow those IP
addresses in the predefined class = ALL-Internal-Hosts to use FTP, HTTP,
Telnet, and SMTP applications.
Alert Generation
The capability of a firewall is significantly enhanced by an alert generation
capability, enabling a firewall to alert a network manager or administrator to a
possible attack on their network. Figure 9.13 illustrates the Technologic Interceptor Add Alert screen display with the IP-Spoof pattern shown selected.
In the example shown in Figure 9.13 the IP Spoof alert is used as a mechanism to denote a connection request occurring from a host claiming to have an
IP address that does not belong to it. In actuality, it can be very difficult to note
Figure 9.13 Using the Technologic Interceptor firewall Add Alert configuration screen.
508
chapter nine
the occurrence of IP spoofing. Although it is relatively easy to consider such
source IP addresses as those in RFC 1918, all zeros, all ones, loopback and
your network as spoofed IP addresses, others are difficult, if not impossible,
to detect. This is because unless the firewall previously obtained information about IP addresses, such as their locations on segments whose access
is obtained via different firewall ports, or notes restrictions on service by IP
address, it will assume an IP address is valid. In comparison, other patterns
such as refused connections or failed authentication are much easier to note.
For each alert you would first specify a name for the alert definition, such
as IP-Spoof for that pattern. After selecting the pattern, you can specify the
days and times of day and the frequency of occurrence that, when matched,
should generate an alert. The Interceptor supports two methods of alert generation — via either email or pager. If you select the use of a pager to transmit
an alert you can include a message, such as a numeric alert code, to inform
the recipient of the type of alert.
Packet Filtering
In concluding our brief examination of the operation of the Interceptor firewall,
we will examine the initiation of packet filtering. Although the packet filtering
capability of firewalls functions in a manner similar to that router feature, the
firewall is usually easier to configure and provides more flexibility in enabling
or disabling access based upon the set of rules that can be developed.
Figure 9.14 illustrates the Technologic Interceptor Network Services display, which lists the protocols for which this firewall accepts connections.
Note that the HTTP protocol is shown selected as we will edit that service. Also note the columns labeled ‘‘Max’’ and ‘‘Rate.’’ The column labeled
‘‘Max’’ indicates the maximum number of simultaneous connections allowed
for each service, while the column labeled ‘‘Rate’’ indicates the maximum
rate of new connections for each service on a per minute basis. By specifying entries for one or both columns, you can significantly control access to
the network services you provide as well as balance the loads on heavily
used services.
Figure 9.15 illustrates the Technologic Interceptor Edit Service display
configuration screen. In this example, HTTP service is enabled for up to 256
connections, and a queue size of 64 was entered, limiting TCP HTTP pending
connections to that value. The Max Rate entry of 300 represents the maximum
rate of new connections that will be allowed to an HTTP service. Once this
rate is exceeded, the firewall will temporarily disable access to that service
for a period of one minute. If you allow both internal and external access to
an internal Web server, the ability to control the maximum rate of incoming
security
509
Figure 9.14 Using the Technologic Interceptor firewall configure screen to
edit HTTP network services.
connections to a particular service can be an important weapon in the war
against so-called denial-of-service attacks. Under this technique, a malicious
person or group of hackers program one or more computers to issue bogus
service initiation requests using RFC 1918, loopback addresses, or random IP
addresses that more than likely do not exist. Since each access request results
in a server initiating a handshake response, the response is directed to a bogus
address that does not respond. The server will typically keep the connection
open for 60 or 120 seconds, which represents a period of time a valid user
may not be able to access the server when its connection capability is at
the maximum. If you compare this method of limiting HTTP requests to the
TCP intercept capability of Cisco routers presented in the first section in this
chapter, you will note that the latter represents a better weapon against SYN
flooding when compared to earlier versions of Cisco’s IOS. This illustrates an
important concept. That is, there are certain security features that a firewall
510
chapter nine
Figure 9.15 Using the Technologic Interceptor firewall Edit Service configuration display to set a series of rules to govern access to HTTP.
provides that may not be available from routers and vice versa. Thus, it is
important to carefully consider the capabilities of each device.
CheckPoint FireWall-1
FireWall-1 represents a scaleable, modular product from CheckPoint Systems
that runs on a variety of platforms ranging from Microsoft’s Windows NT to
Sun’s Solaris, HP-UX and IBM’s AIX version of UMX. FireWall-1 consists
of a troika of components, including a Graphical User Interface client that
operates on the previously mentioned platforms as well as on Windows 95
and Windows 98, a management server onto which security policies created
via the GUI are saved, and a FireWall module. The last represents software
that operates on a device that performs actual packet examination. The device
that operates the FireWall module is normally considered as a firewall and
security
511
is located at network choke points, where it can examine all packets flowing
between a public and a private network or another network access location.
The Graphic User Interface
The Graphic User Interface provides the mechanism for defining an organization’s security policy. To accomplish this, you would first review your network
configuration and devise a security policy that corresponds to your organization’s security requirements and its network configuration. For example, let’s
assume your organization plans to place public-access devices, such as mail
servers and Web servers, on a separate LAN behind a firewall, while placing your organization’s private network resources onto a separate connection
behind the firewall. Figure 9.16 illustrates this network configuration. Note
that in this example the public network resources placed on a segment located
behind the firewall are referred to as a DMZ.
For the configuration illustrated in Figure 9.16 the security policy that could
be developed can vary considerably from organization to organization. One
possible policy for illustrative purposes would be to allow external users to
Internet
Firewall
Router
Private
local
network
DMZ LAN
Private network
resources
Web
server
Mail
server
Figure 9.16 Protecting both private and publicly accessible network segments through the use of a firewall.
512
chapter nine
access the local network facilities only to surf the organization’s public Web
server or to send mail. In the outbound direction let’s assume internal users
will be provided with the ability to access all network resources to include the
servers on the DMZ LAN as well as the Internet. Note that this policy protects
private network resources from the untrusted side of the network represented
by the Internet but does not place any restriction on local users on the private
network segment.
Network Objects
Once you have developed a proposed security policy, you would use FireWall1’s object manager to define one or more network objects that are used by the
program’s rule base. Network objects can represent any device that has an IP
address, such as a router interface, a Web server, or the hosts on an internal
or external network.
From the Rule Base Editor you can invoke the program’s Network Properties and Router Properties dialog boxes to define various types of network
objects. Figure 9.17 illustrates the entry of configuration information into the
two previously mentioned dialog boxes during the network object definition
process. The box on the left illustrates the entry of general network properties,
including the name assigned to a local network, its IP address, network mask,
and applicable comment. Note that, by a simple cursor click, you can define
the location as either internal or external and either permit or deny broadcasts.
The right dialog box provides a similar network object definition process for
a router.
Figure 9.17
a router.
Assigning network object definitions for a local network and
security
513
For our previously described example, the firewall, Web server, mail server,
and the private local network would represent network objects to be defined.
Once this action has been accomplished, you would define the services to be
used to implement your organization’s security policy. In doing so you would
create individual rules that form a rule base.
Firewall-1 Rules
The individual rules supported by Firewall-1 includes seven fields and follows
the format shown in the column headings of Table 9.5. In this format both
source and destination fields can be set to represent a specific object or any
device. For example, let’s assume that you previously used the program’s
object manager to assign the label web-server to that server. If you want to
provide any internal or external user with the ability to access the Web server,
you would develop a rule with ‘‘any’’ in the source field and ‘‘web-server’’
in the destination field. Because the Web server supports HTTP, you would
configure the rule for that service. Concerning the action field, since you want
to allow traffic you would configure that field to ‘‘accept.’’ In comparison, if
you wish to block packet flow to the Web server the action field would be set
to ‘‘drop.’’ The track field can be used to log information or to generate alerts.
The field labeled ‘‘Install On’’ denotes the platform upon which the rule being
developed will operate. Finally, the field labeled ‘‘Time’’ provides you with
the ability to control when the rule remains in effect. The first two rules given
in Table 9.5 permit external users to send requests to the Web server and mail
to the mail server at any time.
Because we also want to allow users on the internal private network to
access both the Internet and the organization’s public accessible network, we
need the third rule given in Table 9.5.
Similar to router access lists, FireWall-1 will drop all packets that are not
explicitly permitted by a security policy. Also similar to our discussion of
access lists, rules like access-list statements are examined sequentially and
TABLE 9.5
Setting Rules in Firewall-1
Source
Destination
Service
Action
Any
Web-server
http
Accept
Any
Mail-server
smtp
Localnet
Any
Any
Any
Track
Install On
Time
Log
Gateways
Any
Accept
Log
Gateways
Any
Any
Accept
Log
Gateways
Any
Any
Reject
Long Log
Gateways
Any
514
chapter nine
the first rule that matches a connection is applied. Although the implicit rule
at the end of the FireWall-1 rule base is to drop all connections that do not
match the previous rules, on occasion you may wish to log such connection
attempts. To do so you could enter the final rule in Table 9.5 to terminate your
rule base.
Once you complete the creation of one or more rules necessary to establish
your organization’s security policy, you would load the code generated by the
rules onto one or more gateways.
One of the more interesting aspects of FireWall-1 is that it can convert your
general rules into a series of access-list statements that can be installed on any
router that supports the Open Security Extension (OSE). This means that you
can install your policy on a pure firewall as well as on certain router products.
Figure 9.18 illustrates an example of the use of the CheckPoint Software
policy editor to create a series of six rules. Note that the last rule in effect
represents an implicit deny all; however, instead of simply sending such
traffic to the bit bucket, its occurrence will be used to generate alert messages.
Management Functions
Although the previous rule creation example reminds us of access lists, in
actuality you can assign a sophisticated series of properties to one or more
rules. Properties are assigned through the use of a Properties Setup dialog
Figure 9.18 The CheckPoint Policy editor uses six fields to define rules that
are evaluated top to bottom similar to an access list.
security
515
box and specify both general and specific aspects associated with the flow of
information. For example, through the Properties Setup dialog box you can
define the direction by which gateway rules will be applied as well as the
length of TCP session timeouts, encryption, authentication and whether or
not to allow RIP, DNS, and ICMP packets.
Figure 9.19 illustrates the Properties Setup dialog box. At the top of the
box are a series of tabs that enable you to perform the indicated functions
labeled on each tab. In the foreground the Security Policy tab was activated.
As you can note from the display, you can use a drop-down menu to select
the direction of packet filtering as well as a simple click to enable numerous
functions similar to an applicable router access-list permit statement.
Concerning authentication, the FireWall-1 supports eight authentication
schemes, which range from password to S/Key, SecurID Tokens, RADIUS
server and digital certificates. In addition, FireWall-1 also supports the
creation of virtual private networks by providing a mechanism to encrypt
communications destined for other networks attached to the Internet, while
passing packets flowing to non-VPN locations in clear. While many of the
functions performed by this firewall are similar to the capabilities performed
by a router’s access list, it is important to note two key differences between the
two. First, as a stand-alone device a firewall provides more security-related
Figure 9.19
log box.
Examining the CheckPoint FireWall-1 Properties Setup dia-
516
chapter nine
functions than commonly available through router access lists. For example,
FireWall-1 can provide content checking for HTTP, SNMP and FTP to include
anti-virus checking. Second, by off-loading communications security checking
to a firewall you should obtain a higher level of router performance.
9.3 The Role of the Virus Scanner and Encryption
In concluding this chapter we will become acquainted with a key network
security-related area — virus scanning. In this section we will briefly review
some of the methods by which malicious code can adversely affect the
operation of a computer. Once this is accomplished, we will describe and
discuss the use of virus scanners to protect hosts on a network from receiving
malicious code.
Virus Overview
A virus represents a program that reproduces its own code, placing its code
in other programs so that, when those programs are executed, the infection
spreads. Although not technically viruses, there are three additional types of
malicious software that a good virus scanner will check for and which we
should note. Those additional types of malicious software are logic bombs,
worms, and Trojan horses.
Logic Bombs
A logic bomb represents a program that lies dormant until it is triggered by
a specific event. The trigger can range in scope from a specific date and time
encoded in the software to the receipt of a command to initiate its predefined
operation. Once activated, the logic bomb will normally perform some type
of insidious operation, such as changing the value of data previously stored
on disk, altering the value of a RAM memory location, or attacking a third
party. Concerning the latter, the distributed denial-of-service (DDoS) attack
employed against such popular Internet sites as Yahoo and EBay resulted
from code placed on third-party hosts that were activated by command.
Worms
A work is a program that reproduces itself. Unlike a virus that both adversely
infects and reproduces, the worm is relatively benign as it does not alter
or destroy data. However, each copy of the worm creates additional copies,
which can rapidly consume all available computer resources.
security
517
Worms are usually planted on networks and in multi-processing operating
systems where they can do the most harm. In fact, the well-known ‘‘Internet
virus’’ that brought several thousand computers connected to that network to
a halt during the early 1990s was actually a worm.
Trojan Horse
A Trojan horse represents a program that functions as a delivery vehicle for
destructive code that will adversely affect the operation of a computer. Named
after the Trojan horse that hid soldiers, a Trojan program will appear useful
but when executed can function as a logic bomb, worm, or virus.
Virus Effect
According to several virus-scanning programs with a built-in database describing different malicious software, there are thousands of identified codes that
perform different types of malicious functions. Thus, it is a bit difficult to
attempt to categorize the effect of a virus. However, we can note that the
actual effect of a virus can vary from humorous to catastrophic. Some viruses
may simply display the message ‘‘X Shopping Days till Christmas,’’ while
other viruses can alter your hard drive’s boot sector, resulting in the apparent
inability to access information previously stored on your hard drive.
One of the most popular methods used by virus developers to replicate
their code is to name their program as an external command file with the
extension .COM when the real command program has the same name but the
extension .EXE. The reason for doing so is that the entry of a command name
results in the operating system executing the file with the extension .COM if
it encounters both .COM and .EXE file extensions with the same file name.
A smartly constructed virus using the .COM extension will also execute the
actual .EXE command as it spreads its infection. Thus, for a period of time
until the cumulative effect of the virus becomes noticeable it will be hidden
because the command entered by the user will execute as expected. This type
of virus is referred to as a .COM infection.
Types of Viruses
The types of viruses you can encounter depend upon the ingenuity of the
hackers and pranksters who typically develop these programs. Although the
types of viruses are virtually unlimited, they can be categorized by the most
common areas they attempt to infect. These areas are the boot sector and file
allocation table (FAT), system files, and operating system commands.
518
chapter nine
Boot and FAT Infectors
There are two locations that many viruses commonly attack due to the harm
their alterations can cause. Those locations are the boot sector and the file
allocation table. By changing information in the boot sector, a virus can make
your hard disk appear to be unusable. By changing data in the FAT, a virus
can make portions of your disk literally ‘‘disappear’’ simply by marking good
sectors as bad or changing a pointer to the next cluster used by a file to an
end-of-file cluster value. A few viruses have been known to swap the starting
FAT entry address in the root directory, which when performed on a database
file can result in the subsequent execution of a database processing program
using an output file as an input file and vice versa, a truly insidious trick!
System File Infectors
A system file virus infector changes one of the operating system files. Since
system files are always executed upon power-on, their modification functions
as an easy mechanism to replicate a virus. The most popular method to
develop a system file infector is for the virus developer to attach their program
code to a system file. A virus scanner can easily note the occurrence of this
type of file by comparing its size to the size of the legitimate system file. Far
more difficult to detect is the virus that rearranges system file code so it can
insert itself into the file without altering the size of the file. Fortunately, very
few persons have the knowledge and are willing to devote the time required
to develop this type of virus. Even if someone does, many virus scanners can
be configured to prevent any executable system file from being sent over a
corporate network.
Command Infectors
As previously noted, a command (.COM) infector virus either attaches itself
to an external command file or hides itself by naming itself after an .EXE file
using the extension .COM. since operating system commands are frequently
executed many times during the day by many computer users, a command
infector achieves the ability to replicate itself frequently by pretending to be
a command. Now that we have a basic knowledge of the types of software
attacks we can encounter and the manner in which most viruses designed to
cause harm operate, let’s focus our attention upon methods we can consider
to prevent a software attack and, if unsuccessful in our prevention methods,
recover from the effect of an attack.
security
519
Infection Prevention
According to an article published in a recent edition of a national magazine,
over 5000 types of worms, logic bombs, and viruses have been identified. This
means that it is highly probable that no one prevention and detection strategy
or virus scanning program can be expected to represent an absolute guarantee
against infection. However, such programs do represent a good starting point
to protect your computer.
Infection prevention software functions as a filter, examining computer
operations susceptible to modification by different types of attack software.
Most infection prevention software monitors all I/O operations and disallows
any attempt by code to modify critical disk areas, such as the boot sector,
FAT and areas where executable programs that are part of the operating
system reside.
Typically, you load the infection prevention software on your hard drive
and either modify your AUTOEXEC.BAT file to load the program each time
you use the computer or place the program for automatic startup under
Windows. For either method the infection program functions as a barrier to
most of the commonly used attack software techniques known to result in
destructive operations.
Since infection prevention software is designed to detect attack software
once it begins to operate, such methods are normally used to locate dormant
infections. To identify previously stored attack software commonly requires
the use of detection software.
Detection Software
A detection program, now commonly referred to as a virus scanner, searches
your disk looking for identifying attributes associated with different types of
attach software. Such software may examine the names and size of files to
determine if they are disguised executable files, look for replication code, and
search system files and the contents of memory looking for any suspicious
alterations.
Virus Scanning
Today both infection prevention and detection of malicious code is commonly
performed by one program referred to as a virus scanner. To obtain an
appreciation of the role of a virus scanner, we will turn our attention to two
types of products. One type of product operates as a stand-alone program on
a desktop, while the second type of program guards the hosts on a network
from malicious email, typically by examining inbound email.
520
chapter nine
Desktop Scanning
To illustrate the operation of a desktop virus scanner, this author will use
Command Software System’s Command AntiVirus program. Figure 9.20 illustrates the main window of this program. Note that this program is installed
with five predefined tasks, which are listed within the window. To illustrate
the operation of this program, this author will use it to scan his computer’s
hard drives. However, note that the program is also capable of scanning network drives. In addition, you can also create new tasks by clicking on the
button with that label. Once you do so you can enter a name for the task and
use a resulting dialog box labeled Properties to specify the type of scanning to
be performed by the task. Later in this section we will examine the dialog box
labeled Properties and the properties that can be configured for a new task.
General Features
One of the key features of the Command AntiVirus program is its ability to
schedule a virus scan as well as send a message when an infection is found.
Through the use of the program’s Preferences menu you can send a message.
Figure 6.2 illustrates the program’s Network dialog box, which is selected
from the Preferences menu. In examining Figure 9.21, note that it provides
you with the ability to display a message when an infection is found as well as
to send an email. This program integrates very well with Microsoft’s Outlook,
allowing a user to email a report as well as, if desired, the infected files
discovered by the scanning process. Thus, you can configure the program to
execute at predefined times and transmit via email any encountered infections
to a help desk or another central location.
When you use the program dynamically, you can control both the scanning
process and the action when an infection is noted. To do so you would return
Figure 9.20 The main window of Command Software’s Command AntiVirus
program includes predefined tasks as well as allows the user to create
new tasks.
security
521
Figure 9.21 The Command AntiVirus program supports a messaging option,
which enables other members of an organization to be alerted when a scan
indicates the discovery of the infection of a computer.
to the Preferences menu and select the entry labeled Active Protection. This
action will result in the display of a dialog box labeled Active Protection,
which contains two tabs labeled Dynamic Virus Protection and Memory
Scanning. Figure 9.22 illustrate this dialog box with the tab labeled Dynamic
Virus Protection in the foreground.
In examining Figure 9.22 note that the checkbox next to enable DVP must
be selected to enable Dynamic Virus Protection. When enabled, the program
will operate in the background and dynamically scan the computer resources
selected as well as perform the selected action when an infection is encountered. Concerning the latter, note that the right column provides you with five
options concerning the action to be performed in the event an infection is
encountered.
Figure 9.22 The Dynamic Virus Protection tab in the Active Protection dialog
box provides users with the ability to specify what computer resources will
be scanned and the action to be performed if an infection is determined.
522
chapter nine
Figure 9.23 The Memory Scanning tab in the Active Protection dialog box
provides users with the ability to control the scanning of the first 640 kbytes
of computer RAM.
The second tab in the Active Protection dialog box is labeled Memory Scanning. Figure 9.23 illustrates the tab, which is now located in the foreground of
the dialog box. Note that the version of the Command AntiVirus program used
by this author was limited to allowing a scan of the first 640 KB of memory.
File Scanning Control
If you are scanning a large-capacity hard drive or many network drives,
the scanning process can require a considerable amount of time. Perhaps
recognizing this fact, the Command AntiVirus program provides users with
the ability to specify the types of files to be included or excluded from a scan.
To do so you would return to the program’s Preferences menu and select the
Files to Include/Exclude option. Figure 9.24 illustrates the dialog box labeled
Files to Include/Exclude. Note that in addition to predefining file extensions
you can add new file extensions for inclusion or exclusion in a scan.
The Scanning Process
Since the old adage ‘‘the proof of the pudding is in the eating’’ holds true
today, this author decided to run a scan. To do so, the task labeled ‘‘Scan
Hard Drives,’’ previously shown in Figure 9.20, was first selected. The box
in the right portion of the display labeled ‘‘Execute Task’’ was then click
on. Figure 9.25 illustrates the initial execution of the Command AntiVirus
program at a point in time where 856 files were scanned. As the program
security
523
Figure 9.24 The dialog box labeled Files to Include/Exclude provides users
with the ability to specify files via predefined or entered extensions for
inclusion or exclusion in a subsequent scan.
Figure 9.25
Viewing the progress of a virus scan.
scans your hard drive it displays a running progress of the scanning activity
as a horizontal bar chart as well as by informing you of the path to the file
being scanned. Note that at the particular point in time when the screen was
captured no viruses were found.
Virus Information
One of the more interesting tutorial aspects of the Command AntiVirus
program is its database of virus information. This database is arranged alphabetically by virus name, as is illustrated in Figure 9.26. As you select a virus
524
chapter nine
Figure 9.26 Using the Command AntiVirus program to display information
about a virus.
in the left portion of the screen, the right portion of the screen will display
information about the selected virus. This information can be very helpful in
allowing one to sleep at night. For example, assume a virus scan performed
once a week located the 3Tunes virus, which you then deleted. You would
now more than likely want to know the potential effect of the virus. Although
you could more than likely guess its action from the name of the virus in this
example, this is not always the case. Thus, the ability to view a description
of encountered viruses can be an important consideration when selecting a
virus scanner in addition to the number of viruses in its scanning database
and other program features.
Controlling the Virus Scan
In concluding our examination of the Command AntiVirus program we will
turn our attention to its Properties dialog box, which represents the key to
controlling program scans. Figure 9.27 illustrates the Properties dialog box
after the Scan Hard Drives task was selected.
In examining Figure 9.27 note that you can specify the action to be taken if
an infected file is encountered. You can also specify the drive and paths to be
scanned, the types of files to be scanned, and other information, which you
can note by examining the screen. The Properties dialog box also includes a
button labeled Schedule. As you might expect, you can select that button to
have scans automatically performed at predefined times. Now that we have
an appreciation of the operation of a typical virus scanning program, let’s turn
our attention to the scanning of email.
security
525
Figure 9.27 The Command AntiVirus Properties dialog box provides users
with the ability to control both when and how a virus scan is performed as
well as the actions to occur if infected files are discovered.
Email Scanning
According to several surveys of Internet users, one of the highest ranking
reasons for the use of the ‘‘mother of all networks’’ is email. While email is
certainly a key productivity enhancement method for both residential and
business oriented persons, i