Logical communication - McGraw

Logical communication - McGraw

365c01.qxd 11/28/01 4:33 PM Page 1

H

Basic Concepts,

Network Operation,

1

A

P

and Cabling

T

E

ITINERARY

••

Objective 1.01

Internetworking Overview

••

Objective 1.02

OSI Structure and Definitions

••

Objective 1.03

Ethernet and Other Media Types

••

Objective 1.04

Network Devices and Relation Within the Networking Model

••

Objective 1.05

Cabling the LAN and WAN

ETA

NEWBIE

5+ hours

SOME EXPERIENCE

3 hours

VETERAN

1 hour

1

365c01.qxd 11/28/01 4:33 PM Page 2

2

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

To make a network run efficiently, there must be organization and coordination.

There are millions of networks connected together to form the Internet. There are millions of unconnected networks used for educational, corporate, and other purposes. To organize this huge networking structure, models are developed, technical terms defined, and technology and protocols invented. There are basics that internetworking specialists must use as a starting point of reference. These basic models are put in place to make sense of the complexities involved in connecting users.

Objective 1.01

Internetworking Overview

I n network design, there is a three-tier model to present a simplified way of looking at the general structure of internetworks. The three tiers are

Core

Distribution

Access

These terms attempt to simplify how internetworks should be designed and structured. They provide structure that not only describes, but also logically assists in planning and maintaining an internetwork.

Local Lingo

Network, internetwork

The terms network and internetwork have two distinct meanings. Network is generally defined as a single routed subnet. Internetwork refers to the entire network. Many times the terms are used interchangeably to mean the entire network.

Core

The core is the backbone of an internetwork. It is what allows individual networks to pass traffic to other areas quickly. It is not enough to have a transit area to pass information from one place to another; it should do so as fast as possible. Anything that would hinder the quick delivery of information from one area of the internet-

365c01.qxd 11/28/01 4:33 PM Page 3

CHAPTER 1

Basic Concepts, Network Operation, and Cabling work to another should be eliminated. In an internetwork with well-defined core, distribution, and access layers, the core should be kept free of potential bottlenecks and failure points such as servers, routers, and slow protocols. Anything that would slow the speed of the core needs to be placed in a lower layer.

3

Distribution

The distribution layer is defined as the layer of policy. This is where departments and sections of your network are partitioned off so that there is structure, security, and routing, and an administrator can control and manage the internetwork.

WAN connections from other sites are connected here, and VLANs and access controls are implemented. Think of the distribution layer as the layer that allows you to manage and implement policy in the internetwork. Everything flows from policy. Policy controls routing, placement of access lists, implementation of

VLANs, and even the overall structure of the network.

Access

The access layer is where the user interfaces to the network. Whether by LAN or

WAN connections, the user gains access here. For equipment, there are workgroup switches, access routers running technologies like ADSL, cable modems, ISDN, and asynchronous dialup. This area uses media types like Ethernet, Token Ring, or WAN media types as listed previously. Whereas the distribution layer uplinks to the core through 100 Mbps, Gigabit Ethernet, or ATM, the access layer uplinks to the distribution layer through Ethernet, Fast Ethernet, Token Ring, or WAN connections. Figure 1-1 shows the three well-defined tiers of a network.

FIGURE 1.1

Three-tier model

365c01.qxd 11/28/01 4:33 PM Page 4

4

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

Objective 1.02

OSI Structure and Definitions

C onnecting users to accomplish tasks may seem simple on the surface—two users talking to one another. Implementation of this is much harder. From a human perspective, communication is simple, as sound is transmitted in certain patterns that another human listens to and interprets. However, if we begin to dissect the sum of what makes up human communication, there is great complexity.

The functions of the brain, throat, ears, physical features, a common protocol (language) that has been learned, sound, and air all play a part. To break these down in further divisions allows for understanding of the parts and how they relate to the whole. The same is true with internetworking. The big picture is internetwork communication; internetwork models help present things in an organized manner.

We have seen the three-tier model of design; now let’s focus on the Open

Systems Interconnection (OSI) model. Where the three-tier model is concerned with the structure of building networks, the OSI is concerned with the communication that transverses that network. The OSI describes the protocols and functions necessary to pass information across network devices so that communication between end systems is accomplished. More specifically, the OSI is used to do the following:

Simplify complex procedures into an easy-to-understand structure

Allow vendors to interoperate

Isolate problems from one layer to be passed to other areas

Allow modular plug-and-play functionality

Provide independence of each layer

There are seven layers of the OSI, as follows:

Application

Presentation

Session

Transport

Network

Data link

Physical

These layers can be grouped into two main sections called the upper and lower layers. The upper layers refer to the application, presentation, and session,

365c01.qxd 11/28/01 4:33 PM Page 5

CHAPTER 1

Basic Concepts, Network Operation, and Cabling while the lower layers are the transport, network, data link, and physical. The data link is further subdivided into the Logical Link Control (LLC) and the Media

Access Control (MAC).

Another way of dividing the model is the logical and the physical. This functional division describes the upper five layers as logical in nature and the lower two layers as physical in nature. As the model is explained, these two divisions will become apparent.

It is important to understand that the OSI is progressive in many ways.

Intelligence varies between layers. Each layer progressing up the model is generally more intelligent than the layer below it, with greater capabilities and complexities. However, intelligence comes at a price—each layer progressing up the model is generally slower than the layer below it. The application layer has the most intelligence of the OSI, but it is also the slowest layer. The physical is the fastest layer, but has little intelligence.

The model is designed so that each layer communicates with the corresponding layer in the remote system. The application layer doesn’t communicate with the presentation layer below it. It communicates with the application layer of the remote system. To accomplish this, headers are placed on the information as it progresses down the model. On the remote computer, the headers are stripped one by one as each layer recognizes and reads the header placed on the data from the respective layer in the remote system. This allows each layer to be independent of the layer above and below it to accomplish its tasks.

The basic building block of the OSI is the PDU, the protocol data unit. As the data is being passed from each layer, it is referred to as a PDU. When the information is passed from one layer to another, it is said that the PDU is passed to that layer. Each layer may add a header or a trailer to the PDU; this process is known as encapsulation. When the header is placed on the PDU, encapsulation means that the corresponding layer can only read the PDU in another system. The encapsulation process can be compared to mailing a Christmas gift to a friend. The gift might be a new shirt that is placed in a box, then wrapped and placed in a shipping box. The shirt, the box, the gift wrap, and the shipping box all may have a tag, which identifies each piece. It could be said that each piece is encapsulated within the other. The shirt when placed in the box is not seen and is referred to as a box.

When the gift wrap is applied, it is called a gift, yet when it is placed in the shipping box, it is now a parcel. It is very similar with the OSI. Each layer encapsulates the PDU as it is moves down the model. As Figure 1-2 shows, while the PDU is at the application, presentation, or session layer, it is referred to as data. At the transport layer, though, it is known as a segment and the network layer encapsulation is referred to as a packet and the data link as a frame.

5

365c01.qxd 11/28/01 4:33 PM Page 6

6

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

FIGURE 1.2

The encapsulation process

In the early days of networking before the Internet, the TCP/IP model described all these functions within five layers. TCP/IP is much like the OSI, except the single application layer in the TCP/IP model defines the top three layers of the OSI. The other four layers are the same as the lower four of the OSI. What is important about that model is the terminology it uses when describing the PDUs being passed between layers. The TCP/IP model uses the keywords data, segment, packet, frame, and bits to describe the encapsulation process. As was seen in Figure 1-2, the data is passed from the top of the model downward, and each of the lower layers encapsulates the data with a header. The data link layer adds a trailer known as the frame check sequence (FCS) as well. This encapsulation process is then reversed—that is, headers are stripped off as the PDU is passed up the model on the remote system.

The Seven OSI Layers

Each layer of the OSI is responsible for activities that take data and move it with the proper formatting and coordination, determining where it will be sent, setting up connections with remote hosts, and ensuring integrity. The process of how this is accomplished will be described as we look at the individual layers of the OSI, beginning at the top of the stack with Layer 7.

Application

The application is the layer of the user. This is where the user interfaces with the network. It is important to understand that some applications are not considered network applications, such as word processors, spreadsheet programs, and database programs. These applications do not require network connectivity to work.

You can save each of these files on the computer’s hard drive. On the other hand,

365c01.qxd 11/28/01 4:33 PM Page 7

CHAPTER 1

Basic Concepts, Network Operation, and Cabling there are applications that require the network to work. Without network connectivity, the application is useless. Examples of network applications and computer applications can be seen in Table 1-1. These applications are used constantly in today’s internetworks.

7

Presentation

The presentation layer takes over when the data is passed down from the application. As the name would suggest, it is the job of the presentation layer to take the data and format it when passed to the remote user. The best term is code formatting. This layer puts the data in the form that will be required within the network communication process. Examples are ASCII, JPEG, TIFF, or even compression and/or encryption. Any manipulation of the data in terms of representation and formatting is done at this layer.

Session

Once the session layer takes over, it manages the inter-host communication. It keeps track of data from one application and keeps it separate so that the lower layers are able to distinguish what data is going to what host. Therefore, the session layer initiates, manages, and terminates data communication between hosts.

This layer typically handles logins and session parameter negotiation.

Transport

The transport layer is responsible for achieving what the session later began. It sets up reliable and unreliable communication from end systems. This end-to-end communication is dependent upon ports used for a particular application. When

TABLE 1.1

Network Applications vs. PC Applications

Network Applications

E-mail

Remote access

Network management

File transfer

PC Applications

Spreadsheet

Word processing

Database

Presentation

365c01.qxd 11/28/01 4:33 PM Page 8

8

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

HTTP data is sent down, the transport layer sets up communication from a source port on the local device to a destination port on the remote device. The transport layer is not concerned with the devices in between, only the end systems.

When it receives the data, it encapsulates this data in segments. The segments are created based on an agreed-upon segment size between the two end devices. This segment size is important for many reasons, such as reliable communication, flow control, and sequencing. The size of the segment is calculated by and communicated during the initial setup of the communication. With connection-oriented communication, there is communication between the two hosts at all times. It is because of this communication that each side knows what the other side is expecting.

Flow control is an important concept with segments because of the need to deliver the data without overwhelming the other end device. There are three basic forms of flow control at the transport layer:

Buffering allows space for unprocessed data until it can be attended to by the system.

Source-quench messaging is used to notify when the receiver has reached its limit of how much it can receive. When the receiver is being overwhelmed, it sends source-quench messages to stop the other side from sending data. Much like a traffic light, these ready/not ready messages allow the data to be sent in a measured amount without loss. However, if there is loss, there are retransmission messages that can notify the sender to resend a particular segment.

Windowing is perhaps the most important form of flow control for TCP.

It works by negotiating window sizes with the remote host. Think of the window as a buffer that constantly changes in size to allow more or less data to be received as necessary. It is the receiver that sets the window size. This number is used to tell the sender how much data the receiver can accept before the sender must stop sending and wait for an acknowledgment from the receiver.

Not only does flow control play a part, but reliability is also very important.

The two sides need to be on the same page for reliable, connection-oriented communication. If something is sent but not received, this is damaging for communication. TCP deals with this by using acknowledgments for every communication.

Acknowledgment is the most important part of reliable communication, because missing acknowledgments are a glaring light that means that something wasn’t sent properly or there was a problem in transit.

Sequencing is used so that when segments are received out of order, they can be put back in the proper order. This will make it apparent if something is missing in the transfer. Say a computer sent three segments, all with sequence numbers, but

365c01.qxd 11/28/01 4:33 PM Page 9

CHAPTER 1

Basic Concepts, Network Operation, and Cabling when they were received, one was missing. The receiving computer would acknowledge the two received but not the third, and the transmitting station could then resend the missing segment.

Connection-oriented communication can be explained by using the analogy of a phone call. In a phone call between two people, there are the following three components:

Call setup phase

Data transfer phase

Call termination phase

The call setup phase is when a person picks up the receiver, dials, and is answered on the remote end. The data transfer phase is when the person talks and the other person acknowledges and communicates in return. The termination phase is simply saying goodbye and hanging up. Connection-oriented protocols share in this behavior, having formal setup, transfer, and termination phases.

Connectionless communication is used when the protocol doesn’t need to provide reliable segment delivery. Connectionless data transfer can be likened to a megaphone approach to communication. The words are sent out without checking to see if those around are hearing them. There is no concept of flow control, acknowledgment, and sequencing. These functions are left up to the session or application layer processes of the remote host to perform. A transport layer protocol such as UDP is not responsible for reliability; it is up to the application to provide reliable communication as necessary.

9

Network

The network layer of the OSI is used to set up paths to various locations. Path determination is a good way to explain routing. Routing is simply the determination of the next hop that should be taken en route to the destination network. Networks are defined with a network-addressing scheme such as IP or IPX addresses. These addresses are used to give a host a location and address. Network layer addresses are not used to communicate with as much as to determine where the host resides in the internetwork. The network layer has the following characteristics:

Two-part addresses containing host and network portions

Typically, the first half is the network portion defining where hosts can reside, and the second half is the host portion itself.

Logical communication

It is logical in that hosts don’t communicate directly by network layer address. They provide for location and logical upper layer communication between hosts. Physical communication

365c01.qxd 11/28/01 4:33 PM Page 10

10

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT never occurs by network layer address. Instead, a lower layer address is used for this purpose.

Flat or hierarchical in nature

Flat means that there is little rhyme or reason in the relation of one address to another. There is only a comparison of network to host. IP is hierarchical, in that there are classes of addresses and, with the invention of the subnet mask, there are even more divisions in the class system of addressing.

Connectionless

The network layer protocols IP and IPX are connectionless in that they provide best-effort delivery of packets to other devices.

Typically includes a time-to-live (TTL) field in headers

This keeps packets caught in loops from being circulated forever. Each router hop decrements (IP) or increments (IPX) the TTL field by one.

Path selection

This is accomplished with routing tables on the device.

Network layer devices as well as most PCs have routing tables, also known as forwarding tables because they forward information based on interface and next-hop addresses toward the destination.

Unique addresses within the internetwork

Two hosts cannot share the same network and host address in the internetwork. No matter how large the network, there cannot be any duplicates. Not only does IP reside at the network layer, but so does ICMP. ICMP is a messaging protocol used by all TCP/IP hosts. When TCP/IP is installed on a device, ICMP is running and ready to send messages as needed.

Data Link

The data link layer is broken into two parts: the Logical Link Control (LLC) and the Media Access Control (MAC). The division between these two sublayers represents a major division in the OSI. Everything above this division is logical in nature; everything below is physical. This layer provides for physical addressing of network devices, flow control, and error detection. The physical device’s address must be known in order for communication between two devices to occur. This communication is always segment by segment. This is much different than what is defined at the transport layer. In the upper layers, communication is between end systems. At the data link layer, communication is between connected systems only. The addressing system at the data link is physical. For Ethernet and Token

Ring, it is the MAC address. This address is also known as a burned-in address

(BIA) and must be known between two systems before communication can take place. For WAN technologies like frame relay, the data link layer address is the

DLCI; ISDN uses the TEI.

365c01.qxd 11/28/01 4:33 PM Page 11

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

Characteristics of data link layer addressing are as follows:

Two-part 6-byte MAC address

The first 3 bytes are known as the vendor code or the organizational unique identifier (OUI), and the last

3 bytes are the serial number. In an address like 0000.0c1f.2a4f, 0000.0c

identifies the manufacturer of the network device as Cisco. The last half

(1f.2a4f) is the serial number of the device as defined by Cisco.

Flat addressing space

There is no routing or hierarchy in flat addressing space. Data link addresses don’t indicate where a device is located; rather, it is the identification of the device used for local communication.

Logical addresses are used for location. Physical addresses are used for direct communication and identification. The difference can be likened to an individual’s house address and Social Security number. Network layer addressing is similar to a house address, which is logical and hierarchical, used for location of the person. Data link addresses, which reside at the data link layer, are more like a Social Security number—the address doesn’t indicate a person’s location.

Unique only on the segment on which they reside

Because of the sheer number of devices manufactured, many times the MAC addresses are duplicated from older equipment to newer equipment. There is no problem with duplication as long as the duplicate MACs are not used on the same logical segment. As will be described later in the chapter, a logical segment consists of the devices called a network because they border a router interface.

Flow control and error detection

Data link layer communication means two hosts on the same network communicating. Any time there is communication, there must be some type of flow control. Because of

Layer 2 having less intelligence than the upper layers, there is error detection, not correction.

11

Exam Tip

While flow control is a description of the transport layer, the description topology and flow control are only possible at the data link layer.

Switches and bridges are examples of data link devices. A switch is the most popular Layer 2 device at the data link layer. Because switches and bridges are

Layer 2, the only addressing they support is MAC addresses, not IP.

365c01.qxd 11/28/01 4:33 PM Page 12

12

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

Physical

The physical layer is the layer of cabling, physical topology, and electrical, optical, and procedural standards. Cabling, transceivers, interfaces, hubs, and repeaters reside here. Some of the physical layer standards include the following:

V.35

EIA/TIA-449

EIA/TIA-232

HSSI

Note that topology is defined at both the physical and data link layers. When defined at the data link, it is referring to encapsulations such as Ethernet and

Token Ring. Originally, Ethernet was defined in an industry agreement to run over a bus topology. When topology is mentioned at the physical layer, it is referring to the media over which the data link encapsulation runs. Frames are broken down into binary at the physical later to be sent across the wire. Depending on the technology and media, signaling types will vary, but all have the goal of signaling in such a way that patterns of binary 1s and 0s are understood by the other side.

Objective 1.03

Ethernet and Other

Media Types

M edia types are defined as encapsulations at Layer 2 running over a specified physical topology. This can lead to confusion because “media types” sounds as if it indicates only the physical medium. It does define physical media, but usually the emphasis is on the data link layer encapsulation.

Exam Tip

Media types indicate Layer 2 encapsulations used to transfer over a physical medium.

Media is actually a plural for medium; however, at the data link layer, media types indicate topology.This is crucial because transparent switches and bridges don’t translate between media types. Either a translational switch/bridge or a router must be used to change encapsulations.Transparent switches/bridges only forward, filter, and flood; they don’t modify frames or translate between media types such as Ethernet and Token Ring.

365c01.qxd 11/28/01 4:33 PM Page 13

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

Ethernet is defined by the IEEE standard 802.3. Originally intended to run over a bus topology at 10 Mbps half duplex, the standard has been adapted over time to allow for 10, 100 or 1000 Mbps using half or full duplex and a physical star topology in a logical bus topology. Figure 1-3 illustrates the difference between physical star and bus topologies.

So engrained was the concept of a bus topology with Ethernet, the term “wire” was used to indicate the medium to which every station was connected. On this wire, the IEEE defined Ethernet as a CSMA/CD technology denoting that Ethernet is contention-based. CSMA/CD stands for Carrier Sense Multiple Access Collision

Detection. Because all stations shared the wire, there had to be a way of securing access to the medium. Ethernet implemented a listening mechanism so that stations could sense when the wire was available for transmission. The Ethernet station was required to sense the carrier—that is, it would listen to see if another station was transmitting on the wire. If there was traffic on the wire, the station would wait to send until it sensed that the transmission was over and the wire was free; then it would begin to send. Because the wire is made up of several stations

(known as multiple access) listening and sending when able, collisions may occur.

This made it mandatory for Ethernet to employ a collision detection mechanism.

This mechanism defined a notification called a jam signal and a back-off interval so that sending stations would wait at random intervals before sending again.

13

Full/Half Duplex

Half duplex refers to bidirectional sending on the same wire, but not at the same time—that is, one direction at a time. Similar to a one-way bridge, half duplex

Ethernet can transmit or receive, but not simultaneously. This means that half duplex is unable to make use of all wires simultaneously for data transfer. When a

FIGURE 1.3

Star and bus topologies

365c01.qxd 11/28/01 4:33 PM Page 14

14

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT station sends, it must also listen and detect collisions. Because the shared wire means there is the possibility of a collision, a sending station must always be listening to detect a collision.

There are certain devices that demand half duplex mode be used. The most common is the hub. An eight-port hub may have eight separate stations plugged into it, but as far as Ethernet is concerned, it is a single wire. The hub can do nothing to divide the logical wire into separate collision domains. Thus, when a hub is used, half duplex must be employed. Repeaters also force a station into using half duplex. Any time two or more stations are contending for the wire, half duplex must be used.

In recent years there has been a move from shared to dedicated bandwidth.

This change is referred to as microsegmentation. When stations share the wire, they can never make full use of it. Collisions inevitably waste bandwidth.

Microsegmentation reduces contention so that stations enjoy increased bandwidth. When contention is removed, collision detection can be turned off and full usage of the media can be achieved. The advent of the switch allows for full duplex; however, even with a switch, there are three requirements:

Point-to-point connection

NIC support

Full duplex turned on

Collision domains refer to shared media in which stations contend for access to the wire. When each station is placed on a separate logical wire, contention with other stations is removed and a separate collision domain is formed. In half duplex mode, the station must still contend for use of the wire with the switch, which may be trying to send data to the station. Reducing the size of collision domains in a network means reduced contention for the wire. Certain devices have the intelligence to segment the wire. Layer 1 devices such as hubs and repeaters cannot create new collision domains. Only devices that reside above Layer 1 can do this. The main two devices capable of segmenting collision domains are the switch and the bridge.

Obviously, higher layer devices can do so also; the focus here is that Layer 2 devices have the intelligence to create separate wires. In fact, they do it naturally—just plug the device in and it is on its own logical wire. Every port on a switch is its own collision domain, and this is one of the reasons why some switches can support full duplex. Bridges are older devices that never developed many of the capabilities that a switch has. Therefore, bridges usually operate only at half duplex.

Once a switch is in place, all that is needed is an interface supporting full duplex plugged directly into it. Realize that a hub may be plugged into a switch port, so that many stations might be sharing the wire. In this case, there is no point-to-point connection and therefore half duplex must be employed. All the stations off the single switch port reside in the same collision domain.

365c01.qxd 11/28/01 4:33 PM Page 15

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

10/100/1000 Mbps

Ethernet has been so developed that today, gigabit speeds are attainable, even using existing unshielded twisted-pair copper wiring. To do this, the signaling rate had to be increased. Fast Ethernet is 10 times faster than 10 Mbps Ethernet, therefore, the time slots had to be decreased by 10. This was not without a price. When using shared-media hubs, 100 Mbps operates poorly at best. Plug some stations into a hub and watch the collision light during data transfers. The light will be blinking constantly. Shrinking the time slots increases the chance of collisions on shared media. If 100 Mbps is chosen, use a switch. Hubs will only kill the improved performance that 100 Mbps was intended to have over 10 Mbps.

15

Objective 1.04

Network Devices and

Relation Within the

Networking Model

A s described earlier, hubs and repeaters are Layer 1 devices. These devices have little intelligence and are used to amplify signals and add stations to the logical wire. Bridges have greater intelligence than their Layer 1 counterpart, a repeater, but they were never developed fully, because of the popularity of the switch. The reason for this popularity is speed. The bridge used software and CPU in order to forward frames between devices. This method of forwarding can be found in low-end routers as well. The CPU is interrupted repeatedly to make simple forwarding decisions.

The switch is far superior to a bridge. The main difference that separates the switch from the bridge is the ASIC. With the advent of ASICs, logic is programmed into chips so that decisions can be offloaded from the CPU and performed in hardware. This is what gives the switch the ability to forward data at “wire speed.”

Even though the bridge is at the same layer of the OSI, it cannot forward frames nearly as fast as a switch. The demands of modern networks are such that only the switch can satisfy the need for increased speed and flexibility. This is why the switch is the most popular Layer 2 device in the world. With the increase of speed, the switch has evolved so that it is much more capable than a bridge. New capabilities include VLANs, trunking, and low-latency forwarding mechanisms. These technologies offer superior control and scalability to existing networks.

As can be seen in Table 1-2, different terminology should be used when referring to the device’s ports and what it is forwarding. The devices to focus on here are the repeater, hub, bridge, switch, and router. Higher layer switches are beyond the scope of this book.

365c01.qxd 11/28/01 4:33 PM Page 16

16

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

TABLE 1.2

Layer

Transport

Network

Data link

Physical

Terminology and Devices Used at the Lower

Four Layers of the OSI

Device

Layer 4 switch

Router,

Layer 3 switch

Bridge, switch

Repeater, hub

Connection Type

Sockets

Interface

Ports

Connectors

Terminology

Segments

Packets/Datagrams

Frames

Bits

Collisions are not the only problem on a network. Often, it is necessary for a device to send a message to all other devices on the network. Essentially, the term

“network” refers to the distance a broadcast travels; the network is known as a broadcast domain. The bridge/switch has the intelligence to segment collision domains, but cannot help with broadcasts. The bridge/switch floods broadcast and multicast frames out every port on the device.

Local Lingo

Broadcast, multicast, unicast A broadcast originates from a single host to every host in the network. A multicast originates from a single host to many hosts on the network. A unicast originates from a single host and is limited to a single other host.The following summarizes these terms: unicast, one to one; multicast, one to many; and broadcast, one to all.

The bridge/switch cannot filter broadcasts/multicasts because the frame is destined for all stations. Any time the destination is unknown, the bridge/switch must replicate the frame out to every port.

Routers reside at Layer 3 of the OSI and have more intelligence than a bridge/switch. The main capability they add over a switch is the ability to segment broadcast domains. A router divides networks, and a network is a broadcast domain, therefore the router prevents broadcasts from one network from passing to another network. This helps by saving bandwidth and CPU utilization that the broadcast/multicast would use on other segments.

365c01.qxd 11/28/01 4:33 PM Page 17

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

The main problem with broadcasts is not bandwidth (although this can be a problem), it is that they are sent out to the local network and every host on that network must process them. In addition to its own MAC address, each host must also pay attention to the broadcast address FFFF.FFFF.FFFF, meaning everybody on the network. Destination MAC addresses that do not match the local host are dropped by the NIC without interrupting the CPU. A broadcast address means that this frame is intended for all hosts, and even if the host doesn’t need the information contained in the broadcast, the CPU must be interrupted by the network interface and must process the frame. If there are many broadcasts crossing the network, the host’s performance can be degraded, as it will be processing broadcast packets rather than doing other things the user needs the device to do.

17

Exam Tip

Broadcasts and multicasts are not forwarded by a router. Bridges and switches must replicate these frames on every port.

In addition to limiting broadcast/multicast, a router is able to naturally change between media types. Recall that a bridge/switch is not able to do this. In order for media type translation, many things must be changed within the frame. The most basic is that the frame encapsulation must be changed, which the bridge/switch cannot do. Translational bridges can do this, but performance suffers. Routers naturally translate media types because frames are read and the Layer 2 encapsulation is discarded by the data link layer on its way up the router’s stack. When the router forwards the packet out an interface, it rebuilds a new frame to be sent out on that segment. Thus, media type translation is no problem—the frame is destroyed anyway and a new frame with the necessary encapsulation is created on the outgoing interface. This process is known as rewriting a frame.

What routers really do is connect broadcast domains to each other. A network address is defined for each domain and the router passes traffic from one broadcast domain to another. The IP address defines each individual network so that each device can be located and logical communication set up. Within each broadcast domain, the devices communicate physically with each other by MAC address. When a host needs to reach a host in a remote broadcast domain, it will send the message to a designated router and the router will be responsible for forwarding it to the destination. Many times the message will have to be forwarded over many networks to reach its ultimate destination. Within each of these hops, physical communication will have to be set up between devices. Whereas logical communication is end to end, physical communication is done by at least two

365c01.qxd 11/28/01 4:33 PM Page 18

18

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT devices (usually routers on the intermediate segments) on each routed network between the two end devices.

The penalty for the intelligence of the router is performance. The router is much slower than the typical switch. Most routers don’t have ASICs and must make forwarding decisions with software and CPU. This can lead to bottlenecks within an internetwork. For all the benefits and intelligence a router brings, it can still be the device that slows down the network. Because of this, placement of routers and design of the network are very important. It is also why it was stated earlier that routers should be kept out of the core of the network if possible.

Objective 1.05

Cabling the LAN and WAN

T o set up a network, basic knowledge of wiring must be understood. LAN cabling, while different from WAN cabling, has many common elements.

Both can be run over copper or fiber wiring and both make connections to repeaters, switches, and routers.

Connecting to the Cisco Device

Accessing a Cisco router/switch can be accomplished by direct cable connection through the console or auxiliary port located on the front of the device. The ports accept RJ-45 connectors and attach a computer’s serial port to the console, or a

DB25 is used to connect an external modem to the auxiliary port. Either connection requires a rollover cable. A rollover cable is so called because the pins are opposite each other from one end to the other as follows:

Pin 1-8

Pin 2-7

Pin 3-6

Pin 4-5

Pin 5-4

Pin 6-3

Pin 7-2

Pin 8-1

When connecting to the console, an application like Hyper Terminal (available with the Windows operating system) can be used. Simply use a rollover cable

365c01.qxd 11/28/01 4:33 PM Page 19

CHAPTER 1

Basic Concepts, Network Operation, and Cabling to connect the PC’s serial port to the router’s console port and set the following parameters under Properties for the serial connection.

9600 baud

8 bits

No parity

1 stop bit

No flow control

The console port doesn’t support Ready To Send (RTS) or Clear To Send

(CTS), which is used for flow control, so it should be turned off in the terminal program. By default, the console port operates at 9600 baud; it is not recommended that you change it.

Cisco provides cabling kits with the device to access the router properly. The router can also be accessed by Telnet.

19

Cabling the LAN

There are different cabling and signaling standards used for Ethernet. Table 1-3 displays these specifications.

Specifications such as 10BaseT should be understood in three parts. As can be seen following, the first section defines the speed (here 10 Mbps), and the second section refers to the signaling type, in this case baseband signaling. Finally, the last section is the cabling that is used. The T stands for twisted pair.

To connect end devices to their respective intermediate devices, different cabling is used depending on the topology. For Ethernet, the most common cabling is CAT 3 or CAT 5 unshielded twisted pair (UTP). When connecting devices, a straight-through cable is usually used. A straight-through cable is defined as a cable in which all the wires in the connector on one side match the wires in the connector on the other. Pin 1 on one side of the wire is connected to pin 1 on the other and so on. When connecting a PC to a switch or a hub, a straight-through connection is required. In terms of wiring, a router interface should be viewed like a host in that it requires a straight-through connection to the switch or hub. However, when connecting a switch to another switch or a host

365c01.qxd 11/28/01 4:33 PM Page 20

20

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

TABLE 1.3

Description

10Base2

10BaseT

100BaseTX

100BaseFX

Cabling Specifications

Topology

Bus

Star

Star

Point to point

Media

Coax

UTP

UTP

Fiber

Distance

185 meters

100 meters

100 meters

400 meters

Local Lingo

End device, intermediate device An end device is a host that is the ultimate destination for logical communication. An intermediate device allows remote end devices to be logically connected. In a Telnet communication between two remote hosts, the two end devices communicate by passing the information through the intermediate devices. Examples of end devices are most often computers, while intermediate devices are typically routers and switches.

However, in the case of Telnet and other applications, a router or a switch can be an end device. In this case, the devices used to connect these end hosts would be the intermediate devices.

to a host, a crossover cable is required. With a crossover cable, the TX and RX wires on one end are reversed on the other end. In particular for UTP (CAT 3 or

5), pins 1, 3 and 2, 6 are reversed on the opposite side as shown here.

Cabling the WAN

When cabling a WAN connection, the most important thing to know is DCE and

DTE. Typically, the router or PC is the DTE device, while the device that it plugs into for the WAN connectivity is DCE. DCE devices are important because they

365c01.qxd 11/28/01 4:33 PM Page 21

CHAPTER 1

Basic Concepts, Network Operation, and Cabling provide the clocking for the connection between the local device and the remote device. For frame relay, a CSU/DSU is used to supply the clock rate for the connection. With asynchronous communication, the modem does the clocking. In these examples, the CSU/DSU and the modem are considered DCE.

21

Local Lingo

DCE, DTE Data circuit-terminating equipment and data terminal equipment, respectively.

Recall that the physical layer of the OSI defined connectors and cabling. The following physical layer specifications are used to connect a router to a CSU/DSU:

V.35

EIA/TIA 232

EIA/TIA-449

X.21

EIA-530

The most common cable types for the CCNA material are the V.35 and

EIA/TIA 232. V.35 is the faster, typically supporting clock rates of up to 2 Mbps, while EIA/TIA 232 supports up to 115 Kbps.

CHECKPOINT

Objective 1.01: Internetworking Overview

Models help simplify and organize concepts in order to describe internetwork communication. The three-tier design model describes the core, distribution, and access areas within a network. The core is the backbone, the distribution implements policy, and the access is where the user interfaces with the network.

Objective 1.02: OSI Structure and Definitions

The seven-layer OSI describes the actual process that a PDU goes through as it moves up and down the model. Each layer adds its requirements to the frame and passes it up or down to the next layer, and each layer communicates with its corresponding layer in the remote device. The model is progressive, so that intelligence is the greatest at the top; however, with more complexity, speed is diminished.

Consequently, moving up usually means slower performance.

365c01.qxd 11/28/01 4:33 PM Page 22

22

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

Objective 1.03: Ethernet and Other Media Types

Ethernet is an encapsulation that works on several different media types, including UTP, coax, and fiber.

Ethernet is the most common LAN encapsulation, and though conceived originally as 10 Mbps half duplex, today it supports gigabit speeds and full duplex.

Objective 1.04:Network Devices and Relation Within the Networking Model

Devices that help move information throughout the internetwork have evolved over time, allowing the network to run faster and more efficiently. The invention of the switch contributed much of the increase in speed because of the

ASIC, which allowed forwarding decisions to be carried out with hardware.

This is the reason some switches can run at wire speed. However, there are things the switch was unable to help with which the router is needed for.

Switches cannot connect broadcast domains and move information between them. The router is the device that literally divides the internetwork into a collection of networks. It is the device used to set up logical communication by determining the path information takes to get to remote devices.

Objective 1.05: Cabling the LAN and WAN

Finally, though the LAN and

WAN have some similarities, there are many differences. Two important differences are the cables used to connect devices and the requirement for a device to supply clocking. The cabling for the direct connection to the router’s console is rolled. For connecting Ethernet devices, most often the straightthrough cable is used, unless it is host to host or switch to switch, in which case a crossover connection is necessary.

REVIEW QUESTIONS

1.

The LLC and MAC sublayers are part of what OSI layer?

A.

B.

C.

D.

E.

Network

Data link

Hardware

Physical

Presentation

2.

Which of the following is referred to as a best-effort protocol?

A.

B.

C.

D.

E.

IP

FTP

TCP

NCP

SPX

365c01.qxd 11/28/01 4:33 PM Page 23

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

3.

Which of the following descriptions best describes TCP windowing?

A.

B.

C.

D.

The period in which routes are exchanged

The number of seconds before the router disconnects

How much data a receiving station can accept before an acknowledgment must be sent

How large a TCP packet may be

4.

Which of the following flow control methods is used by the transport layer?

A.

B.

C.

D.

E.

Read ready

Gateway buffering

Sequencing

Delay

Source-quench

5.

What two layers of the OSI model support flow control and connection-oriented services?

A.

B.

C.

D.

E.

Presentation

Session

Transport

Network

Data link

23

6.

What types of addresses are seen at the data link layer? (Choose two.)

A.

B.

C.

D.

E.

DLCI

IP

IPX

MAC

E-mail

7.

A business has two buildings on the same campus. The computers within both buildings are on the same LAN; however, the network has been flooded with excessive broadcasts. What network device would you recommend to go between the two buildings?

A.

B.

C.

D.

E.

Switch

Bridge

Router

Hub

None of the above

365c01.qxd 11/28/01 4:34 PM Page 24

24

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT

8.

Topology is defined at which layer of the OSI model?

A.

B.

C.

D.

Physical

Data link

Network

Transport

9.

Which two devices are used to segment collision domains?

A.

B.

C.

D.

E.

Hub

Repeater

Switch

Router

Bridge

10.

A company has two networks and on each network resides 40 stations. Each station is connected to a hub that is connected to a router interface, and the network administrator says that the hub collision light is continually on.

What device would you replace the hub with to help this situation?

A.

B.

C.

D.

Bridge

Router

Repeater

Switch

REVIEW ANSWERS

1.

B

Layer 2 (data link) contains the LLC and MAC sublayers. The IEEE 802 committee subdivided the data link layer into two sublayers. The Logical Link

Control (LLC) provides connectionless and/or connection-oriented services for high-level protocols at the data link layer. The Media Access Control

(MAC) provides a unique address so that multiple devices can share the same medium and still identify each other.

2.

A

The Internet Protocol is referred to as a best-effort protocol because it simply receives data, applies the proper header information, and sends it along. IP, like most network layer protocols, is connectionless.

3.

C

A TCP window is the amount of outstanding data a sender can send on a particular connection before it gets an acknowledgment back from the receiver that it has received some data. For example, if a receiving host in a

365c01.qxd 11/28/01 4:34 PM Page 25

CHAPTER 1

Basic Concepts, Network Operation, and Cabling

TCP connection has specified a TCP window size of 64KB, the sender can only send 64KB of data and then it must stop and wait for an acknowledgment from the receiver that some or all of the data has been received. When the receiver acknowledges that all the data has been received, the sender is free to send another 64KB.

4.

E

The three flow controls mentioned at the transport layer are buffering, windowing, and source-quench messaging. Buffering stores the incoming segments for processing. Windowing is a dynamic method of flow control that changes based on receiver conditions. Source-quench messages are used to send a message to “back off ” when segments are being sent too fast for the receiver to handle.

5.

C

and

E

The transport and data link layers both support connection-oriented and connectionless protocols and flow control. The difference is that the transport layer supports these for segments being sent end to end, and the data link layer supports them for frames as they are sent on the local network.

The transport protocols deal with logical communication and the data link for physical communication.

6.

A

and

D

DLCI (data link connection identifiers) and MAC addresses both reside at the data link layer. MAC addresses are used in networks to establish communication between hosts. MAC addresses are physical, while

IPX and IP addresses are logical. Logical addresses allow one host to find another, but all communication must be done physically. DLCIs are used in frame relay and are similar functionally to a MAC address; they identify the data link address of the frame relay connection.

7.

C

Of the devices listed, only a router will help the situation. Switches and bridges segment collisions but not broadcasts. Hubs are Layer 1 devices with no intelligence—hubs pass everything.

8.

B

Topology is defined at the data link layer. When we say “topology,” we’re most often referring to logical topology, which is the access method used and the encapsulation type. Because of the emphasis on the encapsulation rather than the physical cabling, it is defined at Layer 2. If the question had said

“physical topology,” the physical layer would have been correct.

9.

C

and

E

While the router will also do this, it does much more. The switch and the bridge are Layer 2, which is the layer at which wire segmentation

25

365c01.qxd 11/28/01 4:34 PM Page 26

26

M I K E M E Y E R S ' C C N A C E RT I F I C AT I O N PA S S P O RT occurs. Therefore, they are the best answers. A hub and repeater cannot create collision domains, and all devices connected to them are on the same collision domain.

10.

D

The switch is the appropriate device here because the problem is collisions. While the bridge and router segment collision domains, the switch is more capable than the bridge and faster than the router. The 40 stations don’t even approach the maximum per network for TCP/IP or IPX, so a router is not necessary.

Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement