advertisement
DAS – Disk Array Storage Systems
Planning a DAS Installation
Fibre Channel Environments
ORDER REFERENCE
86 A1 94JX 04
Bull DAS – Disk Array Storage
Systems
Planning a DAS Installation
Fibre Channel Environments
Hardware
July 2000
BULL CEDOC
357 AVENUE PATTON
B.P.20845
49008 ANGERS CEDEX 01
FRANCE
ORDER REFERENCE
86 A1 94JX 04
The following copyright notice protects this book under the Copyright laws of the United States of America and other countries which prohibit such actions as, but not limited to, copying, distributing, modifying, and making derivative works.
Copyright Bull S.A. 1992, 2000
Printed in France
Suggestions and criticisms concerning the form, content, and presentation of this book are invited. A form is provided at the end of this book for this purpose.
To order additional copies of this book or other Bull Technical Publications, you are invited to use the Ordering Form also provided at the end of this book.
Trademarks and Acknowledgements
We acknowledge the right of proprietors of trademarks mentioned in this book.
AIX
R
is a registered trademark of International Business Machines Corporation, and is being used under licence.
UNIX is a registered trademark in the United States of America and other countries licensed exclusively through the Open Group.
Year 2000
The product documented in this manual is Year 2000 Ready.
The information in this document is subject to change without notice. Groupe Bull will not be liable for errors contained herein, or for incidental or consequential damages in connection with the use of this material.
Preface
This planning guide provides an overview of CLARiiON
â
Fibre Channel disk-array storage system models and offers essential background information and worksheets to help you with the installation and configuration planning.
Please read this guide
• if you are considering purchase of a CLARiiON
â
Fibre Channel disk-array storage system and want to understand its features; or
• before you plan the installation of a storage system.
Audience for the manual
You should be familiar with the host servers that will use the storage systems and with the operating systems of the servers. After reading this guide, you will be able to
• determine the best storage system components for your installation
• determine your site requirements
• configure storage systems correctly
Organization of the manual
Provides background information on the Fibre Channel protocols and explains the two major types of storage: shared and unshared.
Chapter 2 Describes the RAID groups and the different ways they store data.
Chapter 3 Describes the configurations for shared storage.
Chapter 4 Describes the configurations for unshared storage.
Chapter 5 Describes the hardware components for shared and unshared storage.
Chapter 6 Describes storage-system management utilities.
014-002912-01 v
Contents
Chapter 1 – About Fibre Channel storage systems and networks (SANs)
Chapter 2 – RAID types and tradeoffs
Chapter 3 – Planning file systems and LUNs with shared storage in a SAN
014-002912-01 vii
Contents
Chapter 4 – Planning LUNs and file systems with unshared storage
Chapter 5 – Storage-system hardware
Storage hardware — rackmount DPE-based storage systems ......................................... 5-2
viii
014-002912-01
Contents
Chapter 6 – Storage-system management
Managing shared or unshared storage systems using Navisphere Manager software ......... 6-2
Managing unshared storage systems using Navisphere Supervisor software ....................... 6-3
014-002912-01 ix
1
About Fibre Channel storage systems and networks (SANs)
CLARiiON
â
Fibre Channel disk-array storage systems provide terabytes of disk storage capacity, high transfer rates, flexible configurations, and highly available data at low cost.
A storage system package includes a host interface kit with hardware and software to connect with a server, storage management software, Fibre
Channel interconnect hardware, and one or more storage systems.
This chapter introduces Fibre Channel, its components, and shared and unshared storage systems. Major topics are
•
Fibre Channel background
About shared storage and SANs (storage area networks)
Fibre Channel is a high-performance serial protocol that allows transmission of both network and I/O channel data. It is a low level protocol, independent of data types, and supports such formats as SCSI and IP.
The Fibre Channel standard supports several physical topologies, including switched fabric point-to-point and arbitrated loop (FC-AL). The topologies used by the Fibre Channel storage systems described in this manual are switched fabric and FC-AL.
014-002912-01 -
1-1
1-2
A switched fabric is set of point-to-point connections between nodes, the connection being made through a Fibre Channel switch. Each node may have its own unique address, but the path between nodes is governed by the switch. The nodes are connected by optical cable.
A Fibre Channel arbitrated loop is a circuit consisting of nodes. Each node has a unique address, called a Fibre Channel arbitrated loop address. The nodes are connected by copper or optical cables. An optical cable can transmit data over great distances for connections that span entire enterprises and can support remote disaster recovery systems. Copper cable serves well for local connections; its length is limited to 30meters (99 feet).
Each connected device in a switched fabric or arbitrated loop is a server
(initiator) or a target (storage array). The switches and hubs are not considered nodes.
Nodes connected by switch - initiator and target
Server (initiator) Server (initiator)
Nodes
FC optical cable
Switch
Path 1
Path 2 Storage system (target) Storage system (target)
Nodes connected by hub - initiator and target
Server (initiator) Server (initiator)
Nodes
FC cable copper or optical
Hub
FC loop Storage system (target) Storage system (target)
The Fibre Channel standards include rules about cable types and lengths.
014-002912-01
Fibre Channel components
Fibre Channel components
A Fibre Channel storage system has three main components:
•
Server component (host interface kit with software)
•
Interconnect components (cables and Fibre Channel switches and hubs)
•
Storage components (storage systems and their hardware)
Server component (host interface kit with software)
The host interface kit includes a host-bus adapter (HBA) and support software. An HBA is a printed-circuit board that slides into an I/O slot in the server’s cabinet. It transfers data between server memory and one or more disk-array storage systems over Fibre Channel — as controlled by the support software (adapter driver).
Storage system installations
With hub
Direct
Server Server Server
Server
With switch
Server Server
Hub Swiitch
014-002912-01
Disk-array storage systems
Depending on your server type, you may have a choice of adapters. The adapter is designed for a specific host bus; for example, a PCI bus or SBUS.
Some adapter types support copper or optical cabling; some support copper cabling only.
Interconnect components
The interconnect components include the cables, Fibre Channel switch (for shared storage), and Fibre Channel hub (for unshared storage).
Cables
Depending on your needs, you can choose copper or optical cables.
The maximum length of copper cable is 30 meters (99 feet) between nodes or hubs. The maximum length of optical cable between server and hub or storage system is much greater, depending on the cable type. For example,
62.5 micron multimode cable can span up to 500 meters (1,640 feet) while 50
1-3
Fibre Channel components micron single-mode cable can span up to 10 kilometers (6.2 miles). This ability to span great distances is a major advantage of optical cable.
Some nodes have connections that require copper or optical cable. Other nodes allow for the conversion from copper to optical using a conversion device called a GigaBit Interface Converter (GBIC) or Media Interface
Adapter (MIA). In most cases, a GBIC or MIA lets you substitute long-distance optical connections for shorter copper connections. A GBIC, inserted in the back of the storage system SP, looks like this:
GBIC (GigaBit Interface Converter) seen from back of storage system
B
A
Optical cable
GBIC inserted in cable connector on SP
Details on cable lengths and rules appear later in this manual.
Fibre Channel switches
A Fibre Channel switch, which is a requirement for shared storage (a
Storage Area Network, SAN) connects all the nodes cabled to it using a fabric topology. A switch adds serviceability and scalability to any installation; it allows on-line insertion and removal of any device on the fabric and maintains integrity if any connected device stops participating. A switch also provides host-to-storage-system access control in a multiple-host shared storage environment. A switch has several advantages over a hub: it provides point-to-point connections (as opposed to a hub’s loop that includes all nodes) and it offers zoning to specify paths between nodes in the switch itself.
1-4
014-002912-01
Fibre Channel components
Server
Switch and hub toplogies compared
Switch topology
Server Server
Server
Hub topology (loop)
Server
Server
Switch uses discrete connections between ports
Hub uses loop between ports
SP SP SP
Storage systems
SP
Fibre Channel switches are available with 16 or 8 ports. They are compact units that fit in 2 U (3.5 inches) for the 16-port or 1 U (1.75 inches) for the
8-port. They are available to fit into a rackmount cabinet or as small deskside enclosures.
16-port switch, back view
Port with GBIC
If your servers and storage systems will be far apart, you can place the switches closer to the servers or the storage systems, as convenient.
Switch interconnect options
Optical cable can span great distances between nodes. The SPs for shared storage support optical cable. And all shared storage systems ship with devices called GBICs (GigaBit Interface Converters) for use with optical cable. Many types of host-bus adapter support optical cable, and can thus use GBICs to connect to the cables. For adapters that require copper cable, a device called a MIA (Media Interface Adapter) lets the adapter communicate with the copper cable.
014-002912-01
1-5
Fibre Channel components
Adapters with optical and copper interfaces, shared storage
Adapter with optical interface
Server
Adapter with copper interface
Server
Optical cable
Switch
Copper cable
Optical cable
Storage systems
SP SP
= GBIC
= MIA
A switch is technically a repeater, not a node, in a Fibre Channel loop.
However, it is bound by the same cabling distance rules as a node.
Fibre Channel hubs
A hub connects all the nodes cabled to it into a single logical loop. A hub adds serviceability and scalability to any loop; it allows on-line insertion and removal of any device on the loop and maintains loop integrity if any connected device stops participating.
Fibre channel hubs are compact units that fit in 1 U (1.75 inches) of storage space. They are available to fit into a rackmount cabinet or as small deskside units.
Nine-port hub
1-6
The nine-pin port can connect to a server, storage system, or another hub.
If your servers and storage systems will be far apart, you can place the hubs closer to the servers or the storage systems, as convenient. You can connect one hub to another, at great distances using optical cable, for disaster recovery systems.
Hub interconnect options
Some host-bus adapter types support copper cabling only; some support either optical or copper cabling. Hubs and disk-array storage systems can connect via copper cables or optical cables using MIAs.
014-002912-01
About shared storage and SANs (storage area networks)
Adapters with copper and optical interfaces, unshared storage
Server Server
Server Server
Copper cable
Optical cable
Hub
Hub
Storage systems
SP SP SP SP SP SP SP SP
A hub is technically a repeater, not a node, in a Fibre Channel loop.
However, it is bound by the same cabling distance rules as a node.
= MIA
About shared storage and SANs (storage area networks)
This section explains the features that let multiple servers share disk-array storage systems on a SAN (storage area network).
A SAN is a collection of storage devices connected to servers via Fibre
Channel switches to provide a central location for disk storage. Centralizing disk storage among multiple servers has many advantages, including
• highly available data
• flexible association between servers and storage capacity
• centralized management for fast, effective response to users’ data storage needs
• easier file backup and recovery
A SAN includes two or more servers and has Fibre Channel switches between servers and storage systems.
014-002912-01
1-7
About shared storage and SANs (storage area networks)
Components of a SAN
Server Server Server
Fibre Channel switch
SP A SP B
Fibre Channel switch
SP A SP B
Storage systems
Fibre Channel switches can control data access to storage systems through the use of switch zoning. With zoning, an administrator can specify groups
(called zones) of Fibre Channel devices (such as host-bus adapters, specified by worldwide name), and SPs between which the switch will allow communication.
However, switch zoning cannot selectively control data access to LUNs in a storage system, because each SP appears as a single Fibre Channel device to the switch. So switch zoning can prevent or allow communication with an
SP, but not with specific disks or LUNs attached to an SP. For access control with LUNs, a different solution is required: Storage Groups.
Storage Groups
A Storage Group is one or more LUNs (logical units) within a disk-array storage system that is reserved for one or more servers and is inaccessible to other servers.
When you configure the storage system for a SAN, you specify servers and the Storage Group(s) each server can read from and/or write to. The
Licensed Internal Code firmware running in each storage system enforces the server-to-Storage Group permissions.
The following figure shows a simple SAN configuration consisting of one
CLARiiON storage system with two Storage Groups. One Storage Group serves a cluster of two servers running the same operating system and the other Storage Group serves a UNIX database server. Each server is configured with two independent paths to its data, including separate host-bus adapters, switches, and SPs, so there is no single point of failure for access to its data.
1-8
014-002912-01
Sample SAN configuration
Highly available cluster
File Server
Operating system A
Mail Server
Operating system A
About shared storage and SANs (storage area networks)
Database Server
Operating system B
014-002912-01
Cluster
Fibre Channel switch
Storage Group
Database Server
Storage Group
Fibre Channel switch
SP A
LUN
LUN
LUN
LUN
LUN
LUN
LUN
SP B
Physical storage system with up to
120 disks
Access Control in a SAN
Access control permits or restricts a server’s access to SAN storage. There are two kinds of access control:
• Data access control
• Configuration access control
Data access control — Storage Groups provide data access control.
During storage system configuration, using a management utility, the system administrator associates a server with one or more LUNs. The associated LUNs constitute a Storage Group. The server HBA port’s unique worldwide name identifies the port and allows access to the Storage Group.
The storage-system SP will prevent any server from accessing LUNs the server is not configured to access.
Each server sees its Storage Group as if it were an entire storage system and never sees the other LUNs on the storage system. Therefore, it cannot access or modify data on LUNs that are not part of its Storage Group.
Configuration access control — Configuration access control lets you restrict the servers through which a user can send configuration commands to an attached storage system. Configuration access is governed by a password. The administrator can set the password during storage system setup; or if the site doesn’t want to use configuration access control, it can skip the password.
The following figure shows both data access control (Storage Groups) and configuration access control. Each server has exclusive read and write access to its designated Storage Group. Of the four servers connected to the
1-9
About shared storage and SANs (storage area networks)
SAN, only the Admin server can send configuration commands to the storage system.
Data and configuration access control in a SAN
Highly available cluster
Admin server
Operating system A
Inventory server E-mail server
Operating system A
Operating system B
Web server
Operating system B
01
02 03 04 05 06 07 08
Fibre Channel switch
Admin Storage Group
Dedicated
Data access by adapters 01, 02
Inventory Storage Group
Dedicated
Data access by adapters 03, 04
E-mail and Web server
Storage Group
Shared
Data access by
adapters 05, 06, 07, 08
SP A
LUN
LUN
LUN
LUN
LUN
LUN
LUN
LUN
LUN
LUN
SP B
Fibre Channel switch
Configuation access, by adapters 01and 02
(Admin server only)
Storage system hardware for shared storage
For shared storage, you need a Disk-array Processor Enclosure (DPE) storage system.
A DPE is a 10-slot enclosure with hardware RAID features provided by one or two storage processors (SPs). For shared storage, two SPs are required.
In addition to its own disks, a DPE can support up to 11 ten-slot Disk Array
Enclosures (DAEs) for a total of 120 disks.
1-10
014-002912-01
About unshared storage
Disk-array processor enclosure (DPE) with two DAEs, for shared storage
About unshared storage
Unshared storage systems are less costly and less complex than shared storage systems. They offer many shared storage system features; for example, you can use multiple unshared storage systems with multiple servers. However, with multiple servers, unshared storage offers less flexibility and security than shared storage, since any user with write access to privileged server files can enable access to any storage system.
Storage system hardware for unshared storage
For unshared storage, there are four types of storage system, each using the
FC-AL protocol. Each type is available in a rackmount or deskside (office) version.
• Disk-array Processor Enclosure (DPE) storage systems. A DPE is a
10-slot enclosure with hardware RAID features provided by one or two storage processors (SPs). In addition to its own disks, a DPE can support up to 110 additional disks in 10-slot Disk Array Enclosures (DAEs) for a total of 120 disks. This is the same type of storage system used for shared storage, but it has a different SP and different Licensed Internal
Code (LIC).
• Intelligent Disk Array Enclosure (iDAE). An iDAE, like a DPE, has SPs and thus all the features of a DPE, but is thinner and has a limit of 30 disks.
• Disk Array Enclosure (DAE). A DAE does not have SPs. A DAE can connect to a DPE or an iDAE, or you can use it without SPs. A DAE used without an SP does not inherently include RAID, but can operate as a
RAID device using software running on the server system. Such a DAE is also known as Just a Box of Disks, or JBOD.
• 30-slot SCSI-disk storage systems. Like the DPE, these offer RAID features provided by one or two SPs. However, they use SCSI, not Fibre
Channel, disks. Each has space for 30 disks.
014-002912-01
1-11
About unshared storage
Disk-array processor enclosure (DPE)
Deskside DPE with DAE
Rackmount DPE, one enclosure, supports up to 11 DAEs
30-slot deskside
Intelligent disk-array enclosure (iDAE)
10-slot deskside Rackmount
30-slot SCSI-disk storage system
Deskside Rackmount
1-12
014-002912-01
What next?
What next?
For information about RAID types and RAID tradeoffs, continue to the next chapter. To plan LUNs and file systems for shared storage, skip to
Chapter 3; or for unshared storage, Chapter 4. For details on the
storage-system hardware -- shared and unshared -- skip to Chapter 5.
For storage-system management utilities, skip to Chapter 6.
014-002912-01
1-13
2
RAID types and tradeoffs
This chapter explains RAID types you can choose for your storage system
LUNs. If you already know about RAID types and know which ones you want, you can skip this background information and skip to the planning
chapter (Chapter 3 for shared storage; Chapter 4 for unshared storage).
Topics are as follows:
• Introducing RAID
• RAID 5 group (individual access array)
• RAID 3 group (parallel access array)
• RAID 1 mirrored pair
• RAID 0 group (nonredundant array)
• RAID 1/0 group (mirrored RAID 0 array)
• Individual disk
• Hot spare
• RAID benefits and tradeoffs
• Guidelines for RAID groups
• Sample applications for RAID groups
IMPORTANT This chapter applies primarily to storage systems with storage processors (SPs). For a storage system without SPs (a
DAE-only system), RAID types are limited by the RAID software you run on the server. The RAID terms and definitions used here conform to generally accepted standards.
Introducing RAID
The storage system uses RAID (redundant array of independent disks) technology. RAID technology groups separate disks into one logical storage unit (LUN) to improve reliability and/or performance.
The storage system supports five RAID levels and two other disk configurations, the individual unit and the hot spare (global spare). You group the disks into one RAID group by binding them using a storage-system management utility.
Four of the RAID groups use disk striping and two use mirroring .
2-1
014-002912-01
Introducing RAID
Disk striping
Using disk stripes, the storage-system hardware can read from and write to multiple disks simultaneously and independently. By allowing several read/write heads to work on the same task at once, disk striping can enhance performance. The amount of information read from or written to each disk makes up the stripe element size. The stripe size is the stripe element size multiplied by the number of disks in a group. For example, assume a stripe element size of 128 sectors (the default) and a five-disk group. The group has five disks, so you would multiply five by the stripe element size of 128 to yield a stripe size of 640 sectors.
The storage system uses disk striping with most RAID types.
Mirroring
Mirroring maintains a second (and optionally through software, a third) copy of a logical disk image that provides continuous access if the original image becomes inaccessible. The system and user applications continue running on the good image without interruption. There are two kinds of mirroring: hardware mirroring, in which the SP synchronizes the disk images; and software mirroring, in which the operating system synchronizes the images.
With a storage system, you can create a hardware mirror by binding disks as a RAID 1 mirrored pair or a RAID 1/0 group (a mirrored RAID 0 group); the hardware will then mirror the disks automatically. Or you can use software mirroring with RAID 0 groups or individual units that have no inherent data redundancy. With software mirroring, the operating system mirrors the images. Some operating systems support software mirroring; others do not.
RAID groups and LUNs
On full-fibre systems, some RAID types let you create multiple LUNs on one
RAID group. You can then allot each LUN to a different user, server, or application. For example, a five-disk RAID 5 group that uses 36-Gbyte disks offers 144 Gbytes of space. You could bind three LUNs, say with 24, 60, and
60 Gbytes of storage capacity, for temporary, mail, and customer files.
One disadvantage of multiple LUNs on a RAID group is that I/O to each
LUN may affect I/O to the others in the group; that is, if traffic to one LUN is very heavy, I/O performance with other LUNs may degrade. The main advantage of multiple LUNs per RAID group is the ability to divide the enormous amount of disk space provided by RAID groups on newer, high-capacity disks.
2-2
014-002912-01
RAID 5 group (individual access array)
RAID Group
Disk Disk
Key
LUN 0 - temp
LUN 1 - mail
LUN 2 - customers
Disk Disk Disk
RAID types
You can choose from the following RAID types: RAID 5, RAID 3, RAID 1,
RAID 0, RAID 1/0, individual disk unit, and hot spare.
RAID 5 group (individual access array)
A RAID 5 group usually consists of five disks (but can have three to sixteen).
A RAID 5 group uses disk striping. With a RAID 5 group on a full-fibre storage system, you can create up to 32 RAID 5 LUNs to apportion disk space to different users, servers, and applications.
The storage system writes parity information that lets the group continue operating if a disk fails. When you replace the failed disk, the SP rebuilds the group using the information stored on the working disks. Performance is degraded while the SP rebuilds the group. However, the storage system continues to function and gives users access to all data, including data stored on the failed disk.
The following figure shows user and parity data with the default stripe element size of 128 sectors (65,536 bytes) in a five-disk RAID 5 group. The stripe size comprises all stripe elements. Notice that the disk block addresses in the stripe proceed sequentially from the first disk to the second, third, and fourth, then back to the first, and so on.
014-002912-01
2-3
RAID 3 group (parallel access array)
Stripe element size
Stripe
Blocks
0-127
RAID 5 group
First disk
512-639
1024-1151
1536-1663 Parity
. . .
128-255
Second disk
640-767 1152-1279
Parity 2048-2175
Stripe size
256-383
768-895
Third disk
Parity
1664-1791
2176-2303
. . .
384-511
Parity
Fourth disk
1280-1407
1792-1919
2304-2431
. . .
Parity
Fifth disk
896-1023
1408-1535
1920-2047
2432-2559
. . .
User data
Parity data
The storage system performs more steps when writing data to a RAID 5 group than to any other type of group. For each write, the storage system
1.
Reads user data from the sectors and parity data for the sectors.
2.
Recalculates the parity data.
3.
Writes the new user and parity data.
RAID 5 groups benefit greatly from storage-system caching, particularly write caching.
RAID 3 group (parallel access array)
A RAID 3 group consists of five or more disk disks. The hardware always reads from or writes to all the disks. A RAID 3 group uses disk striping .
To maintain the RAID 3 performance, you can create only one LUN per RAID 3 group.
The storage system writes parity information that lets the group continue operating if a disk fails. When you replace the failed disk, the SP rebuilds the group using the information stored on the working disks. Performance is degraded while the SP rebuilds the group. However, the storage system continues to function and gives users access to all data, including data stored on the failed disk.
The following figure shows user and parity data with a data block size of
2 Kbytes in a RAID 3 group. Notice that the byte addresses proceed from the first disk to the second, third, and fourth, then the first, and so on.
2-4
014-002912-01
RAID 1 mirrored pair
Bytes
0-511
RAID 3 group
Data block
First disk
2048-2559
4096-4607
6144-6655
8192-8603
. . .
Stripe element size
512-1023 2560-3071
4608-5119
6656-7167
8604-9115
Stripe size
Third disk
1024-1535 3072-3583
5120-5631
7168-7679
9116-9627
. . .
Fourth disk
1536-2047 3584-4095
5632-6143
7680-8191
9628-10139
. . .
User data
Parity data
Parity Parity
Fifth disk
Parity Parity Parity
. . .
RAID 3 differs from RAID 5 in several important ways. First, in a RAID 3 group the hardware processes disk requests serially; whereas in a RAID 5 group the hardware can interleave disk requests. Second, with a RAID 3 group, the parity information is stored on one disk; with a RAID 5 group, it is stored on all disks. Finally, with a RAID 3 group, the I/O occurs in small units (one sector) to each disk. A RAID 3 group works well for single-task applications that use I/Os of blocks larger than 64 Kbytes.
Each RAID 3 group requires some dedicated SP memory (6 Mbytes recommended per group). This memory is allocated when you create the group and becomes unavailable for storage-system caching. For top performance, we suggest that you do not use RAID 3 groups with RAID 5,
RAID 1/0, or RAID 0 groups, since SP resources (including memory) are best devoted to the RAID 3 groups. RAID 1 mirrored pairs and individual units require less SP attention and therefore work well with RAID 3 groups.
For each write to a RAID 3 group, the storage system
1.
Calculates the parity data.
2.
Writes the new user and parity data.
RAID 1 mirrored pair
A RAID 1 group consists of two disks that are mirrored automatically by the storage-system hardware.
RAID 1 hardware mirroring within the storage system is not the same as software mirroring or hardware mirroring for other kinds of disks.
014-002912-01
2-5
RAID 0 group (nonredundant array)
Functionally, the difference is that you cannot manually stop mirroring on a
RAID 1 mirrored pair, and then access one of the images independently. If you want to use one of the disks in such a mirror separately, you must unbind the mirror (losing all data on it), rebind the disk in as the type you want, and software format the newly bound LUN.
With a storage system, RAID 1 hardware mirroring has the following advantages:
• automatic operation (you do not have to issue commands to initiate it)
• physical duplication of images
• a rebuild period that you can select during which the SP recreates the second image after a failure.
With a RAID 1 mirrored pair, the storage system writes the same data to both disks, as follows.
0
0
1
RAID 1 mirrored pair
First disk
2 3 4
. . .
1
Second disk
2 3 4
. . .
= User data
RAID 0 group (nonredundant array)
A RAID 0 group consists of three to a maximum of sixteen disks. A RAID 0 group uses disk striping , in which the hardware writes to or reads from multiple disks simultaneously. In a full-fibre storage system, you can create up to 32 LUNs per RAID group.
Unlike the other RAID levels, with RAID 0 the hardware does not maintain parity information on any disk; this type of group has no inherent data redundancy. RAID 0 offers enhanced performance through simultaneous I/O to different disks.
If the operating system supports software mirroring, you can use software mirroring with the RAID 0 group to provide high availability. A desirable alternative to RAID 0 is RAID 1/0.
RAID 1/0 group (mirrored RAID 0 group)
A RAID 1/0 group consists of four, six, eight, ten, twelve, fourteen, or sixteen disks. These disks make up two mirror images, with each image including two to five disks. The hardware automatically mirrors the disks. A RAID 1/0 group uses disk striping .
It combines the speed advantage of RAID 0 with the redundancy advantage of mirroring. With a RAID 1/0 group on a
2-6
014-002912-01
Individual disk unit full-fibre storage system, you can create up to 32 RAID 5 LUNs to apportion disk space to different users, servers, and applications.
The following figure shows the distribution of user data with the default stripe element size of 128 sectors (65,536 bytes) in a six-disk RAID 1/0 group. Notice that the disk block addresses in the stripe proceed sequentially from the first mirrored disks (first and fourth disks) to the second mirrored disks (second and fifth disks), to the third mirrored disks
(third and sixth disks), and then from the first mirrored disks, and so on.
Stripe element size
Stripe size
Stripe
Blocks
0-127
RAID 1/0 group
First disk of primary image
384-511 768-895 1152-1279 1536-1663
. . .
128-255
Second disk of primary image
512-639 896-1023 1280-1407
1664-1791
. . .
256-383
Third disk of primary image
640-767
1024-1151
1408-1535
1792-1919
. . .
0-127
First disk of secondary image
384-511
768-8951152-1279 1536-1663
128-255
Second disk of secondary image
512-639 896-1023 1280-1407 1664-1791
. . .
256-383
Third disk of secondary image
640-767
1024-1151
1408-1535
1792-1919
. . .
User data
A RAID 1/0 group can survive the failure of multiple disks, providing that one disk in each image pair survives.
Individual disk unit
An individual disk unit is a disk bound to be independent of any other disk in the cabinet. An individual unit has no inherent high availability, but you can make it highly available by using software mirroring with another individual unit. You can create one LUN per individual disk unit. If you want to apportion the disk space, you can do so using partitions, file systems, or user directories.
014-002912-01
2-7
Hot spare
Hot spare
A hot spare is a dedicated replacement disk on which users cannot store information. A hot spare is global: if any disk in a RAID 5 group, RAID 3 group, RAID 1 mirrored pair, or RAID 1/0 group fails, the SP automatically rebuilds the failed disk’s structure on the hot spare. When the SP finishes rebuilding, the disk group functions as usual, using the hot spare instead of the failed disk. When you replace the failed disk, the SP copies the data from the former hot spare onto the replacement disk.
When the copy is done, the disk group consists of disks in the original slots, and the SP automatically frees the hot spare to serve as a hot spare again. A hot spare is most useful when you need the highest data availability. It eliminates the time and effort needed for someone to notice that a disk has failed, find a suitable replacement disk, and insert the disk.
IMPORTANT When you plan to use a hot spare, make sure the disk has the capacity to serve in any RAID group in the storage-system chassis. A RAID group cannot use a hot spare that is smaller than a failed disk in the group.
You can have one or more hot spares per storage-system chassis. You can make any disk in the chassis a hot spare, except for a disk that serves for
Licensed Internal Code storage or the write cache vault. That is, a hot spare can be any of the following disks:
DPE or iDAE system without write caching:
DPE system with write caching: iDAE system with write caching:
30-slot SCSI-disk system: disks 3-119 disks 9-119 disks 5-29 disks A1-E1, A2-E2,
B3-E3, A4-E4
An example of hot spare usage for a deskside DPE storage system follows.
2-8
014-002912-01
RAID benefits and tradeoffs
6
7
4
5
8
9
2
3
0
1
14
15
16
17
18
19
10
11
12
13
Hot spare
1. RAID 5 group consists of disk modules 0-4; RAID 1 mirrored pair is modules 5 and 6;
hot spare is module 9.
2. Disk module 3 fails.
3. RAID 5 group becomes modules 0, 1, 2, 9, and 4; now no hot spare is available.
4. System operator replaces failed module 3 with a functional module.
5. RAID 5 group once again is 0-4 and hot spare is 9.
RAID benefits and tradeoffs
This section reviews RAID types and explains their benefits and tradeoffs.
You can create seven types of LUN:
• RAID 5 group (individual access array)
• RAID 3 group (parallel access array)
• RAID 1 mirrored pair
• RAID 1/0 group (mirrored RAID 0 group); a RAID 0 group mirrored by the storage-system hardware
• RAID 0 group (nonredundant individual access array); no inherent high-availability features, but can be software mirrored if the operating system supports mirroring
• Individual unit; no inherent high-availability features but can be software mirrored, if the operating system supports mirroring
• Hot spare; serves only as an automatic replacement for any disk in a
RAID type other than 0; does not store data during normal system operations
014-002912-01
2-9
RAID benefits and tradeoffs
IMPORTANT Plan the disk unit configurations carefully. After a disk has been bound into a LUN, you cannot change the RAID type of that
LUN without unbinding it, and this means losing all data on it.
The following table compares the read and write performance, tolerance for disk failure, and relative cost per megabyte (Mbyte) of the RAID types and their software-mirrored counterparts. Figures shown are theoretical maximums.
Relative performance, availability, and cost of disk RAID types (individual unit = 1.0)
Disk configuration
RAID 5 group with fivedisks
Relative read performance without cache
Relative write performance without cache
Up to 5 with five disks (for small I/O requests, 2 to
8 Kbytes)
Up to 1.25 with five disks (for small I/O requests, 2 to 8 Kbytes)
Up to 4 (for large I/O requests) Up to 4 (for large I/O requests)
Relative cost per
Mbyte
1.25
RAID 3 group with fivedisks
RAID 1 mirrored pair
RAID 1/0 group with
10 disks
Individual unit
Up to 2
Up to 10
Up to 1
Up to 5
1.25
2
Not software mirrored 1
1 1
Software mirrored Up to 2
Up to 1 2
Notes:These performance numbers are not based on storage-system caching. With caching, the performance numbers for RAID 5 writes improve significantly.
Performance multipliers vary with load on server and storage system.
Performance
RAID 5, with individual access, provides high read throughput for small requests (blocks of 2 to 8 Kbytes) by allowing simultaneous reads from each disk in the group. RAID 5 write throughput is limited by the need to perform four I/Os per request (I/Os to read and write data and parity information). However, write caching improves RAID 5 write performance.
RAID 3, with parallel access, provides high throughput for sequential, large block-size requests (blocks of more than 64 Kbytes). With RAID 3, the system accesses all five disks in each request but need not read data and parity before writing – advantageous for large requests but not for small ones. RAID 3 employs SP memory without caching, which means you do not need the second SP and BBU that caching requires.
Generally, the performance of a RAID 3 group increases as the size of the
I/O request increases. Read performance increases rapidly with read requests up to 1Mbyte. Write performance increases greatly for sequential write requests that are greater than 256 Kbytes. For applications issuing
2-10
014-002912-01
014-002912-01
RAID benefits and tradeoffs very large I/O requests, a RAID 3 LUN provides significantly better write performance than a RAID 5 LUN.
We do not recommend using RAID 3 in the same storage-system chassis with RAID 5 or RAID 1/0.
A RAID 1 mirrored pair has its disks locked in synchronization, but the SP can read data from the disk whose read/write heads are closer to it.
Therefore, RAID 1 read performance can be twice that of an individual disk while write performance remains the same as that of an individual disk.
A RAID 0 group (nonredundant individual access array) or RAID 1/0 group
(mirrored RAID 0 group) can have as many I/O operations occurring simultaneously as there are disks in the group. Since RAID 1/0 locks pairs of RAID 0 disks the same way as RAID 1 does, the performance of RAID 1/0 equals the number of disk pairs times the RAID 1 performance number. If you want high throughput for a specific LUN, use a RAID 1/0 or RAID 0 group. A RAID 1/0 group requires at least six disks; a RAID 0 group, at least three disks.
An individual unit needs only one I/O operation per read or write operation.
RAID types 5, 1, 1/0, and 0 allow multiple LUNs per RAID group. If you create multiple LUNs on a RAID group, the LUNs share the RAID group disks, and the I/O demands of each LUN affect the I/O service time to the other LUNs. For best performance, you may want to use one LUN per RAID group.
Storage flexibility
Certain RAID group types — RAID 5, RAID 1, RAID 1/0, and RAID 0 — let you create up to 32 LUNs in each group. This adds flexibility, particularly with large disks, since it lets you apportion LUNs of various sizes to different servers, applications, and users. Conversely, with RAID 3, there can be only one LUN per RAID group, and the group must include five or nine disks — a sizable block of storage to devote to one server, application, or user. However, the nature of RAID 3 makes it ideal for that single-threaded type of application.
Data availability and disk space usage
If data availability is critical and you cannot afford to wait hours to replace a disk, rebind it, make it accessible to the operating system, and load its information from backup, then use a redundant RAID group: RAID 5,
RAID 3, RAID 1 mirrored pair, or RAID 1/0. Or bind a RAID 0 group or individual disk unit that you will later mirror with software mirroring. If data availability is not critical, or disk space usage is critical, bind an individual unit or RAID 0 group without software mirroring.
A RAID 1 mirrored pair or RAID 1/0 group provides very high data availability. They are more expensive than RAID 5 or RAID 3 groups, since only 50 percent of the total disk capacity is available for user data, as shown
2-11
RAID benefits and tradeoffs
A RAID 5 or RAID 3 group provides high data availability, but requires more disks than a mirrored pair. In a RAID 5 or RAID 3 group of five disks,
80 percent of the disk space is available for user data. So RAID 5 and
RAID 3 groups use disk space much more efficiently than a mirrored pair. A
RAID 5 or RAID 3 group is usually more suitable than a RAID 1 mirrored pair for applications where high data availability, good performance, and efficient disk space usage are all of relatively equal importance.
Disk space usage in the RAID configurations
2-12
RAID 5 group
1st disk user and parity data
2nd disk user and parity data
3rd disk user and parity data
4th disk user and parity data
5th disk user and parity data
RAID 3 group
1st disk user data
2nd disk user data
3rd disk user data
4th disk user data
5th disk parity data
80% user data
20% parity data
100% user data
RAID 0 group
(nonredundant array)
1st disk user data
2nd disk user data
3rd disk user data
RAID 1/0 group
Disk mirror (RAID 1 mirrored
pair or software mirror)
1st disk user data
2nd disk redundant user data
50% user data
50% redundant data
Individual disk unit
50% user data
50% redundant data
1st disk user data
2nd disk user data
3rd disk user data
4th disk redundant user data
5th disk redundant user data
6th disk redundant user data User data
100% user data
Hot spare
Reserved No user data
If your operating system supports software mirroring, you can use software mirroring for an individual unit or RAID 0 group. For a comparison of
software and hardware mirroring, see the section on mirroring, page 2-2.
A RAID 0 group (nonredundant individual access array) provides all its disk space for user files, but does not provide any high availability features. For high availability, you can use a RAID 1/0 group or use software mirroring with the RAID 0 group. A RAID 1/0 group or software-mirrored RAID 0
014-002912-01
Guidelines for RAID groups group provides the best combination of performance and availability, at the highest cost per Mbyte of disk space. A RAID 1/0 group offers higher reliability than a software-mirrored RAID 0 group.
An individual unit, like a RAID 0 group, provides no high-availability features. All its disk space is available for user data, as shown in the figure
on page 2-12. For high availability with an individual disk, you can use
software mirroring.
Guidelines for RAID groups
To decide when to use a RAID 5 group, RAID 3 group, mirror (that is, a
RAID 1 mirrored pair, RAID 1/0 group, or software mirroring), a RAID 0 group, individual disk unit, or hot spare, you need to weigh these factors:
•
Importance of data availability
•
Importance of performance
•
Amount of data stored
•
Cost of disk space
The following guidelines will help you decide on RAID types.
Use a RAID 5 group (individual access array) for applications where
•
Data availability is very important
•
Large volumes of data will be stored
•
Multitask applications use I/O transfers of different sizes
•
Good read and moderate write performance are important (write caching can improve (RAID 5 write performance)
•
You want the flexibility of multiple LUNs per RAID group.
Use a RAID 3 group (parallel access array) for applications where
• Data availability is very important
• Large volumes of data will be stored
• A single-task application uses large I/O transfers (more than 64 Kbytes).
The operating system must allow transfers aligned to start at disk addresses that are multiples of 2 Kbytes from the start of the LUN.
Use a RAID 1 mirrored pair for applications where
•
Data availability is very important
•
Speed of write access is important and write activity is heavy
014-002912-01
2-13
Sample applications for RAID types
Use a RAID 1/0 group (mirrored nonredundant array) for applications where
• Data availability is critically important
• Overall performance is very important
Use a RAID 0 group (nonredundant individual access array) for applications where
•
High availability is not important (or you plan to use software mirroring by the operating system)
•
Overall performance is very important
Use an individual unit for applications where
•
High availability is not important (or you plan to use software mirroring by the operating system)
•
Speed of write access is somewhat important
Use a hot spare where
•
In any RAID 5, RAID 3, RAID 1/0 or RAID 1 group, high availability is so important that you want to regain data redundancy quickly without human intervention if any disk in the group fails
•
Minimizing the degraded performance caused by disk failure in a
RAID 5 or RAID 3 group is important
Sample applications for RAID types
This section describes some types of applications in which you would want to use a RAID 5 group, RAID 3 group, RAID 1 mirrored pair, RAID 0 group
(nonredundant array), RAID 1/0 group, or individual unit.
RAID 5 group (individual access array) — Useful as a database repository or a database server that uses a normal or low percentage of write operations (writes are 33 percent or less of all I/O operations). Use a
RAID 5 group where multitask applications perform I/O transfers of different sizes. Write caching can significantly enhance the write performance of a RAID 5 group.
For example, a RAID 5 group is suitable for multitasking applications that require a large history database with a high read rate, such as a database of legal cases, medical records, or census information. A RAID 5 group also works well with transaction processing applications, such as an airline reservations system, where users typically read the information about several available flights before making a reservation, which requires a write operation. You could also use a RAID 5 group in a retail environment, such as a supermarket, to hold the price information accessed by the point-of-sale terminals. Even though the price information may be updated daily, requiring many write operations, it is read many more times during the day.
2-14
014-002912-01
014-002912-01
Sample applications for RAID types
RAID 3 group — A RAID 3 group (parallel access array) works well with a single-task application that uses large I/O transfers (more than 64 Kbytes), aligned to start at a disk address that is a multiple of 2 Kbytes from the beginning of the logical disk. RAID 3 groups can use SP memory to great advantage without the second SP and battery backup unit required for storage-system caching.
You might use a RAID 3 group for a single-task application that does large
I/O transfers, like a weather tracking system, geologic charting application, medical imaging system, or video storage application.
RAID 1 mirrored pair or individual unit to be software mirrored —
A RAID 1 mirrored pair or individual unit to be software mirrored is useful for logging or record-keeping applications because it requires fewer disks than a RAID 0 group (nonredundant array) and provides high availability and fast write access. Or you could use it to store daily updates to a database that resides on a RAID 5 group, and then, during off-peak hours, copy the updates to the database on the RAID 5 group.
RAID 0 group (nonredundant individual access array) — Use a
RAID 0 group where the best overall performance is important. In terms of high availability, a RAID 0 group is less available than an individual unit.
You can improve availability by software mirroring the RAID 0 group if the operating system supports mirroring. A RAID 0 group (like a RAID 5 group) requires a minimum of three disks; software mirroring a RAID 0 group requires the same number of disks used in the original image. A RAID 0 group is useful for applications using short-term data to which you need quick access, such as mail.
RAID 1/0 group (mirrored RAID 0 group) — A RAID 1/0 group provides the best balance of performance and availability. You can use it very effectively for any of the RAID 5 applications. A RAID 1/0 group requires a minimum of six disks.
Individual unit — An individual unit is useful for print spooling, user file exchange areas, or other such applications, where high availability is not important or where the information stored is easily restorable from backup.
Individual disk units are flexible; later, if the operating system supports software mirroring, you can always use software mirroring with the disk.
The performance of an individual unit is slightly less than a standard disk not in an storage system. The slight degradation results from SP overhead.
Hot spare — A hot spare provides no data storage but enhances the availability of each RAID 5, RAID 3, RAID 1, and RAID 1/0 group in a storage system. Use a hot spare where you must regain high availability quickly without human intervention if any disk in such a RAID group fails.
A hot spare also minimizes the period of degraded performance after a
RAID 5 or RAID 3 disk fails.
2-15
What next?
What next?
This chapter explained RAID group types and tradeoffs. To plan LUNs and
file systems for shared storage, continue to Chapter 3; or for unshared
storage, skip to Chapter 4. For details on the storage-system hardware —
shared and unshared — skip to Chapter 5.
For storage-system management utilities, skip to Chapter 6.
2-16
014-002912-01
3
Planning file systems and LUNs with shared storage in a SAN
This chapter shows a sample RAID, LUN, and Storage Group configuration with shared storage, and then provides worksheets for planning your own shared storage installation. Topics are
• Paths to LUNs
• Sample shared storage configuration
• Shared storage planning worksheets
Dual paths to LUNs
A shared storage system includes one or more servers, two Fibre Channel switches, one or more storage systems, each with two SPs, and the shared model of storage system software (Licensed Internal Code).
With shared storage, there are two paths to each LUN in the storage system. The shared storage LIC detects both paths and, using a built-in application called Application Transparent Failover (ATF), can automatically switch to the other path if a device (like a host-bus adapter or cable) fails.
The SP on which you bind a disk becomes the default owner of the resulting
LUN. The route through the SP that owns a unit determines the primary route to that LUN. The route through the other SP is the secondary route to the LUN.
With a single server, you can bind disks to each SP to share the I/O load equally. With dual servers, each server has one primary SP; the SP that binds a LUN determines the server that’s the primary owner of the LUN.
Sample shared storage configuration
The following figure shows a sample shared storage system connected to three servers: two servers in a cluster and one server running a database management program.
3-1
014-002912-01
Sample shared storage configuration
Sample shared storage configuration
Highly available cluster
File Server (FS) Mail Server(MS)
Operating system A
Operating system A
Database Server(DS)
Operating system B
Private storage
Cluster
Storage Group
Database Server
Storage Group
Switch Switch
FS R5
Apps
Unbound disks
FS R5
Users
Disk IDs
6_0-6_9
5_0-5_9
FS R5
Files A
FS R5
Files B
4_0-4_9
MS R5
ISP A mail
MS R5
ISP B mail
3_0-3_9
MS R5
Users
MS R5
Specs
2_0-2_9
DS R5
Users
DS R5
Dbase2
1_0-1_9
SP A
DS R5 (6 disks) Dbase1
0_0-0_9
SP B Path 1
Path 2
The storage system disk IDs and servers’ Storage Group LUNs are as follows.
Clustered System LUNs
File Server LUNs (FS) - SP B
Disks RAID type, storage type
5_0-5_4–RAID 5, Applications
5_5-5_9 – RAID 5, Users
4_0-4_4 – RAID 5, Files A
4_5-4_9 – RAID 5, Files B
Mail Server LUNs (MS) - SP A
Disks RAID type, storage type
2_0-2_4 – RAID 5, ISP A mail
2_5-2_9 – RAID 5, ISP B mail
3_0-3_4 – RAID 5, Users
3_5-3_9 – RAID 5, Specs
Database Server LUNs (DS) -
SP A
Disks RAID type, storage type
0_0,0_ 1 – RAID 1, Log file for database Dbase1
0_2, 0_3 – RAID 1, Log file for database Dbase2
0_4-0_9 – RAID 5 (6 disks), Dbase1
1_0-1_4 – RAID 5, Users
1_5-1_9 – RAID 5, Dbase 2
6_0, 6_1 – Hot spare (usable for any server’s LUN)
3-2
014-002912-01
014-002912-01
Sample shared storage configuration
With 18-Megabyte disks, the LUN storage capacities and drive names are as follows.
File Server — 288 Gbytes on four LUNs
FS R5
Apps
Unit S on five disks bound as a RAID 5 group for 72 Gbytes of storage; for applications.
FS R5
Users
Unit T on five disks bound as a RAID 5 group for 72 Gbytes of storage; for user directories and files.
FS R5
FilesA
Unit U on five disks bound as a RAID 5 group for 72 Gbytes of storage; for file storage.
FS R5
FilesB Unit V on five disks bound as a RAID 5 group for 72 Gbytes of storage; for file storage.
Mail Server — 288 Gbytes on four LUNs
MS R5
ISP mail Unit O on five disks bound as a RAID 5 group for 72 Gbytes of storage; for the mail delivered via ISP A.
MS R5
ISP mail
Unit P on five disks bound as a RAID 5 group for 72 Gbytes of storage; for the mail delivered via ISP B.
MSR5
Users
Unit Q on five disks bound as a RAID 5 group for 72 Gbytes of storage; for user directories and files.
MS R5
Specs Unit R on five disks bound as a RAID 5 group for 72 Gbytes of storage; for specifications.
Database Server — 208 Gbytes on four LUNs
DS R5
Users
Unit users on five disks bound as a RAID 5 group for 72 Gbytes of storage; for user directories.
DS R5
Dbase2
Unit dbase2 on five disks bound as a RAID 5 group for 52 Gbytes of storage; for the second database system.
DS R1
Logs
Unit logfiles on two disks bound as a RAID 1 mirrored pair for
18 Gbytes of storage; for the database log files.
DS R5
Dbase1
Unit dbase on eight disks bound as a RAID 5 group for 144 Gbytes of storage; for the primary database system.
3-3
Planning applications, LUNs, and Storage Groups
Planning applications, LUNs, and Storage Groups
This section helps you plan your shared storage use — the applications to run, the LUNs that will hold them, and the storage groups that will belong to each server. The worksheets to help you do this include
• Application and LUN planning worksheet - lets you outline your storage needs.
• LUN and Storage Group planning worksheet - lets you decide on the disks to compose the LUNs and the LUNs to compose the storage groups for each server.
• LUN details worksheet - lets you plan each LUN in detail.
Make as many copies of each blank worksheet as you need. You will need this information later when you configure the shared storage system.
Sample worksheets appear later in this chapter.
3-4
014-002912-01
Planning applications, LUNs, and Storage Groups
Application
Application and LUN planning
Use the following worksheet to list the applications you will run and the
RAID type and size of LUN to hold them. For each application that will run in the SAN, write the application name, and file system (if any), RAID type,
LUN ID (ascending integers, starting with 0), disk space required, and finally the name of the servers and operating systems that will use the
LUN.
Application and LUN planning worksheet
File system, partition, or drive
RAID type of LUN
LUN
ID (hex)
Disk space req’d
(Gbytes)
Server name and operating system
Application
Mail 1
Mail 2
Database index
A sample worksheet begins as follows:
File system, partition, or drive
RAID type of LUN
LUN
ID (hex)
RAI D 5 0
RAI D 5 1
RAI D 1 2
Disk space req’d
(Gbytes)
72 Gb
72 Gb
18 Gb
Server name and operating system
Server1, NT
Server1, NT
Server2, NT
014-002912-01
Completing the application and LUN planning worksheet
Application . Enter the application name or type.
File system, partition , or drive . Write the drive letter (for NT only) and the partition , file system , logical volume , or drive letter (NT only) name.
With a system such as Windows NT, the LUNs are identified by drive letter only. The letter does not help you identify the disk configuration
(such as RAID 5). We suggest that later, when you use the operating system to create a partition on the unit, you use the disk administrator
3-5
Planning applications, LUNs, and Storage Groups software to assign a volume label that describes the RAID configuration. For example, for drive T, assign the volume ID
RAID5_T. The volume label will then identify the drive letter.
RAID type of LUN . This is the RAID group type you want for this partition, file system, or logical volume. The features of RAID types are
explained in Chapter 2. For a RAID 5, RAID 1, RAID 1/0, and RAID 0
group, you can create one or more LUNs on the RAID group. For other
RAID types, you can create only one LUN per RAID group.
LUN ID.
The LUN ID is a hexadecimal number assigned when you bind the disks into a LUN. By default, the ID of the first LUN bound is 0, the second 1, and so on. Each LUN ID must be unique within the storage system, regardless of its storage group or RAID group.
The maximum number of LUNs supported on one host-bus adapter depends on the operating system.
Disk space required (Gbytes) . Consider the largest amount of disk space this application will need, then add a factor for growth.
Server hostname and operating system . Enter the server hostname (or, if you don’t know the name, a short description that identifies the server) and the operating system name, if you know it.
LUN and Storage Group planning worksheet
Use the following worksheet to select the disks that will make up the LUNs and Storage Groups in the SAN. A shared storage system can include up to
120 disks, numbered 0 through 119, left to right from the bottom up.
3-6
014-002912-01
Planning applications, LUNs, and Storage Groups
LUN and Storage Group planning worksheet
11_0 11_1 11_211_3 11_4 11_5 11_6 11_7 11_811_9
10_0 10_1 10_210_3 10_4 10_5 10_6 10_7 10_8 10_9
9_0 9_1 9_2 9_3 9_4 9_5 9_6 9_7 9_8 9_9
8_0 8_1 8_2 8_3 8_4 8_5 8_6 8_7 8_8 9_9
7_0 7_1 7_2 7_3 7_4 7_5 7_6 7_7 7_8 7_9
6_0 6_1 6_2 6_3 6_4 6_5 6_6 6_7 6_8 6_9
5_0 5_1 5_2 5_3 5_4 5_5 5_6 5_7 5_8 5_9
4_0 4_1 4_2 4_3 4_4 4_5 4_6 4_7 4_8 4_9
3_0 3_1 3_2 3_3 3_4 3_5 3_6 3_7 3_8 3_9
2_0 2_1 2_2 2_3 2_4 2_5 2_6 2_7 2_8 2_9
1_0 1_1 1_2 1_3 1_4 1_5 1_6 1_7 1_8 1_9
0_0 0_1 0_2 0_3 0_4 0_5 0_6 0_7 0_8 0_9
Storage system number or name:_______________
Storage Group ID or name:______ Server hostname:_____________________ Dedicated Shared
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
Storage Group ID or name:______ Server hostname:_____________________ Dedicated Shared
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
Storage Group ID or name:______ Server hostname:_____________________ Dedicated Shared
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________________
014-002912-01
3-7
Planning applications, LUNs, and Storage Groups
Part of a sample LUN and Storage Group worksheet follows.
3_0 3_1 3_2 3_3 3_4 3_5 3_6 3_7 3_8 3_9
2_0 2_1 2_2 2_3 2_4 2_5 2_6 2_7 2_8 2_9
LUN 2
RAID 1
1_0 1_1 1_2 1_3 1_4 1_5 1_6 1_7 1_8 1_9
LUN 0
RAID 5
0_0 0_1 0_2 0_3 0_4 0_5 0_6 0_7 0_8 0_9
LUN 1
RAID 5
0
1
72
72
Server1 X
5
5
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs_________________________________
X
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs________________________________
LUN ID or name_______RAID type ___ Cap. (Gb) _____ Disk IDs________________________________
Completing the LUN and Storage Group planning worksheet
As shown, draw circles around the disks that will compose each LUN, and within each circle specify the RAID type (for example, RAID 5) and LUN ID.
This is information you will use to bind the disks into LUNs. For disk IDs, use the form shown. This form is enclosure _ diskID , where enclosure is the enclosure number (the bottom one is 0, above it 1, and so on) and diskID is the disk position (left is 0, next is 1, and so on).
IMPORTANT None of the disks 0_0 through 0_8 may be used as a hot spare.
Next, complete as many of the Storage System sections as needed for all the
Storage Groups in the SAN. Copy the (blank) worksheet as needed for all
Storage Groups in each storage system.
A storage system is any group of enclosures connected to a DPE; it can include up to 11 DAE enclosures for a total of 120 disks. If a Storage Group will be dedicated (not accessible by another system in a cluster), mark the
Dedicated box at the end of its line; if the Storage Group will be accessible to one or more other servers in a cluster, write the hostnames of all servers and mark the Shared box.
LUN Details worksheet
Use the following LUN details worksheet to plan the individual LUNs.
Complete as many of these as needed for all LUNs in your SAN.
3-8
014-002912-01
Planning applications, LUNs, and Storage Groups
LUN Details Worksheet
Storage system (complete this section once for each storage system)
Storage-system number or name:______
Storage-system configuration
V Shared storage
V Unshared storage: V Basic V Dual-adapter/dual-SP V Single-initiator/dual-loop
V Dual-initiator/dual-loop V Hub,single loop V Hub,dual loop
SP FC-AL address ID (unshared only):
SP memory (Mbytes): SP A:______
V Use for caching
V Use for RAID 3
SP A:_____SP B:_____
SP B: ______
Read cache size:___ MB Write cache size: ___ MB Cache page size:___KB
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
V
V
RAID 1 mirrored pair
Hot spare
SP: V
V
A V B
RAID 0
Servers that can access this LUN:
Operating system information: Device name: File system, partition, or drive:
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
014-002912-01
3-9
Planning applications, LUNs, and Storage Groups
LUN Details Worksheet
Storage system (complete this section once for each storage system)
Storage-system number or name:__
SS1
____
Storage-system configuration
X
Shared storage
V Unshared storage: V Basic V Dual-adapter/dual-SP V Single-initiator/dual-loop
V Dual-initiator/dual-loop V Hub,single loop V Hub,dual loop
SP FC-AL address ID (unshared only):
SP memory (Mbytes):
X
Use for caching
V Use for RAID 3
SP A:_____SP B:_____
SP A:_
256
__ SP B: __
256
__
Read cache size:_
80
_ MB Write cache size: _
160
_ MBCache page size:_
2
__KB
LUN ID:__
0
___
RAID Group ID:
0
Size, GB:
72
LUN sIze,GB:
72
Disk IDs:
RAID type:
X
RAID 5
V RAID 1/0
V RAID 3 - Memory, MB:___
V Individual disk
Caching:
X
Read and write V Write V Read V None
0_0, 0_1, 0_2, 0_3, 0_4
V RAID 1 mirrored pair
V Hot spare
SP:
X
A V B
V RAID 0
Servers that can access this LUN:
Server1
Operating system information: Device name: File system, partition, or drive:
T
RAID Group ID:
LUN ID:__
1
__
1
Size,GB:
72
LUN sIze,GB:
72
Disk IDs:
0_5, 0_6, 0_7, 0_8, 0_9
RAID type:
X
RAID 5 V RAID 3 - Memory, MB:___ V RAID 1 mirrored pair
V RAID 1/0 V Individual disk
Caching:
X
Read and write V Write V Read V None
V Hot spare
Servers that can access this LUN:
Server1
SP:
X
A V B
V RAID 0
Operating system information: Device name: File system, partition, or drive:
U
RAID Group ID:
2
Size,GB:
18
LUN ID:__
2
__
LUN sIze,GB:
18
Disk IDs:
RAID type: V RAID 5 V RAID 3 - Memory, MB:___
V RAID 1/0 V Individual disk
Caching:
X
Read and write V Write V Read V None
Servers that can access this LUN:
Server1
Operating system information: Device name:
1_0, 1_1
X
RAID 1 mirrored pair
V Hot spare
File system, partition, or drive:
SP:
X
A V B
V RAID 0
V
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
3-10
014-002912-01
Planning applications, LUNs, and Storage Groups
014-002912-01
Completing the LUN details worksheet
Complete the header portion of the worksheet for each storage system as described below. Copy the blank worksheet as needed.
Storage-system entries
Storage-system configuration , specify Shared storage.
SP FC-AL address ID . This does not apply to shared storage, in which the switch determines the address of each device.
Use memory for caching.
You can use SP memory for read/write caching or RAID 3. (Using both caching and RAID 3 in the same storage system is not recommended.) If you choose caching, check the box and continue to the next step; for RAID 3, skip to the RAID Group
ID entry.
Read cache size.
If you want a read cache, it should generally be about one third of the total available cache memory.
Write cache size.
The write cache should be two thirds of the total available. Some memory is required for system overhead, so you cannot determine a precise figure at this time. For example, for 256
Mbytes of total memory, you might have 240 Mbytes available, and you would specify 80 Mbytes for the read cache and 160 Mbytes for the write cache.
Cache page size.
This applies to both read and write caches. It can be
2, 4, 8, or 16 Kbytes. As a general guideline, we suggest
• For a general-purpose file server — 8 Kbytes
• For a database application — 2 or 4 Kbytes
The ideal cache page size depends on the operating system and application.
Use memory for RAID 3 . If you want to use the SP memory for
RAID 3, check the box.
RAID Group/LUN entries
Complete a RAID Group/LUN entry for each LUN and hot spare.
LUN ID . The LUN ID is a hexadecimal number assigned when you bind the disks into a LUN. By default, the ID of the first LUN bound is 0, the second 1, and so on. Each LUN ID must be unique within the storage system, regardless of its storage group or RAID group.
The maximum number of LUNs supported on one host-bus adapter depends on the operating system.
3-11
Planning applications, LUNs, and Storage Groups
RAID Group ID . This ID is a hexadecimal number assigned when you create the RAID Group. By default, the number of the first RAID group in a storage system is 0, the second 1, and so on, up to the maximum of 1F (31).
Size (RAID group size). Enter the user-available capacity in gigabytes
(Gbytes) of the whole RAID group. You can determine the capacity as follows:
RAID5 or RAID-3 group: disk-size * (number-of-disks - 1)
RAID 1/0 or RAID-1 group: (disk-size * number-of-disks) / 2
RAID 0 group:
Individual unit: disk-size * number-of-disks disk-size
For example,
•
A five-disk RAID 5 or RAID 3 group of 18-Gbyte disks holds 72
Gbytes;
•
An eight-disk RAID 1/0 group of 18-Gbyte disks also holds
72Gbytes;
•
A RAID 1 mirrored pair of 18-Gbyte disks holds 18 Gbytes; and
•
An individual disk of an 18-Gbyte disk also holds 18 Gbytes.
Each disk in the RAID group must have the same capacity; otherwise, you will waste disk storage space.
LUN size . Enter the user-available capacity in gigabytes (Gbytes) of the LUN. You can make this the same size as the RAID group, above.
Or, for a RAID 5, RAID 1, RAID 1/0, or RAID 0 group, you can make the LUN smaller than the RAID group. You might do this if you wanted a RAID 5 group with a large capacity and wanted to place many smaller capacity LUNs on it; for example, to specify a LUN for each user. However, having multiple LUNs per RAID group may adversely impact performance. If you want multiple LUNs per RAID group, then use a RAID Group/LUN series of entries for each LUN.
Disk IDs . Enter the ID(s) of all disks that will make up the LUN or hot spare. These are the same disk IDs you specified on the previous worksheet. For example, for a RAID-5 group in the DPE (enclosure 0 disks 2 through 6, enter 0_2, 0_3, 0_4, 0_5, and 0_6.
SP . Specify the SP that will own the LUN: SP A or SP B. You can let the management program automatically select the SP to balance the workload between SPs; to do so, leave this entry blank.
RAID type . Copy the RAID type from the previous worksheet. For example, RAID 5 or hot spare. For a hot spare (not strictly speaking a
3-12
014-002912-01
What next?
LUN at all), skip the rest of this LUN entry and continue to the next
LUN entry (if any).
If this is a RAID 3 group, specify the amount of SP memory for that group. To work efficiently, each RAID 3 group needs at least 6 Mbytes of memory.
Caching
. If you want to use caching (entry on page 3-11), you can
specify whether you want caching — read and write, read, or write for this LUN. Generally, write caching improves performance far more than read caching. The ability to specify caching on a LUN basis provides additional flexibility, since you can use caching for only the units that will benefit from it. Read and write caching recommendations follow.
Cache recommendations for different RAID types
RAID 5
Highly
Recommended
RAID 3
Not allowed
RAID 1 RAID 1/0 RAID 0
Individual
Unit
Recommended Recommended Recommended Recommended
Servers that can access this LUN . Enter the name of each server
(copied from the LUN and Storage Group worksheet).
Operating system information: Device name . Enter the operating system device name, if this is important and if you know it. Depending on your operating system, you may not be able to complete this field now.
File system, partition, or drive . Write the name of the file system, partition, or drive letter you will create on this LUN. This is the same name you wrote on the application worksheet.
On the following line, write any pertinent notes; for example, the file system mount- or graft-point directory pathname (from the root directory). If this storage system’s chassis will be shared with another server, and the other server is the primary owner of this disk, write
“secondary.” (As mentioned earlier, if the storage system will be used by two servers, we suggest you complete one of these worksheets for each server.)
What next?
This chapter outlined the planning tasks for shared storage systems. If you have completed the worksheets to your satisfaction, you are ready to learn
about the hardware needed for these systems as explained in Chapter 5.
014-002912-01
3-13
4
Planning LUNs and file systems with unshared storage
This chapter shows sample RAID and LUN configurations with unshared storage, and then provides worksheets for planning your own shared storage installation. Topics are
• Paths to LUNs
• Sample unshared storage configurations
• Unshared storage planning worksheets
Dual SPs and paths to LUNs
With two SPs, you have two routes to LUNs. The SP on which you bind a disk is the default owner of the resulting LUN. The route through the SP that owns a unit determines the primary route to that LUN. The route through the other SP is the secondary route to the LUN.
With a single server, you can bind disks to each SP to share the I/O load equally. With dual servers, each server has one primary SP; the SP that binds a LUN determines the server that’s the primary owner of the LUN.
Unshared storage system configurations
This section explains the different options available for connecting storage systems to servers. As needs change, you may want to change a configuration. You can do so without changing your LUN configuration or losing user data.
There are three types of configuration:
• Single-server configurations
• Dual-server configurations
• Configurations with hubs
Single-server configurations
There are two single-server storage-system configurations: basic and dual-adapter/dual-SP. The following table and pictures show the features of each.
4-1
014-002912-01
Unshared storage system configurations
Storage-system configurations for a single server
Configuration type
Basic
One server
One adapter
One cable
One SP
Dual-adapter/dual-SP
One server
Two adapters
Two cables
Two SPs
Features
With a RAID group of any level other than 0, applications can continue after failure of any disk, but cannot continue after failure of an adapter, loop, or SP.
Provides highest availability and best storage-system performance of the single-server configurations; recommended for high-availability servers. With a RAID group of any level other than 0, applications can continue after any disk fails.
If one adapter, loop, or SP fails, the software or system operator can transfer control to the other adapter.
Basic configuration
Server
4-2
Storage system
FC loop
SP A
Dual-adapter/dual-SP configuration
Server
Storage system
DAE
FC loop 1
FC loop 2
Dual-server configurations
Dual-server configurations offer higher availability: if one server fails, another server can take over and run applications. Sharing storage systems between servers can lower costs. When two or more servers connect to the same storage system, the servers should run cluster software to retain possession of their disks.
014-002912-01
014-002912-01
Unshared storage system configurations
There are two dual-server configurations: single-initiator/dual-loop and dual-initiator/dual-loop. The following table and pictures show the features of each.
Dual-server configurations
Configuration type Features
SIngle-initiator/dual-loop
Two servers
Two adapters
(one per server)
Two cables
(one per server)
Two SPs
Dual-initiator/dual-loop
Two servers
Four adapters
(two per server)
Four cables
Two SPs
Resembles two basic configurations using the same storage system, and provides some high availability for two servers. Each server and its applications can continue after failure of any disk. A server using a failed adapter or SP cannot continue after failure, but the other server can continue and run the first server’s applications. Some operating systems support failover software that can automatically direct one server to take over the other’s disks if the other fails. Or the system operator can transfer the disks manually.
Two servers share two Fibre Channel loops. This provides the highest availability and best storage-system performance for dual-server configurations.
With a RAID group of any level other than 0, applications can continue after any disk fails.
If a server, adapter, cable, or SP fails, the system operator can transfer the affected disks to the other server, and then restart applications. Or the transfer and restart can occur automatically with failover software.
Single-initiator/dual-loop configuration
Highly available cluster
Server 1 Server 2
Storage system
SP B SP A
FC loop 1
FC loop 2
4-3
Unshared storage system configurations
Dual-initiator/dual-loop configuration
Highly available cluster
Server 1 Server 2
Storage system
SP B SP A
FC loop 1
FC loop 2
Configurations with hubs
A hub adds flexibility and availability to a Fibre Channel site. Essentially, a hub lets you expand any of the previous configurations to include multiple servers and storage systems. In any of these configurations, you can have one or more host-bus adapters in the servers, one or more hubs, and one or
two SPs in each storage system. Each hub has nine ports as shown on page
In any hub configuration, you cable the server(s) to the hub, and the hub to the storage system(s). To add a storage system, cable it to the hub; all servers connected to the hub then have access to that storage system. To add a server, cable it to the hub; the server has access to all storage systems connected to the hub. This eliminates the vulnerability of the daisy chain, since any storage system can be shut down without affecting the others.
The distance between nodes (servers, hubs, and storage systems) depends on the type of cable between them. An adapter that supports optical cable can exploit optical cable ’s speed and distance features; an adapter that does not support optical cables must use copper cables. A hub supports copper cables and — with MIAs — optical cables.
4-4
014-002912-01
Sample unshared storage configurations
Sample configuration with hubs and high-availability options
Highly available cluster
Server 1 Server 2
Hub Hub
Storage system 1
Cables
Two per server,
Two per storage system
SP B SP A
Storage system 2
FC loop 1
FC loop 2
SP B SP A
Sample unshared storage configurations
This section shows disks in three unshared storage-system configurations
— single-server dual-loop without a hub, dual-initiator/dual-loop without a hub, and multiple server dual-loop with two hubs.
Single server example without hub
Server
Database
RAID 5
Unused
FC loop 1
FC loop 2
SP B
Clients, Mail
RAID 1/0
SP A
Disk IDs
1_0-1_9
0_0-0_9
014-002912-01
4-5
Sample unshared storage configurations
The storage system disk IDs and LUNs are as follows. The LUN capacities shown assume 18-Gbyte disks.
LUNs - SP A and SP B, 234 Gbytes
Disks RAID type, storage type
0_0, 0_1 – RAID 1, System disk, 18 Gbytes
0_2-0_9 – RAID 5 (8 disks), Clients and Mail, 126 Gbytes
1_0-1_4 – RAID 5, Database, 72 Gbytes
1_5 – Disk, Temporary storage, 18 Gbytes
Dual-server example without hub
Server 1 (S1) Server 2 (S2)
System disk
S2 Customers
RAID 1/0
Storage system
SP B
S1Dbase
RAID 5
SP A
FC loop 1
FC loop 2
If each disk holds 18 Gbytes, then the storage-system chassis provides
Server 1 with 108 Gbytes of disk storage, 90 Gbytes highly available; it provides Server 2 with 90 Gbytes of storage, all highly available. Each server has its own SP, which controls that server’s LUNs; those LUNs remain primary to that server. The LUNs are as follows. The LUN capacities shown assume 18-Gbyte disks.
Server1 LUNs (S1) - SP B, 126 Gbytes Server2 LUNs (S2) - SP A, 126 Gbytes
Disks RAID type, storage type, capacity Disks RAID type, storage type, capacity
0_0, 0_1 – RAID 1, System disk, 18
Gbytes
0_2 – Disk, Temporary storage,18
Gbytes
0_3-0_7 – RAID 5, Database, 72
Gbytes
0_8-0_ 9 – RAID 1, Users, 18 Gbytes
1_0-1_7 – RAID 5 (8 disks), Customer
Accounts, 126 Gbytes
4-6
014-002912-01
014-002912-01
Sample unshared storage configurations
Multiple server example with hubs
Highly available cluster
Server1 (S1) Server2 (S2)
Hub
System disk
Hub
S2 Mail
RAID 5
Unused 3_0-3_9
FC loop 1
FC loop 2
S2 Users
RAID 5
S2 Accts
RAID 5
2_0-2_9
S1 R5
Users
1_0-1_9
S1 Dbase
RAID 5 (8 disks)
0_0-0_9
SP A SP B
The storage system disk IDs and servers’ Storage Group LUNs are as follows. The LUN capacities shown assume 18-Gbyte disks.
Server1 LUNs (S1) - SP A, 234 Gbytes
Disks RAID type, storage type
0-0, 0_1 – RAID 1, System disk, 18
Gbytes
0_2-0_9 – RAID 5 (8 disks), Database,
126 Gbytes
1_0-1_4 – RAID 5, Users, 72 Gbytes
1_5,1_6 – RAID 1, Log files, 18 Gbytes
1_7 – Hot spare (usable in any server’s LUN)
Server2 LUNs (S2) - SP B, 216 Gbytes
Disks RAID type, storage type
2_0-2_4 – RAID 5,Users, 72 Gbytes
2_5-2_9 – RAID 5, Accounts, 72 Gbytes
3_0-3_4 – RAID 5, Mail, 72 Gbytes
With a dual-loop configuration, for highest availability, all the servers should run the same operating system and participate in a cluster. The cluster environment will prevent one server from inadvertently overwriting the data owned by another server.
4-7
Planning applications and LUNs - Unshared storage
Planning applications and LUNs - Unshared storage
This section helps you plan your unshared storage use — applications you want to run and the LUNs that will hold them. The worksheets to help you do this include
• Application and file system planning worksheet - lets you outline your storage needs.
• LUN planning worksheet - lets you decide on the disks that will compose the LUNs.
• LUN details worksheet - lets you plan each LUN in detail.
Make as many copies of each blank worksheet as you need. You will need this information later when you configure the shared storage system.
Sample file system, Storage Group, and LUN worksheets appear later in this chapter.
4-8
014-002912-01
Planning applications and LUNs - Unshared storage
Application
Application and LUN planning
Use the following worksheet to plan your file systems and RAID types. For each application, write the application name, file system (if any), RAID type, LUN ID (ascending integers, starting with 0), disk space required, and finally the name of the servers and operating systems that will use the
LUN.
Application and LUN planning worksheet
File system (if any)
RAID type of LUN
LUN
ID (hex)
Disk space req’d
(Gbytes)
Server name and operating system
Application
Mail 1
Mail 2
Database index
A sample worksheet begins as follows:
File system (if any)
RAID type of LUN
LUN
ID (hex)
RAI D 5 0
RAI D 5 1
RAI D 1 2
Disk space req’d
(Gbytes)
72 Gb
72 Gb
18 Gb
Server name and operating system
Server1, NT
Server1, NT
Server2, NT
Completing the application and LUN planning worksheet
Application . Enter the application name or type.
File system, partition , or drive . Write the drive letter (for NT only) and the partition , file system , logical volume , or drive letter (NT only) name.
With a system such as Windows NT, the LUNs are identified by drive letter only. The letter does not help you identify the disk configuration
(such as RAID 5). We suggest that later, when you use the operating system to create a partition on the unit, you use the disk administrator software to assign a volume label that describes the RAID
014-002912-01
4-9
Planning applications and LUNs - Unshared storage configuration. For example, for drive T, assign the volume ID
RAID5_T. The volume label will then identify the drive letter.
The RAID type of LUN is the RAID group type you want for this partition, file system, or logical volume. The features of RAID types are
explained in Chapter 2. For a RAID 5, RAID 1, RAID 1/0, and RAID 0
group, you can create one or more LUNs on the RAID group. For other
RAID types, you can create only one LUN per RAID group.
The LUN ID is a hexadecimal number assigned when you bind the disks into a LUN. By default, the ID of the first LUN bound is 0, the second 1, and so on. Each LUN ID must be unique within the storage system, regardless of its storage group or RAID group.
The maximum number of LUNs supported on one host-bus adapter depends on the operating system. Some systems allow only eight LUNs
(numbers 0 through 7). For an operating system with this restriction, if you want a hot spare, assign the hot spare an ID above 7; for example, 8 or 9. The operating system never accesses a hot spare, so the ID is irrelevant to it.
For Disk space required (Gbytes) , consider the largest amount of disk space this application will need, then add a factor for growth.
For Server hostname and operating system , enter the server hostname (or, if you don’t know the name, a short description that identifies the server) and the operating system name, if you know it.
If this storage system will be used by two servers, provide a copy of this worksheet to the other server. This is particularly important where one server may take over the other’s LUNs. If a LUN will be shared, on the Notes section of the LUN details worksheet, write “Primary to server-name ” or “Secondary to server-name .”
LUN planning worksheet
Use one of the following worksheets (Rackmount or Deskside) to select the disks that will make up the LUNs. Depending on model, a full-fibre rackmount storage system can include up to 120 disks, numbered 0 through
119, left to right from the bottom up. A 30-slot disk system with SCSI disks has a 30-disk maximum and different disk IDs.
Again depending on model, a deskside storage system can hold ten, 20, or 30 disks.
4-10
014-002912-01
Planning applications and LUNs - Unshared storage
LUN planning worksheet - Rackmount
Full-fibre storage system
11_0 11_1 11_2 11_3 11_4 11_5 11_6 11_7 11_8 11_9
10_0 10_1 10_2 10_3 10_4 10_5 10_6 10_7 10_8 10_9
9_0 9_1 9_2 9_3 9_4 9_5 9_6 9_7 9_8 9_9
8_0 8_1 8_2 8_3 8_4 8_5 8_6 8_7 8_8 8_9
7_0 7_1 7_2 7_3 7_4 7_5 7_6 7_7 7_8 7_9
6_0 6_1 6_2 6_3 6_4 6_5 6_6 6_7 6_8 6_9
5_0 5_1 5_2 5_3 5_4 5_5 5_6 5_7 5_8 5_9
4_0 4_1 4_2 4_3 4_4 4_5 4_6 4_7 4_8 4_9
3_0 3_1 3_2 3_3 3_4 3_5 3_6 3_7 3_8 3_9
2_0 2_1 2_2 2_3 2_4 2_5 2_6 2_7 2_8 2_9
1_0 1_1 1_2 1_3 1_4 1_5 1_6 1_7 1_8 1_9
0_0 0_1 0_2 0_3 0_4 0_5 0_6 0_7 0_8 0_9
30-slot SCSI-disk storage system
A0 B0 C0 D0 E0 A3 B3 C3 D3 E3
A1 B1 C1 D1 E1 A4 B4 C4 D4 E4
A2 B2 C2 D2 E2 A5 B5 C5 D5 E5
Storage system number_______
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs____________________________________
014-002912-01
4-11
Planning applications and LUNs - Unshared storage
LUN planning worksheet - Deskside
0_0
Full-fibre storage system
1_0 2_0
0_1
0_2
1_1
1_2
2_1
2_2
0_3
0_4
1_3
1_4
2_3
2_4
0_5 1_5
1_6
2_5
2_6
0_6
0_7 1_7
0_8
0_9
1_8
1_9
2_7
2_8
2_9
C5
D5
E5
30-slot SCSI disk storage system
A2 A1
B2 B1
A0
B0
C2 C1
D2
E2
D1
E1
C0
D0
E0
A5
A4
B5 B4
A3
B3
C4
D4
C3
D3
E4 E3
Storage system number_____
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs___________________________________
4-12
014-002912-01
Planning applications and LUNs - Unshared storage
A sample LUN worksheet follows.
2_0 2_1 2_2 2_3 2_4 2_5 2_6 2_7 2_8 2_9
LUN 2
RAID 1
1_0 1_1 1_2 1_3 1_4 1_5 1_6 1_7 1_8 1_9
LUN 0
RAID 5
0_0 0_1 0_2 0_3 0_4 0_5 0_6 0_7 0_8 0_9
LUN 1
RAID 5
30-slot SCSI-disk storage system
A0 B0 C0 D0 E0 A3 B3 C3 D3 E3
A1 B1 C1 D1 E1 A4 B4 C4 D4 E4
A2 B2 C2 D2 E2 A5 B5 C5 D5 E5
0
1
5
5
72
72
18
0_5, 0_6, 0_7, 0_8, 0_9
LUN number_______RAID type ___ Cap. (Gb) _____ Disk IDs______________________________________
Completing the LUN planning worksheet
As shown, draw circles around the disks that will compose each LUN, and within each circle specify the RAID type (for example, RAID 5) and LUN ID.
This is information you will use to bind the disks into LUNs. For disk IDs, use the form shown. This form is enclosure _ diskID , where enclosure is the enclosure number (the bottom one is 0, above it 1, and so on) and diskID is the disk position (left is 0, next is 1, and so on).
IMPORTANT On a full-fibre system, none of the disks 0_0 through 0_8 may be used as a hot spare. On a SCSI-disk system, none of the disks
A0, B0, C0, or A3 may be used as a hot spare.
Next, complete as many of the LUN sections as needed for each storage system.Copy the (blank) worksheet as needed for all LUNs in each storage system. A storage system is any group of enclosures connected to a DPE; a full-fibre system can include up to 11 DAE enclosures for a total of 120 disks. If a Storage Group
LUN Details worksheet
Use the following LUN details worksheet to plan the individual LUNs.
Complete as many of these as needed for all LUNs.
014-002912-01
4-13
Planning applications and LUNs - Unshared storage
LUN Details Worksheet
Storage system (complete this section once for each storage system)
Storage-system number or name:______
Storage-system configuration
V Shared storage
V Unshared storage: V Basic V Dual-adapter/dual-SP V Single-initiator/dual-loop
V Dual-initiator/dual-loop V Hub,single loop V Hub,dual loop
SP FC-AL address ID (unshared only):
SP memory (Mbytes): SP A:______
V Use for caching
V Use for RAID 3
SP A:_____SP B:_____
SP B: ______
Read cache size:___ MB Write cache size: ___ MB Cache page size:___KB
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
V
V
RAID 1 mirrored pair
Hot spare
SP: V
V
A V B
RAID 0
Servers that can access this LUN:
Operating system information: Device name: File system, partition, or drive:
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
4-14
014-002912-01
Planning applications and LUNs - Unshared storage
LUN Details Worksheet
Storage system (complete this section once for each storage system)
Storage-system number or name:__
1
____
Storage-system configuration
V Shared storage
X
Unshared storage: V Basic V Dual-adapter/dual-SP
X
Single-initiator/dual-loop
V Dual-initiator/dual-loop V Hub,single loop V Hub,dual loop
SP FC-AL address ID (unshared only):
SP memory (Mbytes):
X
Use for caching
V Use for RAID 3
SP A:_____SP B:_____
SP A:_
128
__ SP B: __
128
__
Read cache size:_
40
_ MB Write cache size: _
80
_ MBCache page size:_
2
__KB
LUN ID:__
0
___
RAID Group ID:
0
Size, GB:
72
LUN sIze,GB:
72
Disk IDs:
RAID type:
X
RAID 5
V RAID 1/0
V RAID 3 - Memory, MB:___
V Individual disk
Caching:
X
Read and write V Write V Read V None
Servers that can access this LUN:
Server1
0_0, 0_1, 0_2, 0_3, 0_4
SP:
X
A V B
V RAID 1 mirrored pair
V Hot spare
V RAID 0
Operating system information: Device name: File system, partition, or drive:
T
RAID Group ID:
LUN ID:__
1
__
1
Size,GB:
72
LUN sIze,GB:
72
Disk IDs:
0_5, 0_6, 0_7, 0_8, 0_9
RAID type:
X
RAID 5
V RAID 1/0
V RAID 3 - Memory, MB:___
V Individual disk
Caching:
X
Read and write V Write V Read V None
V RAID 1 mirrored pair
V Hot spare
SP:
X
A V B
V RAID 0
Servers that can access this LUN:
Server1
Operating system information: Device name: File system, partition, or drive:
U
RAID Group ID:
2
Size,GB:
18
LUN ID:__
2
__
LUN sIze,GB:
18
Disk IDs:
RAID type: V RAID 5 V RAID 3 - Memory, MB:___
V RAID 1/0 V Individual disk
Caching:
X
Read and write V Write V Read V None
Servers that can access this LUN:
Server1
Operating system information: Device name:
1_0, 1_1
X
RAID 1 mirrored pair
V Hot spare
File system, partition, or drive:
SP:
X
A V B
V RAID 0
V
RAID Group ID: Size,GB:
LUN ID:_____
LUN sIze,GB: Disk IDs:
RAID type: V RAID 5
V RAID 1/0
V
V
RAID 3 - Memory, MB:___
Individual disk
Caching: V Read and write V Write V Read V None
Servers that can access this LUN:
Operating system information: Device name:
V
V
RAID 1 mirrored pair
Hot spare
SP:
File system, partition, or drive:
V A V B
V RAID 0
014-002912-01
4-15
Planning applications and LUNs - Unshared storage
Completing the LUN details worksheet
Complete the header portion of the worksheet for each storage system as described below. Copy the blank worksheet as needed. Sample completed
LUN worksheets appear later.
Storage-system entries
Storage-system configuration . Specify Unshared. Then specify the unshared configuration; for example , single-initiator, dual-loop.
For any multiple-server configuration, each server will need cluster software.
SP FC-AL address ID . For unshared storage, which uses FC-AL addressing, each SP (and each other node) on a Fibre Channel loop must have a unique FC-AL address ID. You set the SP FC-AL address
ID using switches on the back panel of the SP. The valid FC-AL address ID range is a number 0 through 125 decimal, which is 0 through 7D hexadecimal. For any number above 9, we suggest hexadecimal, since the switches are marked in hexadecimal.
If you have two FC-AL loops, we suggest a unique FC-AL address ID for each SP on both loops.
SP memory . Enter the amount of memory each SP has. If a storage system has two SPs, they will generally have the same amount of memory. You can allocate this memory to storage-system caching or
RAID 3 use.
Use memory for caching . You can use SP memory for read/write caching or RAID 3. (Using both caching and RAID 3 in the same storage system not recommended.) If you choose caching, check the box and continue to the next step; for RAID 3, skip to the RAID Group ID entry.
Read cache size . If you want a read cache, it should generally be about one third of the total available cache memory.
Write cache size . The write cache should be two thirds of the total available. Some memory is required for system overhead, so you cannot determine a precise figure at this time. For example, for 256
Mbytes of total memory, you might have 240 Mbytes available, and you would specify 80 Mbytes for the read cache and 160 Mbytes for the write cache.
Cache page size . This applies to both read and write caches. It can be
2, 4, 8, or 16 Kbytes. As a general guideline, we suggest
For a general-purpose file server — 8 Kbytes
For a database application — 2 or 4 Kbytes
4-16
014-002912-01
014-002912-01
Planning applications and LUNs - Unshared storage
The ideal cache page size depends on the operating system and application.
Use memory for RAID 3 . If you want to use the SP memory for
RAID 3, check the box.
RAID Group/LUN entries
Complete a RAID Group/LUN entry for each LUN and hot spare.
LUN ID . The LUN ID is a hexadecimal number assigned when you bind the disks into a LUN. By default, the ID of the first LUN bound is 0, the second 1, and so on. Each LUN ID must be unique within the storage system, regardless of its storage group or RAID group.
The maximum number of LUNs supported on one host-bus adapter depends on the operating system. Some systems allow only eight LUNs
(numbers 0 through 7). For an operating system with this restriction, if you want a hot spare, assign the hot spare an ID above 7; for example, 8 or 9. The operating system never accesses a hot spare, so the ID is irrelevant to it.
RAID Group ID . This is a hexadecimal number assigned when you create the RAID Group. By default, the number of the first RAID group in a storage system is 0, the second 1, and so on, up to the maximum of 1F (31).
For size (RAID group size), enter the user-available capacity in gigabytes (Gbytes) of the whole RAID group. You can determine the capacity as follows:
RAID-5 or RAID-3 group:
RAID 1/0 or RAID-1 group:
RAID-0 group:
Individual unit:
For example, disk-size * (number-of-disks - 1)
(disk-size * number-of-disks) / 2 disk-size * number-of-disks disk-size
•
A five-disk RAID 5 or RAID 3 group of 18-Gbyte disks holds 72
Gbytes;
•
An eight-disk RAID 1/0 group of 18-Gbyte disks also holds
72Gbytes;
•
A RAID 1 mirrored pair of 18-Gbyte disks holds 18 Gbytes; and
•
An individual disk of a 9-Gbyte disk also holds 18 Gbytes.
Each disk in the RAID group must have the same capacity; otherwise, you will waste disk storage space.
4-17
Planning applications and LUNs - Unshared storage
LUN Size . Enter the user-available capacity in gigabytes (Gbytes) of the LUN. You can make this the same size as the RAID group, above.
Or, for a RAID 5, RAID 1, RAID 1/0, or RAID 0 group, you can make the LUN smaller than the RAID group. You might do this if you wanted a RAID 5 group with a large capacity and wanted to place many smaller capacity LUNs on it; for example, to specify a LUN for each user. However, having multiple LUNs per RAID group may adversely impact performance. If you want multiple LUNs per RAID group, then use a RAID Group/LUN series of entries for each LUN.
Disk IDs . Enter the ID(s) of all disks that will make up the LUN or hot spare. These are the same disk IDs you specified on the previous worksheet. For example, for a RAID-5 group in the DPE (enclosure 0 disks 2 through 6, enter 0_2, 0_3, 0_4, 0_5, and 0_6.
SP . Specify the SP that will own the LUN: SP A or SP B. You can let the management program automatically select the SP to balance the workload between SPs; to do so, leave this entry blank.
RAID type . Copy the RAID type from the previous worksheet. For example, RAID 5 or hot spare. For a hot spare (not strictly speaking a
LUN at all), skip the rest of this LUN entry and continue to the next
LUN entry (if any).
If this is a RAID 3 group, specify the amount of SP memory for that group. To work efficiently, each RAID 3 group needs at least 6 Mbytes of memory.
Caching.
If you want to use caching (entry on page 4-16), you can
specify whether you want caching — read and write, read, or write for this LUN. Generally, write caching improves performance far more than read caching. The ability to specify caching on a LUN basis provides additional flexibility, since you can use caching for only the units that will benefit from it. Read and write caching recommendations follow.
Cache recommendations for different RAID types
RAID 5
Highly
Recommended
RAID 3
Not allowed
RAID 1 RAID 1/0 RAID 0
Individual
Unit
Recommended Recommended Recommended Recommended
4-18
Servers that can access this LUN . Enter the name of each server that will be able to use the LUN. Normally, you need to restrict access by establishing SP ownership of LUNs when you bind them.
Operating system information: Device name . Enter the operating system device name, if this is important and if you know it. Depending
014-002912-01
What next?
on your operating system, you may not be able to complete this field now.
File system, partition, or drive . Write the name of the file system, partition, or drive letter you will create on this LUN. This is the same name you wrote on the application worksheet.
On the following line, write any pertinent notes; for example, the file system mount- or graft-point directory pathname (from the root directory). If this storage system’s chassis will be shared with another server, and the other server is the primary owner of this disk, write
“secondary.” (If the storage system will be used by two servers, we suggest you complete one of these worksheets for each server.)
What next?
This chapter outlined the planning tasks for unshared storage systems. If you have completed the worksheets to your satisfaction, you are ready to learn about the hardware needed for these systems as explained in
014-002912-01
4-19
5
Storage-system hardware
The products described in this chapter compose the storage component of storage systems. The storage systems attach to the server and the interconnect components described in Chapter 1.
Shared storage - with switch
Server Server Server
Unshared storage
With hub
Server Server
Direct
Server
Server component
Interconnect component
Storage component
Switch
Hub
Disk-array storage systems
Topics in this chapter are
• Hardware for shared storage
• Hardware for unshared storage
• Planning your hardware components
• Cabinets for rackmount enclosures
• Cable and configuration guidelines
• Component planning worksheets
Hardware for shared storage
The primary hardware component for shared storage is a ten-slot
Disk-array Processor Enclosure (DPE) with two storage processors (SP).
The DPE can support up to 11 separate 10-slot enclosures called Disk Array
Enclosures (DAEs) for a total of 120 disks. Shared storage requires two SPs and LIC of a specific model.
A DPE with a DAE is available as a deskside system, but with a capacity of
20 disks this cannot provide the expandibility and total storage capacity needed for a SAN (storage area network). So this section does not cover the deskside version.
014-002912-01
5-1
Hardware for shared storage
Storage hardware — rackmount DPE-based storage systems
The DPE rackmount enclosure is a sheet-metal housing with a front door, a midplane, and slots for the storage processors (SPs), link control cards
(LCCs), disk modules, power supplies, and fan packs. All components can be replaced under power. The DPE rackmount model looks like the following figure.
DPE storage-system components – rackmount model
Power supplies
LCC
5-2
Link control card (LCC)
Disk modules (front door removed for clarity)
Storage processors (SPs)
FC ports with GBICs
Drive fan module
(detached for clarity)
A separate standby power supply (SPS) is required to support write caching.
All the shared storage components — rackmount DPE, DAEs, SPSs, and cabinet — are shown in the following figure.
Rackmount system with DPE and DAEs
Front view Back view
DAE
DAE
DPE
SPSs
SPs
Disks
The disks — available in differing capacities — fit into slots in the enclosure. Each module has a unique ID that you use when binding or
014-002912-01
014-002912-01
Hardware for shared storage monitoring its operation. The ID is derived from the enclosure address
(always 0 for the DPE, settable on a DAE) and the disk module slot numbers.
Disk modules and module IDs — rackmount DPE-based system
10 11 12 13 14 15 16 17 18 19
0 1 2 3 4 5 6 7 8 9
Storage processor (SP)
The SP provides the intelligence of the storage system. Using its own operating system (called Licensed Internal Code), the SP processes the data written to or read from the disk modules, and monitors the modules themselves. An SP consists of a printed-circuit board with memory modules
(DIMMs), and status lights.
For high availability, a storage system can support a second SP. The second
SP provides a second route to a storage system and also lets the storage system use write caching for enhanced write performance. Two SPs are required for shared storage.
Shared storage systems
Server Server Server
Fibre Channel switch
SP A SP B
Fibre Channel switch
SP A SP B
Storage systems
There are more examples of shared storage in Chapter 3.
5-3
Hardware for unshared storage
Hardware for unshared storage
Unshared storage systems are less costly and less complex than shared storage systems. They offer many shared storage system features; for example, you can use multiple unshared storage systems with multiple servers. However, with multiple servers, unshared storage offers less flexibility and security than shared storage, since any user with write access to privileged server files can enable access to any storage system.
Types of storage system for unshared storage
For unshared storage, there are four types of storage system, each using the
FC-AL protocol. Each type is available in a rackmount or deskside (office) version.
•
Disk-array Processor Enclosure (DPE) storage systems. A DPE is a
10-slot enclosure with hardware RAID features provided by one or two storage processors (SPs). In addition to its own disks, a DPE can support up to 110 additional disks in 10-slot Disk Array Enclosures (DAEs) for a total of 120 disks. This is the same kind of storage system used for shared storage, but it uses a different storage processor (SP).
•
Intelligent Disk Array Enclosure (iDAE). An iDAE, like a DPE, has SPs and thus all the features of a DPE, but is thinner and has a limit of 30 disks.
•
Disk Array Enclosure (DAE). A DAE does not have SPs. A DAE can connect to a DPE or an iDAE, or you can use it without SPs. A DAE used without an SP does not inherently include RAID, but can operate as a
RAID device using software running on the server system. Such a DAE is also known as Just a Box of Disks, or JBOD.
•
30-slot SCSI-disk storage systems. Like the DPE, these offer RAID features provided by one or two SPs. However, they use SCSI, not Fibre
Channel, disks. Each has space for 30 disks.
5-4
014-002912-01
Hardware for unshared storage
Disk-array processor enclosure (DPE)
Deskside DPE with DAE
Rackmount DPE, one enclosure, supports up to 11 DAEs
30-slot deskside
Intelligent disk-array enclosure (iDAE)
10-slot deskside Rackmount
30-slot SCSI-disk storage system
Deskside Rackmount
014-002912-01
5-5
Hardware for unshared storage
The following figure shows some components of a deskside DPE.
Components for the other types are similar.
DPE components – deskside model
Front
Back (fans and cables omitted for clarity)
Storage processors
(SPs)
DAE link control
card (LCC)
DPE LCC
FC ports
DAE power supplies
DPE power supplies
DPE
SP fan cover
(covers SP fan pack)
DAE
Front doors
(cover disk modules)
Power distribution
units
DPE LCC
SPS units
DAE LCC
Disks
The disks — available in differing capacities — fit into slots in the enclosure. Each disk has a unique ID that you use when binding it or monitoring its operation. The ID is the enclosure address (always 0 for the
DPE, settable on a DAE) and the disk slot number.
Disks and disk IDs in a full-fibre system
7
8
9
4
5
6
0
1
2
3
14
15
16
17
18
19
10
11
12
13
0 1 2 3 4 5 6 7 8 9
For a 30-slot system with SCSI disks, the disk IDs are based on the SCSI internal bus and the number of the disk on that bus, as follows:
5-6
014-002912-01
Hardware for unshared storage
C2
D2
E2
A5
B5
C5
D5
E5
A2
B2
Disks and disk IDs in a 30-slot SCSI disk system
Deskside model Rackmount model
Internal bus A
Internal bus B
Internal bus C
Internal bus D
Internal bus E
A4
B4
C4
D4
E4
A1
B1
C1
D1
E1
A0
B0
C0
D0
E0
A3
B3
C3
D3
E3
Internal bus A
Internal bus B
Internal bus C
Internal bus D
Internal bus E
A0 B0 C0 D0 E0
A3 B3 C3 D3 E3
A1 B1 C1 D1 E1
A4 B4 C4 D4 E4
A2 B2 C2 D2 E2 A5 B5 C5 D5 E5
Storage processor (SP)
The SP provides the intelligence of the storage system. Using its own operating system (called Licensed Internal Code), the SP processes the data written to or read from the disk modules, and monitors the modules themselves. An SP consists of a printed-circuit board with memory modules
(DIMMs), status lights, and switches for setting FC-AL addresses.
For high availability, a storage system can support a second SP. A second SP provides a second route to a storage system, so both SPs can connect to the same server or two different servers, as follows.
Storage system with two SPs connected to the same server
Server
Cables
FC loop 1
FC loop 2
Storage system
DAE(s)
SP A SP B
014-002912-01
5-7
Planning your hardware components
Storage system with two SPs connected to different servers
Highly available cluster
Server 1 Server 2
Storage system
SP B SP A
FC loop 1
FC loop 2
Either SP can control any LUN in the storage system, but only one SP at a time can control a LUN. If one SP cannot access a LUN it controlled
(because of a failure), you can transfer control of the LUN to the other SP, manually or via software.
Storage-system caching provides significant performance enhancement.
Read caching is available with one or two SPs. Mirrored write caching, particularly helpful with RAID 5 I/O, requires two SPs (to mirror one another, for cache integrity) and a Standby Power Supply (SPS) to enable the SPs to write their cached data to disk if power fails.
Planning your hardware components
This section helps you plan the hardware components — adapters, cables and hubs, and storage systems and site requirements — for each server in your installation.
For shared storage, you must use a DPE rackmount system with two SPs and high-availability options. We assume you have some idea of how many servers, adapters, hubs, storage systems, and SPs you want. Skip to the component data sheets following.
For unshared storage, you can use one or two SPs and you can choose among storage system configurations. This section assumes you have examined the
configurations shown starting on page 4-1 and have some idea of how many
servers, adapters, hubs, storage systems, and SPs you want. It ends with blank worksheets and sample worksheets.
Configuration tradeoffs - shared storage
The hardware configuration required for shared storage is very specific: two host-bus adapters in each attached server, two Fibre Channel switches, and two SPs per storage system. Choices you can make with shared storage systems include the number of storage systems (up to 15 are allowed), and for each storage system the cache configuration (maximum or minimum), and one or two standby power supplies (SPS units).
5-8
014-002912-01
014-002912-01
Planning your hardware components
The number of storage systems in the SAN depends on the servers’ processing demands. For each system, the larger cache improves write performance for very large processing loads; the redundant SPS lets write caching continue if one SPS fails.
Configuration tradeoffs - unshared storage
For each storage-system enclosure, you have two important areas of choice: rackmount or deskside model, and high-availability options.
Generally, rackmount systems are more versatile; you can add capacity in a cabinet without consuming more floor space. However, rackmount systems require additional hardware, such as cabinets and mounting rails, and someone must connect power cords and cables within them. For large storage requirements, rackmount systems may be more economical than deskside systems. Deskside systems are more convenient; they ship with all internal cabling in place and require only ac power and connection to the servers.
For high availability, there are many variations. The most important high-availability features are a second SP/LCC pair, second power supply, and standby power supply (SPS). The second SP/LCC and SPS let you use write caching to enhance performance; the second SP provides continuous access to storage-system disks if one SP or LCC fails. Another high-availability option is a redundant SPS.
Yet another option, for a deskside system, is a second power distribution unit (PDU), which lets you route ac power from an independent source.
Used this way, the second PDU protects against failure in one of the two ac power sources. With a rackmount system, you can acquire a cabinet with one or two ac inlet cords. The second inlet cord, connected to a second ac power source, provides the same advantage for all storage systems in the cabinet as the second PDU in the deskside storage system.
For deskside systems, the optional high-availability hardware fits into the deskside cabinet. Deskside high-availability options are as follows.
5-9
Planning your hardware components
High-availability options, deskside unshared storage
Deskside system type HA level
DPE Minimum iDAE
DAE only
30-slot
SCSI-disk
Maximum 2
Minimum
Maximum 2
Minimum
Maximum 2
Minimum
PDUs SPs LCCs
1 1 1 DPE
1 DAE
1
1 n/a
Maximum n/a
2
1
2 n/a n/a
1
2
4 (2 DPE
2 DAE) n/a (10-slot)
2 (30-slot) n/a (10-slot)
4 (30-slot)
1
2 n/a n/a
Power supplies
1 DPE
1 DAE
4
(2 DPE
2 DAE)
1
2 (10-slot)
6 (30-slot)
1
2
1 (called a
VSC)
3 (called
VSCs)
Disks
5 (without write cache)
10 (write cache or
RAID 3)
3 (without write cache)
5 (write cache or
RAID 3)
5 (write cache or
RAID 3)
No minimum
No minimum
5 (write cache or
RAID 3)
5 (write cache or
RAID 3)
SPS units
0 (without write cache)
1 (write cache)
2
0 (without write cache)
1 (write cache
2 n/a n/a
0 (without write cache)
1 (write cache)
1
For rackmount systems, the standby power supply or supplies (SPS or BBU) must be placed in a tray directly beneath the storage system. Typically, any hubs in the cabinet mount at the top or bottom of the cabinet. Rackmount options are as follows.
5-10
014-002912-01
Hardware data sheets
Rackmount system type HA level
DPE Minimum iDAE
DAE only
30-slot
SCSI-disk
High-availability options, rackmount unshared storage
Minimum
Minimum
Minimum
SPs LCCs
1 1
Maximum 2
1
Maximum 2 n/a
Maximum n/a
1
Maximum 2
2 (DPE)
22 (with 11
DAEs) n/a (10-slot)
4 (with two
DAEs) n/a
1
2 n/a n/a
Power supplies
1
2 (DPE)
22 (with 11
DAEs)
1
2
1
2
1 (called a
VSC)
3 (called
VSCs)
Disks
5 (without write cache)
SPS units
0 (without write cache)
1 (write cache)
2 10 (write cache or
RAID 3)
3 (without write cache)
5 (write cache or
RAID 3)
5 (write cache or
RAID 3)
No minimum
No minimum
5 (write cache or
RAID 3)
5 (write cache or
RAID 3)
0 (without write cache)
1 (write cache
2 n/a n/a
0 (without write cache)
1 (write cache)
1
Hardware data sheets
The hardware data sheets shown in this section provide the plant requirements, including dimensions (footprint), weight, power requirements, and cooling needs, for DPE, iDAE, DAE, and 30-slot SCSI disk systems. Sections on cabinets and cables follow the data sheets.
DPE data sheet
For shared storage, a rackmount DPE and one or more rackmount DAEs are required. For unshared storage, you can use a rackmount or deskside DPE and DAE(s). The DPE dimensions and requirements are shown in the following figure.
014-002912-01
5-11
Hardware data sheets
DPE dimensions and requirements
Depth
74.7 cm
(30 in)
Deskside model
Width
52.1 cm
(20.6 in)
Depth
70 cm
(27.6 in)
Rackmount model
Width
44.5 cm
(17.5 in)
Height
68 cm
(26.8 in)
Height
28.6 cm
(11.3 in)
6.5 U
SPS mounting tray, height 4.44 cm
(1.75 in), 1 U; depth 69.9 cm
(27.5 in)
Weight (without packaging)
Maximum (max disks, SPs, LCCs, PSs): with 2 SPSs
Deskside
144 kg (316 lb)
165 kg (364 lb)
Rackmount
52 kg (115 lb)
74 kg (163 lb)
Power requirements
Voltage rating:
Current draw:
Power consumption:
100 V ac to 240 V ac –10%/+15%, single-phase, 47 Hz to 63 Hz; power supplies are auto-ranging
At 100 v ac input – Deskside DPE/DAE: 12.0 A;
Rackmount DPE: 8.0 A max
SPS: 1.0 A max per unit during charge
Deskside DPE/DAE: 1200 VA; Rackmount DPE: 800 VA max
SPS: 100 VA per unit during charge
Power cables (single or dual) ac inlet connector:
Deskside power cord:
IEC 320-C14 power inlet
USA:
Outside USA:
1.8 m (6.0 ft): NEMA 6-15P plug
Specific to country
Operating environment
Temperature:
Relative humidity:
Altitude:
10 o
C to 40 o
C (50 o
F to 104 o
Noncondensing, 20% to 80%
F)
40 o
C to 2,438 m (8,000 ft); 37 o
C to 3,050 m (10,000 ft)
Heat dissipation (max): Deskside DPE/DAE: 3931x10
3
J/hr (2730 BTU/hr) max estimated;
Air flow:
Rackmount DPE: 2520x10
3
Front to back
J/hr (2390 BTU/hr) max estimated
Service clearances
Front:
Back:
30.3 cm (1 ft)
60.6 cm (2 ft)
5-12
014-002912-01
Hardware data sheets
iDAE data sheet
You can use a rackmount or deskside DPE and DAE(s) for unshared storage.
The iDAE dimensions and requirements are shown in the following figure.
Dimensions and requirements, iDAE
Deskside 30-slot model
Depth
74.7 cm
(30 in)
Width
52.1 cm
(20.6 in)
Deskside 10-slot model
Depth
74.7 cm
(30 in)
Width
25 cm
(9.8 in)
Rackmount model
Depth
63.3 cm
(24.9 in)
Width
44.5 cm
(17.5 in)
Height
68 cm
(26.8 in)
SPS mounting tray, height 4.44 cm
(1.75 in), 1 U; depth 69.9 cm
(27.5 in)
Height
15.4 cm
(6.1 in)
3.5 U
Weight (without packaging)
Maximum (max disks, SPs):
Power requirements
Voltage rating:
Current draw
Power consumption:
Deskside 30
144 kg (316 lb)
Deskside 10
60 kg (132 lb)
Rackmount
35.4 kg (78 lb)
100 V ac to 240 V ac +/- 10%, single-phase, 47 Hz to 63 Hz; power supplies are auto-ranging
At 100 v ac input – 30-slot 12.0 A; 10-slot: 4.0 A;
SPS: 1.0 A max per unit during charge
30-slot: 1200;10-slot: 400 VA;
SPS: 100 VA per unit during charge
Power cables (single or dual) ac inlet connector:
Deskside power cord:
IEC 320-C14 power inlet
USA:
Outside USA:
1.8 m (6.0 ft): NEMA 6-15P plug
Specific to country
Operating environment
Temperature:
Relative humidity:
10 o
C to 40 o
C (50 o
F to 104 o
Noncondensing, 20% to 80%
F)
Altitude: 40 o
C to 2,438 m (8,000 ft); 37 o
C to 3,050 m (10,000 ft)
Heat dissipation (max): 30-slot: 4,233 KJ/hr (4,020 BTU/hr)
Air flow:
10-slot: 1,411 KJ/hr (1,340 BTU/hr)
Front to back
Service clearances
Front:
Back:
30.3 cm (1 ft)
60.6 cm (2 ft)
014-002912-01
5-13
Hardware data sheets
DAE data sheet
The DAE storage-system dimensions and requirements are shown in the following figure.
Dimensions and requirements, DAE
Deskside 30-slot model Deskside 10-slot model
Depth
74.7 cm
(30 in)
Width
52.1 cm
(20.6 in)
Depth
74.7 cm
(30 in)
Width
25 cm
(9.8 in)
Rackmount model
Depth
63.3 cm
(24.9 in)
Width
44.5 cm
(17.5 in)
Height
68 cm
(26.8 in)
Height
15.4 cm
(6.1 in)
3.5 U
Weight(without packaging)
Maximum configuration
Deskside 30 Deskside 10
143.6 kg (316 lb) 60 kg (132 lb)
Rackmount
35.4 kg (78 lb)
Power requirements
Voltage rating:
Current draw:
Power consumption:
100 V ac to 240 V ac -10%/+15%, single-phase, 47 Hz to 63 Hz; power supplies are auto-ranging
At 100 V: 30-slot: 12 A; 10-slot 4 A max
30-slot: 1200 VA; 10-slot 400 VA per supply max
Power cables (single or dual) ac inlet connector:
Deskside power cord:
IEC 320-C14 power inlet
USA:
Outside USA:
1.8 m (6.0 ft): NEMA 6-15P plug
Specific to country
Operating environment
Temperature:
Relative humidity:
10 o
C to 40 o
C (50 o
F to 104 o
Noncondensing, 20% to 80%
F)
Altitude: 40 o
C to 2,438 m (8,000 ft); 37 o
C to 3,050 m (10,000 ft)
Heat dissipation (max): 30-slot: 4,233 KJ/hr (4,020 BTU/hr)
10-slot: 1,411 KJ/hr (1,340 BTU/hr)
Service clearances
Front:
Back:
30.3 cm (1 ft)
60.6 cm (2 ft)
5-14
014-002912-01
Hardware data sheets
30-slot system with SCSI disks data sheet
The 30-slot disk-array storage system dimensions and requirements are shown in the following figure.
Dimensions and requirements, 30-slot SCSI-disk storage system
Rackmount model
Depth
Deskside model
76.2 cm
(30.0 in)
Width
48.9 cm
(19.3 in)
Depth
76.2 cm
(30.0 in)
Width
48.3 cm
(19 in)
Height
62.9 cm
(24.8 in)
Height
48.9 cm
(19.3 in)
11 U
Weight(without packaging)
Maximum configuration:
Power requirements
Voltage rating:
Current draw:
Power consumption:
Deskside
119 kg (261.5 lb)
Rackmount
98 kg (216 lb)
200 V ac to 240 V ac -10%/+15%, single-phase, 47 Hz to 63 Hz; power supplies are auto-ranging
At 5.0 maximum at 200 Vac input
1000 VA maximum (apparent)
Power cables (single or dual) ac inlet connector: IEC 320-C14 power inlet
Deskside power cord: USA: 1.8 m (6.0 ft): NEMA 6-15P plug
Outside USA:: Specific to country
Operating environment
Temperature: 10 o
C to 38 o
C (50 o
F to 100 o
F)
Relative humidity:
Altitude:
Noncondensing, 20% to 80%
2,438 m (8,000 ft)
Heat dissipation (max): 300 BTU/hour
Air flow: Front to back
Service clearances
Front: 30.3 cm (1 ft)
Back: 60.6 cm (2 ft)
014-002912-01
5-15
Cabinets for rackmount enclosures
Cabinets for rackmount enclosures
Prewired 19-inch-wide cabinets, ready for installation, are available in the following dimensions to accept rackmount storage systems.
Vertical space Exterior dimensions Comments
173 cm or 68.25
in (39 NEMA units or U; one
U is 1.75 in)
Height: 192 cm (75.3 in)
Width: 65 cm (25.5 in)
Depth: 87 cm (34.25 in) plus service clearances, which are 90 cm (3 ft), 30 cm front and 60 cm back
Accepts combinations of:
DPEs at 6.5 U, iDAEs at 3.5 U,
SPS units at 1 U,
DAEs at 3.5 U each,
30-slot SCSI disk systems at 11 U
Switches or hubs at 1 U
Weight (empty): 134 kg (296 lb)
Requires 200–240 volts ac. Plug options include L6–30 or L7–30
(domestic) and IEC 309 30 A
(international).
Each power strip has 12 IEC-320
CIS outlets.
Filler panels of various sizes are available.
As an example, a rackmount storage system that supports 100 disk modules has the following requirements.
Category
Vertical cabinet space in NEMA units (U, one
U is 1.75 in)
Weight
Power
Cooling
Requirement
Bottom to top: One SPS (1 U), one DPE (6.5 U), and nine
DAEs (9*3.5 U equals 31.5 U) for a total of 39 U.
516 kg (1,137 lb) including the cabinet (134 kg), DPE (52 kg),
SPS (11 kg), and nine DAEs (9 * 35.4 kg equals 319 kg).
4,500 VA max, including the DPE (800 VA), SPS (100 VA), and nine DAEs (9 * 400 VA equals 3600 VA).
15,484 KJ/hour (14,700 BTU/hr), including the DPE (2,520
KJ/hr), SPS (265 KJ/hour, estimated), and nine DAEs (9*1,411
KJ/hr equals 12,699 KJ/hr).
Cable and configuration guidelines
We recommend that all copper-interconnected nodes be connected to a common ground grid. The common grid is not needed for optical interconnections.
Copper cable allows up to 30 meters (99 feet) between nodes or hubs. Optical cable allows significantly longer distances. This is a major advantage of optical cable. However, you can use optical cable from a server only if the server’s adapter supports optical cable; otherwise you must use copper. Not all adapters support optical cable.
5-16
014-002912-01
014-002912-01
Cable and configuration guidelines
To connect a DPE to a DAE, you must use copper cable, whose maximum length is 10 meters (33 ft). So, the distance between a DPE and the DAEs it controls cannot exceed 10 meters (33 ft).
The HBAs and SPs used with shared storage systems require optical cable, as does the switch between the HBAs and SPs. An optical GBIC interface converter, which is required for optical cable, is included with each SP and
HBAs, but are not included with a switch; you must specifically order an optical GBIC for every switch port you will use.
The SPs used with unshared storage systems support copper cables and — with MIAs — optical cables. The hub itself supports copper, or with a MIA, optical. So you can use a copper cable or — with two MIAs per cable — optical cable between any hub and SP. For optical cable to work between an
HBA and hub or SP, the HBA must support optical cable.
Server
Switch or hub
Maximum distance:
Copper cable: 30 m
Optical cable:500 m or more
Maximum distance:
Copper cable: 30 m
Server
Storage system Storage system
You can use any existing FDDI, multimode, 62.5 micron cable with good connections to attach servers, hubs, and storage systems. These cables must be dedicated to storage-system I/O.
5-17
Cable and configuration guidelines
Cable sizes — Optical
Length
5 m (16.5 ft) or
10 m (33 ft)
50 m (164 ft)
Typical use
Within one room, connecting servers to storage systems (adapter must support optical cable) or connecting switches or hubs to storage systems
Within one building, connecting servers to storage systems (adapter must support optical cable) or connecting switches or hubs to storage systems
100 m (328 ft)
250 m (821 ft,.15 mi)
Within one complex, connecting servers to storage systems (adapter must support optical cable) or connecting switches or hubs to storage systems
500 m (1642 ft,.31 mi)
Optical cabling is 50 micron (maximum length is 500 m (1,650 ft) or
62.5 micron (maximum length is 300 m (985 ft). Both types are multimode, dual SC, and require a MIA on a DB-9 or hub connector. The minimum length is 2 m (6.8 ft).
The minimum bend radius is 3 cm (1.2 in).
Cable sizes — Copper
Length
0.3 m (1 ft), non-equalized
Typical use
Connecting DPE/DAE and DAE LCCs
1.0 m (3.3 ft), non-equalized Connecting a hub to an adjacent storage system
3 m (10 ft), non-equalized
5 m (16.5 ft), non-equalized
Connecting a hub to a storage system in the same cabinet, or daisy chaining from one cabinet to an adjacent cabinet
Connecting a hub in one rack to a storage system in another cabinet
10 (33 ft), non-equalized Connecting servers to hubs and/or storage systems
— maximum length for non-equalized copper cable, maximum length between LCCs
30 m (98.5 ft), equalized Connecting servers to hubs and/or storage systems
– maximum length for copper cable
Copper cabling is shielded, 75 ohm twin-axial, shield bonded to DB-9 plug connector shell (360 o
) FC-AL Standard, Revision 4.4 or higher.
Component planning diagrams and worksheets follow.
5-18
014-002912-01
Hardware planning worksheets
Hardware planning worksheets
Following are worksheets to note the hardware components you want.
There are two types of configuration:
• Shared storage
• Unshared storage
Hardware for shared storage
Cable identifier — DPE-based system for shared storage
Server 1
A1 A1
Server n
. . .
An An
Switch 1 Switch 2
E2 E2
DAE
D1
E1 DAE
DPE
SP B SP A
Storage system 1
E1 D1
Dm
E2
E1
DAE
E2
E1 Dm
DAE
Path 1
Path 2
DPE
SP B SP A
Storage system m
The cable identifiers used above apply to shared and unshared storage systems. The worksheet applies to shared storage only.
014-002912-01
5-19
Hardware planning worksheets
Hardware component worksheet for shared storage
Number of servers:____ Adapters in servers:____ Switches: 16-port:____8-port:____ GBICs (1 per port):___
Rackmount DPEs:_____SP/LCC pairs:_____PSs:_____SPSs:____ Rackmount cabinets:___
Rackmount DAEs:_____ LCCs:_____PSs:_____
Cables between server and switch - Cable A, optical only
Cable A
1,
Cable A
2,
Optical: Number:__ __ ...................................................... ..............Length________m or ft
Optical: Number:__ __ ...................................................... ..............Length________m or ft
Cable A n
,Optical: Number:__ __ ....................................................... ..............Length________m or ft
Cables between switches and storage systems - Cable D, copper or optical
Cable D
1,
Cable D
2,
Cable D m,
Optical:Number:_____ ...................................................... ..............Length________m or ft
Optical:Number:_____ ...................................................... ..............Length________m or ft
Optical:Number:_____ ..................................................... ..............Length________m or ft
Cables between enclosures - Cable E, which connects LCCs; between a DPE LCC and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:_____ V Copper V Optical (for DAE to DAE only).. ..............Length________m or ft
Cable E
2
:Number:_____ V Copper V Optical (for DAE to DAE only).. ..............Length________m or ft
5-20
014-002912-01
Hardware planning worksheets
Sample shared storage installation and worksheet
Highly available cluster
File Server Mail Server Database Server
Cable between switch and storage system - D1
Path 1
Path 2
Switch
DAE
DAE
DAE
DAE
DAE
DAE
DPE
SP B SP A
Storage system
Switch
Cable between server and switch - A2
= Cable between storage
systems or enclosures - E1
Hardware component worksheet for shared storage
Number of servers:__
3
_ Adapters in servers:__
6
_ Switches: 16-port:____8-port:__
2
__ GBICs (1 per port):_
6
__
Rackmount DPEs:__
1
___SP/LCC pairs:__
2
___PSs:___
2
__SPSs:__
2
__ Rackmount cabinets:_
1
__
Rackmount DAEs:__
6
___ LCCs:___
12
__PSs:___
12
__
Cables between server and switch - Cable A, optical only
Cable A
1,
Optical: Number:__
2
__ .................................................... ..............Length____
33
_m or ft
Cable A
2,
Optical: Number:__
4
__ .................................................... ..............Length__
1628
_m or ft
Cable A n
,Optical: Number:__ __ ....................................................... ..............Length________m or ft
Cables between switches and storage systems - Cable D, copper or optical
Cable D
1,
Optical:Number:__
2
___ ................................................... ..............Length____
33
__m or ft
Cable D
2,
Optical:Number:_____ ...................................................... ..............Length________m or ft
Cable D m,
Optical:Number:_____ ..................................................... ..............Length________m or ft
Cables between enclosures - Cable E, which connects LCCs; between a DPE LCC and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:__
12
___ V Copper V Optical(for DAE to DAE only) ..........Length____
1
___m or ft
Cable E
2
:Number:_____ V Copper V Optical (for DAE to DAE only).. ..............Length________m or ft
014-002912-01
5-21
Hardware planning worksheets
Hardware for unshared storage
The cable identifiers used below and on the following worksheets apply to all types of unshared storage systems. So, if you want to plan a site with different types of systems, you can consolidate all your unshared storage component entries (from the different system types on a single worksheet).
Cable identifier — unshared full-fibre system without hubs
Server 1
A1 A1
Server 2
A2 A2
E2
E1
DAE
DAE iDAE/DPE
SP B SP A
E2
E1
FC loop 1
FC loop 2
Cable identifier — unshared full-fibre system with hubs
Server 1
A1 A1
Server n
. . .
An An
5-22
Hub 1
E2 E2
DAE
FC loop 1
FC loop 2
Dm
D1
E1 DAE iDAE/DPE
SP B SP A
Storage system 1
E2
E1
DAE
DAE iDAE/DPE
SP B SP A
Storage system m
E1
E2
E1
D1
Dm
Hub 2
014-002912-01
Hardware planning worksheets
Hardware component worksheet for unshared storage
Number of servers: Adapters in servers:_______ Hubs (copper):______ MIAs (copper to optical):_______
DPE-based and DAE-only storage systems:
Rackmount DPEs:_____SP/LCC pairs:_____
Rackmount iDAEs:_____SPs:_____
PSs:_____SPSs:____ Rackmount cabinets:___
PSs:_____SPSs:____ Rackmount cabinets:___
Rackmount DAEs:_____ LCCs:_____PSs:_____
Deskside DPEs:_____SP/ LCC pairs:_____DAE LCCs:____ DPE PSs:____ DAE PSs:____ SPSs:__
Deskside iDAEs:_____SPs:_____ DAE LCCs:____ PSs:____ SPSs:___
Deskside DAEs: 30-slot_____ 10-slot______ LCCs:_____ PSs:_____
30-slot SCSI-disk storage systems:
Rackmount:_______SPs:_______VSCs:_____BBUs:____Rackmount Cabinets:___
Deskside:_______SPs:_______ VSCs:_____BBUs:____
Cables between server and storage system or between server and hub - Cable A, copper or optical
Cable A
1
:Number:__ __ V Copper V Optical .................................. ..............Length________m or ft
Cable A
2
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cable A n
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cables between hubs and storage systems - Cable D, copper or optical
Cable D
1
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D
2
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D m
:Number:_____ V Copper V Optical .................................... ..............Length________m or ft
Cables between storage systems or enclosures - Cable E, which connects LCCs or SP-LCC; between a DPE
LCC or iDAE SP and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:_____ V Copper V Optical(DAE to DAE only)........ ..............Length________m or ft
Cable E
2
:Number:_____ V Copper V Optical(DAE to DAE only)........ ..............Length________m or ft
*Please specify all storage-system components you need, even though you will not need to order them separately, since most or all components will be included with the model of each system you order.
014-002912-01
5-23
Hardware planning worksheets
Sample unshared deskside system — basic configuration
Server
DAE iDAE/DPE
SP A
FC loop 1
Sample component worksheet
Cable between server and storage system - A1
Cable between storage systems or enclosures - E1 - Included with deskside DPE
Component worksheet for any storage system
Number of servers:
1
Adapters in servers:___
1
___ Hubs (copper):______ MIAs (copper to optical):_____
DPE-based and DAE-only storage systems:
Rackmount DPEs:_____SP/LCC pairs:_____...... PSs:_____SPSs:____ Rackmount cabinets:___
Rackmount iDAEs:__
1
___SPs:__
1
___ ............. PSs:___
1
__SPSs:____ Rackmount cabinets:__
1
_
Rackmount DAEs:___
1
__ LCCs:__
1
___PSs:__
1
___
Deskside DPEs:_____SP/ LCC pairs:_____DAE LCCs:____ DPE PSs:____ DAE PSs:____ SPSs:__
Deskside iDAEs:_____SPs:_____ DAE LCCs:____ PSs:____ SPSs:___
Deskside DAEs: 30-slot_____ 10-slot______ LCCs:_____ PSs:_____
Cables between server and storage system or between server and hub - Cable A, copper or optical
Cable A
1
:Number:__
1
__ V Copper V Optical ................................. ..............Length___
10
_____m or ft
Cable A
2
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cable A n
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cables between hubs - Cable C, copper or optical (maximum 1 between hubs)
Cable C:Number:____ V Copper V Optical......................................... ..............Length________m or ft
Cables between hubs and storage systems - Cable D, copper or optical
Cable D
1
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D
2
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D m
:Number:_____ V Copper V Optical .................................... ..............Length________m or ft
Cables between storage systems or enclosures - Cable E, which connects LCCs or SP-LCC; between a DPE
LCC or iDAE SP and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:__
1
__ V Copper V Optical(DAE to DAE only)....... ..............Length__
1
_____m or ft
Cable E
2
:Number:_____ V Copper V Optical (DAE to DAE only)........ ..............Length________m or ft
*Please specify all storage-system components you need, even though you willl not need to order them separately, since most or all components will be included with the model of each system you order.
5-24
014-002912-01
Hardware planning worksheets
Sample unshared deskside system
—
dual-adapter/dual-SP configuration
Server
Cables between server and storage system - A1
Cables between storage systems or enclosures - E1 - included with deskside DPE
LCC LCC LCC LCC LCC
Cables between storage systems or enclosures - E2
LCC LCC
Sample component worksheet
LCC LCC LCC
FC loop 1
FC loop 2
Number of servers:
Hardware component worksheet for unshared storage
1
Adapters in servers:__
2
_____ Hubs (copper):______ MIAs (copper to optical):_______
DPE-based and DAE-only storage systems:
Rackmount DPEs:_____SP/LCC pairs:_____
Rackmount iDAEs:_____SPs:_____
PSs:_____SPSs:____ Rackmount cabinets:___
PSs:_____SPSs:____ Rackmount cabinets:___
Rackmount DAEs:_____ LCCs:_____PSs:_____
Deskside DPEs:__
1
_SP/ LCC pairs:__
2
__DAE LCCs:__
2
__ DPE PSs:__
2
__ DAE PSs:__
2
__ SPSs:_
1
_
Deskside iDAEs:_____SPs:_____
Deskside DAEs: 30-slot__
DAE LCCs:____ PSs:____ SPSs:___
1
___ 10-slot______ LCCs:__
6
___ PSs:__
6
__
Cables between server and storage system or between server and hub - Cable A, copper or optical
Cable A
1
:Number:_
2
_ __ V Copper V Optical ............................... ..............Length___
10
___m or ft
Cable A
2
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cable A n
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cables between hubs and storage systems - Cable D, copper or optical
Cable D
1
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D
2
:Number:_____ V Copper V Optical..................................... ..............Length________m or ft
Cable D m
:Number:_____ V Copper V Optical .................................... ..............Length________m or ft
Cables between storage systems or enclosures - Cable E, which connects LCCs or SP-LCC; between a DPE
LCC or iDAE SP and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:__
Cable E
2
:Number:__
6
___ V Copper V Optical(DAE to DAE only)..... ..............Length___
1
_____m or ft
2
___ V Copper V Optical(DAE to DAE only)..... ..............Length___
5
_____m or ft
014-002912-01
5-25
Hardware planning worksheets
Sample unshared DPE-based system with hubs — two loops
Cable between server and hub - A1
Server 1
Cable between server and hub - A1
Cable between server and hub - A2
Server 2
Cable between server and hub - A2
Cable between hub and storage system - D1
Hub 1
Cable between hub and storage system - D2
FC loop 1
FC loop 2
Hub 2
Storage system 1
DAE
DAE
DAE
DAE
DPE
SP B SP A
Cable between hub and storage system - D1
Cable between hub and storage system - D2
Storage system 2
DAE
DAE
DAE
DAE
DPE
SP B SP A
= Cable between storage systems or enclosures - E1
5-26
014-002912-01
What next?
Sample component worksheet for DPE-based system with hubs — two loops
Component worksheet for any storage system
Number of servers:
2
Adapters in servers:___
4
____ Hubs:___
2
____ MIAs (optical to copper):_______
DPE-based and DAE-only storage systems:
Rackmount DPEs:_____SP/LCC pairs:__
4
___ PSs:___
4
__SPSs:_
2
__ Rackmount cabinets:_
2
__
Rackmount iDAEs:_____SPs:_____
Rackmount DAEs:__
8
___
PSs:_____SPSs:____ Rackmount cabinets:___
LCCs:__
16
_PSs:__
16
___
Deskside DPEs:_____SP/ LCC pairs:_____DAE LCCs:____ DPE PSs:____ DAE PSs:____ SPSs:__
Deskside iDAEs:_____SPs:_____ DAE LCCs:____ PSs:____ SPSs:___
Deskside DAEs: 30-slot_____ 10-slot______ LCCs:_____ PSs:_____
30-slot SCSI-disk storage systems:
Rackmount:_______SPs:_______VSCs:_____BBUs:____Rackmount Cabinets:___
Deskside:_______SPs:_______ VSCs:_____BBUs:____
Cables between server and storage system or between server and hub - Cable A, copper or optical
Cable A
1
:Number:__
Cable A
2
:Number:__
2
__ V Copper V Optical ................................ ..............Length_
2 0
_____m or ft
2
__ V Copper V Optical................................. ..............Length_
10
_____m or ft
Cable A n
:Number:__ __ V Copper V Optical ................................... ..............Length________m or ft
Cables between hubs and storage systems - Cable D, copper or optical
Cable D
1
:Number:__
Cable D
2
:Number:__
2
___ V Copper V Optical .................................. ..............Length__
20
____m or ft
2
___ V Copper V Optical .................................. ..............Length___
2 0
____m or ft
Cable D m
:Number:_____ V Copper V Optical .................................... ..............Length________m or ft
Cables between storage systems or enclosures - Cable E, which connects LCCs or SP-LCC; between a DPE
LCC or iDAE SP and a DAE LCC, Cable E must be copper; between DAE LCCs, it can be copper or optical.
Cable E
1
:Number:_
16
__ V Copper V Optical(DAE to DAE only)....... ..............Length___
1
____m or ft
Cable E
2
:Number:_____ V Copper V Optical (DAE to DAE only)........ ..............Length________m or ft
Please specify all storage-system components you need, even though you will not need to order them separately, since most or all components will be included with the model of each system you order.
What next?
This chapter explained hardware components of shared and unshared storage systems. If you have completed the worksheets to your satisfaction, you are ready to consider ordering some of this equipment. Or you may want to read about storage management in the next chapter.
014-002912-01
5-27
6
Storage-system management
This chapter explains the management applications you can use to manage storage systems from servers. Topics are
•
Managing shared or unshared storage systems using Navisphere ®
Manager software
•
Managing unshared storage systems using Navisphere Supervisor software
•
Monitoring DAE-only storage systems (JBODs)
•
Storage management worksheets
Navisphere software lets you bind and unbind disks, manipulate caches, examine storage-system status and logged events, transfer control from one
SP to another, and examine events recorded in storage system event logs.
Navisphere products have two parts: a graphical user interface (GUI) and an Agent. The GUIs run on a management station, accessible from a common framework, and communicate with storage systems through a single Agent application that runs on each server. The management station and server are often separate systems, but they need not be; that is, you can run one or more GUIs and an agent on one server. The Navisphere products are
• Navisphere Manager, which lets you manage multiple storage systems on multiple servers simultaneously; Navisphere Manager is required for shared storage and optional for unshared storage.
• Navisphere Supervisor software, which lets you manage one storage system on one server at a time, which is included with each storage system. Navisphere Supervisor supports unshared storage systems only.
• Navisphere Analyzer, which lets you measure, compare, and chart the performance of SPs, LUNs, and disks.
• Navisphere Integrator, which provides an interface between Navisphere products and HP OpenView and Tivoli.
• Navisphere Event Monitor, which checks storage systems for fault conditions and can notify you and/or customer service if any fault condition occurs.
• Navisphere Agent, which is included with each storage system, and
Navisphere CLI (Command Line Interpreter), which lets you bypass the
GUIs and issue commands directly to storage systems.
6-1
014-002912-01
Managing shared or unshared storage systems using Navisphere Manager software
Managing shared or unshared storage systems using
Navisphere Manager software
Navisphere Manager software (Manager) lets you manage multiple storage systems connected to servers on a TCP/IP network. Manager offers extensive management features and includes comprehensive on-line help.
Manager is required for shared storage and optional for unshared storage.
Manager runs on a management station which is a Windows NT ® host. The servers connected to a storage system can run Windows NT or one of several
UNIX ® operating systems. With shared storage, servers connected to the
SAN can run different operating systems; with unshared storage, servers connected to the same storage system must run the same operating system.
Another software component for shared storage or other high-availability storage systems is Application-Transparent Failover (ATF). ATF is a software product that works with shared storage or other high-availability storage systems to let applications continue running after the failure in the path to a LUN. If a host-bus adapter, switch, hub, SP, or cable fails, ATF can automatically route I/O to the LUNs the applications need.
The following figures show Navisphere Manager in shared and unshared environments.
Sample shared environment (SAN) with Navisphere Manager
File Server
Management station and server
Operating system A
Navisphere
Manager
Navisphere
Agent
ATF
Mail Server
Management station and server
Operating system A
Navisphere
Manager
Navisphere
Agent
ATF
Switch
Database Server
Operating system B
Navisphere
Agent
ATF
Production Server
Operating system C
Navisphere
Agent
ATF
Network
Switch
6-2
014-002912-01
Managing unshared storage systems using Navisphere Supervisor software
Sample unshared environment with Navisphere Manager
Accts Server
Management station and server
Operating system A
Navisphere
Supervisor
Navisphere
Agent
Database Server
Operating system B
Navisphere
Agent
Development Server
Management station and server
Operating system C
Navisphere
Supervisor
Navisphere
Agent
Network
Managing unshared storage systems using Navisphere
Supervisor software
Navisphere Supervisor software, like Manager, consists of a graphical user interface, called Supervisor, and a storage-system agent, called the
Navisphere Agent — the same Agent used by Manager. However,
Supervisor software lets you manage only one storage system at a time.
The sample management environment below shows a Windows NT server that is both a manager and an agent. You can use this server to manage its own storage system and/or to manage the storage systems on the other server with agents.
Sample unshared environment with Navisphere Supervisor
Accts Server
Management station and server
Operating system A
Navisphere
Supervisor
Navisphere
Agent
Database Server
Operating system B
Navisphere
Agent
Development Server
Management station and server
Operating system C
Navisphere
Supervisor
Navisphere
Agent
Network
014-002912-01
6-3
Monitoring DAE-only storage systems (JBODs)
Monitoring DAE-only storage systems (JBODs)
With a DAE-only (JBOD) system, you can monitor status with Navisphere
Manager or Supervisor. To create LUNs of different RAID types, you must use the software RAID utility designed for the operating system.
Storage management worksheets
This section includes two worksheets: one for shared storage and one for unshared storage. The following worksheet will help you plan your storage system management environment. For each host, complete a section.
For the shared storage worksheet, complete the management station hostname and operating system; then decide whether you want the
Navisphere Analyzer and/or Event Monitor and, if so, mark the appropriate boxes. Then write the name of each managed server, with operating system, storage group, and configuration access specification. You can copy much of the needed information from the LUN and Storage Group planning
Management Utility Worksheet – Shared Storage
Management station hostname:___________________Operating system:_______________
Software: V Navisphere Manager/Navisphere Agent V Navisphere Analyzer V Navisphere Event Monitor
List all the servers this host will manage. Each managed server must run an Agent and ATF software of the same type as its operating system.
Server: Op sys:
Server:
Server:
Server:
Server:
Server:
Server:
Server:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
Management station hostname:___________________Operating system:_______________
Software: V Navisphere Manager/Navisphere Agent V Navisphere Analyzer V Navisphere Event Monitor
List all the servers this host will manage. Each managed server must run an Agent and ATF software of the same type as its operating system.
Server: Op sys:
Server:
Server:
Server:
Server:
Server:
Server:
Server:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Op sys:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
Storage Group number or name:
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
V Config Access
6-4
014-002912-01
Storage management worksheets
For unshared storage, for each host, choose a Navisphere product. The host may be a management station that is not a server (complete only the
Manager section); it may be a management station that is a server
(complete the Manager section and mark the Agent box), or it may be a server (mark the Agent box).
Management Utility Worksheet – Unshared Storage
Hostname: Operating system:
Storage system type: V DPE-based V iDAE-based V 30-slot SCSI disk
Software: V Navisphere Manager and Navisphere Agent V Navisphere Supervisor and Navisphere Agent
List all the servers this host will manage. Each managed server must run an Agent of the same type as its operating system.
Server: Oper sys: Server: Oper sys:
Server:
Server:
Server:
Oper sys:
Oper sys:
Oper sys:
Server:
Server:
Server:
Oper sys:
Oper sys:
Oper sys:
Hostname: Operating system:
Storage system type: V DPE-based V iDAE-based V 30-slot SCSI disk
Software: V Navisphere Manager and Navisphere Agent V Navisphere Supervisor and Navisphere Agent
List all the servers this host will manage. Each managed server must run an Agent of the same type as its operating system.
Server: Oper sys: Server: Oper sys:
Server:
Server:
Server:
Oper sys:
Oper sys:
Oper sys:
Server:
Server:
Server:
Oper sys:
Oper sys:
Oper sys:
Hostname: Operating system:
Storage system type: V DPE-based V iDAE-based V 30-slot SCSI disk
Software: V Navisphere Manager and Navisphere Agent V Navisphere Supervisor and Navisphere Agent
List all the servers this host will manage. Each managed server must run an Agent of the same type as its operating system.
Server: Oper sys: Server: Oper sys:
Server:
Server:
Oper sys:
Oper sys:
Server:
Server:
Oper sys:
Oper sys:
Server: Oper sys: Server: Oper sys:
014-002912-01
6-5
Index
Within this index, a range of page numbers indicates that the reference spans those pages.
Numerics
30-slot SCSI-disk storage systems
disk structure example
physical disk structure example
site requirements 5-15 weight 5-15
A ac power requirements
30-slot SCSI-disk systems 5-15
application planning
application worksheet, completing
applications, sample. for RAID groups 2-14
application-transparent failover (ATF)
disk-array storage system
interface kit
B basic configuration
C
cabinets for rackmount storage systems 5-16
cabling
cache
communication with storage system,
configurations
error recovery features
LUN and file system
RAID
server and storage system
shared storage
storage system
basic 4-2 dual-adapter/dual-SP 4-2
tradeoffs
unshared storage
cooling requirements, cabinet 5-16
copper cable, types and sizes 5-18
CRUs (customer-replaceable units) locating 5-6
D
DAE-only storage systems
site requirements 5-14 weight 5-14
device name, operating system 3-13, 4-18
disk configuration types
see RAID group
shared storage
014-002912-01
Index-1
Index disk (continued)
Disk Array Enclosure (DAE)
disk module(s)
IDs
full-fibre system 3-2, 4-5
disk unit number on worksheet 3-6, 3-11,
disk-array storage system
types
DPE storage systems
dimensions 5-12 site requirements 5-12 weight 5-12
dual-adapter/dual-SP configuration
dual-initiator/dual-loop configuration
disk structure example 4-6–4-7
dual-ported disk
as function of dual SPs 5-3, 5-7
dual-server configurations
disk structure example 4-6–4-7
E enclosure address (EA)
error recovery features
F
Fibre Channel
file system
worksheet
footprint
30-slot SCSI-disk systems 5-15
G
GBIC (Gigabit Interface Converter), about 1-4
see hot spare
GUI (in storage-system management utilities) 6-2
H hardware
planning worksheets
heat dissipation
30-slot SCSI-disk systems 5-15
height
30-slot SCSI-disk systems 5-15
high availability options
high-availability features, see
error recovery features
server host-bus adapter (HBA)
hot spare
hub
sample hardware worksheet 5-26
I iDAE storage systems
dimensions 5-13 site requirements 5-13 weight 5-13
RAID 5 group
Index-2
014-002912-01
individual disk unit
cables, hubs, switches 1-3 interface kit 1-3
L
LCC (link control card)DPE storage system 5-6
see file system
LUN (logical unit) configurations
individual disk
RAID 0
RAID 1/0 group
RAID 3 group
RAID 5 group
shared storage
unshared storage
worksheets 3-9–3-11, 4-14–4-16
number on worksheet 3-6, 3-11, 4-10,
M
MIA (media interface adapter), about 1-3
see RAID 1 mirrored pair
RAID 1/0 group
Index
N
Navisphere Manager utility 6-1
see RAID 0 group
O operating system
device name for disk unit 3-13, 4-18
optical cable, types and sizes 5-18
P
RAID 3 group
see LUN (logical unit)
LUN (logical unit)
planning, LUNs and file systems 3-4
plant requirements
30-slot SCSI-disk systems 5-15
power requirements
30-slot SCSI-disk systems 5-15
power supplies (PSs)
R
rackmount model, DPE storage system 5-2
RAID group configurations
types and tradeoffs, 2-1; see also
RAID groups types and tradeoffs, 2-1;
014-002912-01
Index-3
Index
RAID 0 group
RAID 1 mirrored pair
RAID 1/0 group
RAID 3 group
RAID 5 group
redundant array of independent disks
RAID group
S server
configurations
connection to storage system, see
cabling dual-server configuration
planning worksheet
unshared configurations
service clearance
shared storage systems
hardware
single-server configurations
site requirements
30-slot SCSI-disk systems 5-15
size
SP (storage processor)
SPS (standby power supply), DPE storage system 5-6
sstorage components 5-1
storage managment worksheets 6-4
storage system caching
configurations
basic 4-2 dual-adapter/dual-SP 4-2
storage-system caching, as feature 5-8
stripe
switch
in sample shared storage configuration 3-1
T temperature requirements
30-slot SCSI-disk systems 5-15
terms
tradeoffs configuration
U unshared storage
hardware
storage system configurations 4-1
V
Index-4
014-002912-01
W weight
30-slot SCSI-disk systems 5-15
storage system installation 5-16
Index worksheet
application 3-4 completing 3-4
component planning
LUN configuration 3-9–3-11, 4-14–4-16
014-002912-01
Index-5
Vos remarques sur ce document / Technical publication remark form
Titre / Title : Bull DAS – Disk Array Storage Systems Planning a DAS Installation Fibre Channel
Environments
Daté / Dated : July 2000 Nº Reférence / Reference Nº : 86 A1 94JX 04
ERREURS DETECTEES / ERRORS IN PUBLICATION
AMELIORATIONS SUGGEREES / SUGGESTIONS FOR IMPROVEMENT TO PUBLICATION
Vos remarques et suggestions seront examinées attentivement.
Si vous désirez une réponse écrite, veuillez indiquer ci-après votre adresse postale complète.
Your comments will be promptly investigated by qualified technical personnel and action will be taken as required.
If you require a written reply, please furnish your complete mailing address below.
NOM / NAME :
SOCIETE / COMPANY :
ADRESSE / ADDRESS :
Date :
Remettez cet imprimé à un responsable BULL ou envoyez-le directement à :
Please give this technical publication remark form to your BULL representative or mail to:
BULL CEDOC
357 AVENUE PATTON
B.P.20845
49008 ANGERS CEDEX 01
FRANCE
Technical Publications Ordering Form
Bon de Commande de Documents Techniques
To order additional publications, please fill up a copy of this form and send it via mail to:
Pour commander des documents techniques, remplissez une copie de ce formulaire et envoyez-la à :
BULL CEDOC
ATTN / MME DUMOULIN
357 AVENUE PATTON
B.P.20845
49008 ANGERS CEDEX 01
FRANCE
Managers / Gestionnaires :
Mrs. / Mme :
Mr. / M :
FAX :
C. DUMOULIN
L. CHERUBIN
E–Mail / Courrier Electronique :
+33 (0) 2 41 73 76 65
+33 (0) 2 41 73 63 96
+33 (0) 2 41 73 60 19 [email protected]
Or visit our web site at: / Ou visitez notre site web à: http://www–frec.bull.com
(PUBLICATIONS, Technical Literature, Ordering Form)
CEDOC Reference #
N o Référence CEDOC
_ _ _ _ _ _ _ _ _ [ _ _ ]
Qty
Qté
CEDOC Reference #
N o Référence CEDOC
_ _ _ _ _ _ _ _ _ [ _ _ ]
Qty
Qté
CEDOC Reference #
N o Référence CEDOC
_ _ _ _ _ _ _ _ _ [ _ _ ]
Qty
Qté
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ]
_ _ _ _ _ _ _ _ _ [ _ _ ] _ _ _ _ _ _ _ _ _ [ _ _ ] _ _ _ _ _ _ _ _ _ [ _ _ ]
[ _ _ ] : no revision number means latest revision / pas de numéro de révision signifie révision la plus récente
Date : NOM / NAME :
SOCIETE / COMPANY :
ADRESSE / ADDRESS :
PHONE / TELEPHONE :
E–MAIL :
FAX :
For Bull Subsidiaries / Pour les Filiales Bull :
Identification:
For Bull Affiliated Customers / Pour les Clients Affiliés Bull :
Customer Code / Code Client :
For Bull Internal Customers / Pour les Clients Internes Bull :
Budgetary Section / Section Budgétaire :
For Others / Pour les Autres :
Please ask your Bull representative. / Merci de demander à votre contact Bull.
BULL CEDOC
357 AVENUE PATTON
B.P.20845
49008 ANGERS CEDEX 01
FRANCE
ORDER REFERENCE
86 A1 94JX 04
Utiliser les marques de découpe pour obtenir les étiquettes.
Use the cut marks to get the labels.
DAS
Planning a DAS
Installation
Fibre Channel
Environments
86 A1 94JX 04
DAS
Planning a DAS
Installation
Fibre Channel
Environments
86 A1 94JX 04
DAS
Planning a DAS
Installation
Fibre Channel
Environments
86 A1 94JX 04
advertisement