Storage Area Network
Configurations for
RA8000/ESA12000 on
Windows NT – Intel
Application Note
AA-RHH6B-TE
Visit Our Web Site for the Latest Information
At Compaq we are continually making additions to our storage solutions product line. Please check
our web site for more information on the complete line of Fibre Channel storage products, product
certification, technical information, updates to this document, and other available documentation.
This information can be accessed through our web page at:
www.compaq.com/storageworks
Introduction
This application note describes fibre channel switch based Storage Area Network (SAN)
configurations for RA8000/ESA12000 storage systems on Windows NT Intel platforms. For mixed
platform heterogeneous SAN configurations refer to the "Heterogeneous Storage Area Networks"
applicati o n n o t e, an d the ap p l i cab l e pl at fo rm s p eci fi c ap p l i cat ion notes listed i n Table 3.
Enterprise Network Storage Architecture (ENSA)
The Compaq Enterprise Network Storage Architecture is key to supporting Compaq’s NonStop
eBusiness strategy; through ENSA, Compaq leverages industry standards to allow deployment of
storage where applications need it. ENSA uses the Compaq StorageWorks product family to deliver
the storage solutions that address non-stop computing requirements like availability, reliability,
performance, scalability, manageability, and data backup/restore capabilities. ENSA addresses the
storage is s u es that o u r cust o m ers ex p ect to face n o w an d in the fu t u re. C o mpaq S ANs add res s today’s
issues including:
• Data protect ion
• High availability
• Increased distance
• High connectivity
• High bandwidth
• Multi-vendor platform support
• Enterprise backup support
• Economical capacity growth
• Scalability
• Investment protection
With the addition of multi-switch fibre channel Fabric support in the RA8000/ESA12000 FC storage
products and the integration of Enterprise Backup Solutions into the SAN, Compaq has taken the next
evolutionary step in delivery of the ENSA vision. Our customers can now realize the full benefits of a
scalable Storage Area Network providing the robust backbone needed to address dispersed server
operation with shared storage access and backup acros s t h e ent erpris e.
Copyright © 1999, Com paq C om puter Corporation AA–RHH6B–TE
Printed in U. S.A. Page 1
All Rights Re se rved
SAN Configurations for RA8000/ESA12000 on Windows NT - Intel
Product Overview
The RA8000 and ESA12000 FC storage systems are based on a common architecture and offer
customers centralized management, high availability, and exceptional performance and scalability in
open systems environments. These products provide a Fibre Channel storage solution with industry
leading SAN support utilizing both FC-Fabric (switch) and FC-AL (arbitrated loop) technology.
The RA8000 FC is a mid-range storage system available in a pedestal enclosure for departmental use
in the office. It can also be instal l ed in server rack s fo r small data cent ers . An eas y- t o-d ep l o y, flexible
solution for open systems, the RA8000 supports up to 2.6 TB of storage.
The ESA12000 FC, designed for the data center, is available in three easy-to-order building blocks;
each optimized for specific, high-capacity customer requirements. The ESA12000 FC offers
unprecedented levels of storage, s cali n g in performance and capacit y as user requi rem ent s in creas e.
These storage systems incorporate the latest in RAID technology providing RAID levels 0, 1, 0+1,
adaptive 3/5, and non-RAID disks. Compaq’s implementation of RAID capability assures that data
availability will be maintained despite hardware failure. Features such as read ahead cache and
mirrored writ e b ack cache improve or mini mi ze t h e affect o n p erfo rm an ce while preservin g d at a
availability and supporting high availability.
The RA8000/ESA12000 FC platform kits contain the necessary operating sys tem speci fi c s oft ware
and documentation needed to install, configure, and monitor your storage system. All platform kits
include the StorageWorks Command Console (SWCC) for storage management operations including
configuring, status inquiry and notification facilities. SWCC provides a user oriented graphical user
interface (GUI) s i m p l i fy i ng t he most co m plex o f sto rag e m an agem en t o perat i o n s .
Microsoft Cluster Server (MSCS)
A Windows NT MSCS cluster enables two servers to share an RA8000 or ESA12000 FC storage
system through a Fibre Channel bus. If a server failure should occur, the storage I/O is re-routed
through to the functioning server. This process, called cluster failover, requires no resource
downtime ensuring the high availability of data.
The RA8000 and ESA12000 FC storage systems are supported in Windows NT MSCS cluster
configurations. All of the cluster configurations provide full cluster failover capabilities through a
fibre channel bus connected to both cluster servers.
In addition to cluster failover, the RAID array controllers can be configured in transparent failover
mode to protect against the unlikely event of a single controller failure. This feature eliminates
controllers as single points of failure and is invoked without the need for server intervention.
For the highest availability, two hardware paths between the cluster servers and the RA8000 or
ESA12000 FC storage system can be configured using Compaq’s Secure Path for Windows NT. This
configuration provides two separate physical paths between the servers and the storage.
Secure Path for Windows NT
StorageWorks Secure Path is a high availability software product providing continuous data access
for RA8000 and ESA12000 FC storage products configured on Windows NT platforms. Redundant
hardware, advanced RAID technology and automated failover capability are used to enhance fault
tolerance and availability. Secure Path effectively eliminates controllers, interconnect hardware and
host bus adap t ers as sin g l e poi n t s of fai lure in the storage s y stem .
Key to Secure Path’s functionality is the capability of RA8000/ESA12000 FC dual HSG80
controll ers t o operat e i n t h e activ e/ act i v e M u lti p l e-B u s fail o v er m o de. T his failover mode allo ws each
controller to be configured on its own bus and to process I/O independently under normal operation.
Available s t o rag e u ni ts are “preferred” to one or the other of the two controllers which determines
which controller path is used for access at system boot time. During runtime, storage units may be
moved between paths at anytime through use of the management utility without disrupting operations.
Page 2 AA–RHH6B–TE
Application Note
Controllers in Multip le - Bus failover mode monitor each other and au t o m atical ly failover sto rag e u nits
from the failed member of a controller pair. The Secure Path software detects the failure of I/O
operations to complete on a failed path and automatically re-routes all traffic to the surviving path.
Controller and path failover is completed seamlessly, without process disruption or data loss.
The Secure Path management utility provides continuous monitoring capability and identifies failed
paths and failed-over storage units. To facilitate static load balancing, storage units can be moved
between paths using simple “drag-and-drop” operations.
The integration of Secure Path, Windows NT Clustering, and StorageWorks RAID Array Fibre
Channel technology provides the maximum level of fault-tolerance, data availability, and
performance req uired for miss i o n-critical environ m ents.
Enterprise Backup Solutions
The Enterprise Backup Solution for Legato NetWorker is an integration of industry-standard
application software and hardware that provides a complete enterprise class tape solution. The
implementation combines tape and disk storage on a single SAN providing backup capabilities for up
to 7 Windows NT Intel or Tru64 UNIX servers and 2 disk storage systems using a single shared tape
library. The Legato NetWorker and SmartMedia software applications manage the backup operations
for the solution
. Future releases will support increased capabilities.
SAN Concepts
A Storage Area Network interconnects multiple shared storage environments and servers into a
storage network. All storage can be managed from a single location or from multiple locations. All of
the storage becomes available to any server, regardless of physical location.
SAN Management
SAN management tools and product features provided with the RA8000/ESA12000 storage systems
include the following:
• SAN Storage and Switch Management
SWCC – Utilized to manage and monitor s torag e s y s t em s, sto rag es et s , an d SAN fibre
channel switches including configuration, status inquiry and notification facilities.
• SAN Enterprise Backup Management
Legato NetWorker and SmartMedia software manages and automates the backup-and-restore
process using Compaq servers, disk arrays, tape libraries and FC interconnect hardware.
• SAN Access Management
Switch Fabric Zoning – The FC switch Zoning feature provides a means to control SAN
access at the node port level. Zoning can be used to separate one physical Fabric into many
virtual Fabrics consisting of selected server and storage ports. This capability allows you to
setup barriers between different operating environments, to deploy logical Fabric subsets by
creating defined server and/or storage groups, or to create temporary server and storage zones
for tape backup. Zones may be configured dynamically and the number of zones and zone
membership are effectively unlimited. Nodes can be in multiple zones to allow for
overlapping depending on the desired access control. Use of Zoning is supported in both
homogeneous Windows NT SAN configurations and heterogeneous mixed platform SAN
configurations.
Selective Storage Presentation (SSP) – The RAID Array SSP feat u re p ro vi d es a mean s to
control access at the storages et l evel . SSP is an exclusi v e RA8000/ESA12000 storage system
feature that allows multiple servers or host bus adapters (HBA) to share access to the sam e
RAID array reliabl y , with each server or HBA's sto rag esets (LUNs) present ed ex cl u s i v el y to
AA–RHH6B–TE Page 3
SAN Configurations for RA8000/ESA12000 on Windows NT - Intel
those th at are all o wed acces s . Ad d i tio n al l y, SSP al lows the s etting of host modes and LUN
offsets for each HBA connected t o the array. The host mode is special l y t ai l ored to the
storage com m u nication techni qu es o f the o perat ing system. The LUN offs et feat u re al l o ws
higher numbered LUNs in a RAID Array to be presented in a range required by specific
operating s ys tems. The SSP feature also provi d es a mean s to t rack the n u m ero us F C HB As
within servers at tached to a SAN by ident i fyin g each b y Worl dw i d e ID (WWID).
Additional information about these products and features is available in the documentation listed in
Table 4
SAN Performance Consideration s
The performance of an application on a system that utilizes RA8000 or ESA12000 FC storage can be
limited by many different components of the system and the configuration of the SAN. Some of the
possible component limiting factors include the host CPU(s), memory size, FC HBA, RAID
controllers, or the specific configuration of disks used behind the controllers. At the SAN level,
performance can be limited by the number and arrangement of FC switches and inter-switch links
(ISL) in the F ab ri c, an d t h e way servers an d s torag e s y s t em s are con n ect ed t o t h e Fabr i c. Th e l im i ti ng
factor can move to any of these areas depending on the workload. Identifying the limits will assist in
determining the best configuration for a given application.
Table 1 lists the upper limit performance specifications for the components of the RA8000 and
ESA12000 FC storage systems based on testing using standard storage performance tests and
methodologies. These numbers should be used to compare component level performance as a means
to determine the best configuration from a performance perspective. User application tests may not
necessarily reach t h es e l ev el s of perfo rm an ce as ap p l i cat i o n s may p erform additional l ev el s of
processing fo r each I/ O. The controller s p eci fi cat i o n s li sted s h ow bot h cache (n o d i s k access ) an d
media (with dis k access ) limitations.
The limits are based on I/O performance (I/O’s per second) - typical of small transfer applications
such as databases and mail, and bandwidth performance (MB’s per second) - typical of large transfer
applications such as video and graphics.
Table 1 Performance Limits of FC-Fabric Components
Transfer Size IO/sec (small transfer sizes,
random access)
Operation Type Read Write Read Write
FC HBA - KGPSA-BC 15,500 14,250 86 51
FC Switch (see note)
FC Switch ISL ( s ee not e)
Single HSG80 Controller
(1 Active Port)
Single HSG80 Controller
(2 Active Ports)
Dual HSG80 Controll ers
(2 Active/2 Standby Ports)
Dual HSG80 Controll ers
(4 Active Ports)
400,000
25,000
12,000 Cache
4,100 Media
12,000 Cache
4,100 Media
24,000 Cache
8,200 Media
24,000 Cache
8,200 Media
400,000
25,000
9,000 Cache
4,000 Media
9,000 Cache
4,000 Media
18,000 Cache
8,000 Media
18,000 Cache
8,000 Media
MB/sec (large transfer sizes,
sequential access)
1600
100
77 Cache
54 Media
98 Cache
54 Media
154 Cache
102 Media
195 Cache
102 Media
1600
100
50 Cache
44 Media
100 Cache
47 Media
101 Cache
88 Media
178 Cache
88 Media
EBS TBS TBS TBS TBS
Page 4 AA–RHH6B–TE
Application Note
NOTE
Fibre channel switch and ISL performance limits are
theoretical. Actual measured performance is lower due to
limits unrelated to the switch.
SAN Configuration Guidelines
The following sections provide information about configuring SANs including the base Fabric design
selection, fibre channel Fabric rules, platform and disk storage connectivity rules, Enterprise Backup
integration rules and Compaq Secure Path for Windows NT rules. The configuration information is
presented in the form of rules to provide for flexibility and customization depending on the specific
customer need. In addition to the listed rules, also shown are examples of recommended SAN
configurations based on the number and arrangement of fibre channel switches in a Fabric. The
example “SAN Fabric Configurations” illustrates proper application of the listed rules with regard to
fibre channel switch arrangement, switch interconnection in a Fabric, and platform/storage
connections. The exact SAN Fabric configuration used for your specific application should begin
with one of the base SAN configurations shown. In choosing a Fabric configuration you should
consider capacity, connectivity, availability, distance, backup, and performance needs as well as
future growth requirements.
It is not required that you configure server and storage in the SAN exactly as depicted in the
illustrations, however it is strongly recommended that the switch interconnection rules and platform
connectivity rules be strictly adhered to.
This application note describes homogeneous Windows NT SAN configurations. It is expected that
customers may desire to implement a heterogeneous SAN consisting of many different operating
system platforms. Whether implementing a homogeneous or heterogeneous SAN, it is necessary to
adhere to the p l at fo rm specific rules and maxi m ums fo r each g iven pl atform within the SAN.
SAN Design Selection Process
It is suggested that the following general steps be followed when initially designing a SAN.
1. Considering capacity, connectivity, availability, distance, performance, and backup requirements,
select the SAN Fabric configuration that best fits your needs. Refer to the configuration
illustrations, notes, Fabric rules, and platform server and storage connectivity rules. Using the
maximum server and maximum storage counts shown as the upper bound limits determine the
specific number of servers or storage required number for your particular installation.
• If you want a lower server count than the maximum listed: You may increase the storage
count, but only to the upper limit indicated for maximum storage, and provided you do not
exceed the platform limits listed in Table 2 for HBAs/Server and Controller Ports/HBA
• If you want a lower storage count than the maximum listed: You may increase the server
count, but only to the upper limit indicated for maximum servers, and provided you do not
exceed the platform limits listed in Table 2 for Servers/Storage System
• If capacity or co n n ect i o n n eeds are g reater than prov i d ed in a 4 switch Fabric, consider
implementing multiple 4-switch Fabrics (future releases will support inter-connection of
multiple Fabrics into a single Fabric to provide for growth and scaling of the SAN).
2. If you are configuring a homogeneous Windows NT Intel SAN, refer to the Windows NT Intel
Server/Storage configuration rules in this document to determine how best to configure servers
and storage. Line 1 in Table 2 lists the configuration maximums for Windows NT Intel platforms
in a homogeneous SAN configuration.
3. If you are configuring a heterogeneous SAN, refer to the appropriate platform application note(s)
and the heterogeneous SAN application note referenced in Table 3 to ensure platform interaction
AA–RHH6B–TE Page 5
SAN Configurations for RA8000/ESA12000 on Windows NT - Intel
Line 2 in Table 2 lists the configuration maximums for Windows NT Intel platforms in a
heterogeneous SAN configuration.
4. If implementing a SAN for high availability storage using Compaq Secure Path for Windows NT,
refer to the Secu re Path config ur at ion ru l es.
5. If implementing a SAN integrating Compaq Enterprise Backup on your SAN, refer to the EBS
connectivity and configuration rules
6. Select the desired method(s) of SAN management and access control bas ed on your specific
needs. Use SWCC for storageset management, and controller based SSP or switch based Zoning
(or both) for d isk stor age acces s co n t ro l . Use Secure Path Manager t o m anag e h i gh avai l ab i li t y,
use Legato NetWorker and SmartMedia software to manage the backup-and-restore process.
SAN Configuration Rules and Maximums
The following sections list the SAN design rules as they apply to Fabric configurations, Windows NT
Intel platforms for servers and storage, enterprise backup integration, and high availability storage.
Fabric Rules
• Up to 4 fibre chann el s witch es t otal i n a single Fabric - a SAN with all swit ches in t erco n nected.
All Compaq FC 16-port and 8-port switch models are supported inter-mixed
• A server can attach to multiple multi-switch Fabrics (Figure 1). The number of separate Fabrics
per server or total number of switches per server in separate Fabrics is based on the specific
server model capabilities and the maximum number of FC HBAs supported
• Within a single Fabric, maximum of 1 switch hop between servers and storage, maximum of 2
switch hops worst case with a single fault - a Fabric can re-route to a 2-hop path on a single link
failure. A hop is defined as 1 or more connections between two FC switches, for example; 2
switches cas cad ed = 1 ho p. S erv er t o FC swit ch s eg m en t s and St o rag e t o FC s wit ch s e gments are
not counted as hops
• Within a sin g l e Fab ri c wh ere s wi tches are interconnected, each FC swi t ch mus t hav e a u niqu e
domain number (Domain_ID)
• Up to 15 inter-switch links (ISLs) between any 2 switches
• Minimum cable segment length is 2 meters
• Up to 200 meters per cable segment using short-wavelength laser GBICs and 62.5 micron multi-
mode fiber optic cable. With multiple cable segments 600 meters total distance nominal, 800
meters worst cas e wi t h a sing l e fault re-route, betw een s erv er an d s t or age
• Up to 500 meters per cable segment using short-wavelength laser GBICs and 50u multi-mode
fibre optic cable. With multiple cable segments 1.5 km total distance nominal, 2.0 km worst case
with a single fault re-route, between server and storage
• Up to 10 km per ISL cable segment using long-wavelength laser GBICs and 9u single-mode fibre
optic cable. With multiple cable segments 11 kilometers total distance nominal, 21 km worst case
with a single fault re-route, between server and storage. 10 km links are only supported in switch
to switch connections
Figure 1 One Server to Multiple 4-Switch Fabrics
Fabric 1 Fabric 3
Page 6 AA–RHH6B–TE
Server
Fabric 2
SHR-1585
Application Note
Windows NT Server/Storage Rules
Table 2 describes the Windows NT Intel platform maximums when in a homogenous Windows NT
SAN and when using Windows NT Intel platforms in a heterogeneous mixed platform SAN.
Table 2 Guidelines fo r Windows NT Int el Platf orm Supported Maximums
Active
Maximum
Supported
Number of:
Windows NT SAN 4 4 8 4 16
Heterogeneous SAN
Host Bus
Adapters
per Server
(1)
Controller
Host Ports
(HBA Targets)
per HBA
(2)
LUNs
per HBA
(3)
Target
HBAs per
Server per
Switch Zone
(4)
Active
Servers
per
Storage
System
8
(5)
Table Notes:
1. The actual maxi m um number of host bu s adapters per serv er i s dependent on the specific server mod el.
The recommend ed m aximum number of active HSG80 contro ller host ports a single HBA can simultaneousl y
2.
access, the ac tual number of Targ et s available is 16, inc luding the HBA.
3.
The maximum number of LUNs per SCSI target. T he maximum numbe r of LUN s pe r t arg et a ho st bu s adapter can
address in a SAN may be l ess.
4.
The maximum number of HBAs per ser ver that can be connected to the same sw it ch zone.
5. This column gives th e maximum number o f active hosts for one s torage system in a Homog eneous SAN using the
specific operat ing s ystem. This assumes 1 FC H B A per ho st for non-HA Transpar ent Failover configurations and
2 FC HBAs per host for HA Multi pl e-b us Failover configurat ions. The Heterogen eous Access entry is the
maximum total number of active hos t conn ections when mul tiple ope rating systems are ac cessing the same sto rag e
system.
• All configurations are supported under Windows NT 4.0 with SP4 (minimum)
• For Windows NT, all server connections into the SAN use a FC-Fabric based HBA. FC-AL and
FC FL-Port HBA connections into the SAN are not supported for Windows NT platforms
• The maximum number of FC HBAs per server is 4, or total allowed in the server if less than 4.
Use Selecti v e St o rag e P res en t ation (see bel ow) to as sig n speci fi c L UNs exclusively to each HB A
• The maximum number of active controller ports per FC HBA is 4
• Any combination of Windows NT Servers - standalone or clustered, and RA8000/ESA12000
storage systems is allowed per SAN configuration, provided these rules are followed:
1. A maximum of 16 Windows NT servers (assumes 1 FC HBA per server for non-HA and
2 FC HBAs per server for HA) per RA8000 or ESA12000 dual controller storage system,
with a maxim um of 8 servers co n fi g u red o n each active port. See “Extended
Configurations” description below when g reat er than 4 servers have access t o t h e same
HSG80 controller host port pair
2. All specific SAN configuration rule limits are followed. Refer to the individual SAN
Fabric configurations
• Extended Configurations - If you configure greater than 4 (up to 8) Windows NT servers
(assumes one FC HBA per server) for access to a single host port on an RA8000/ESA12000 Fibre
Channel s torage system , y o u mu st s el ect the “Extended Configuration” check box in the
StorageWorks Windows NT Platform Kit Fibre Channel Software Setup utility custom
installation setup. Each NT server in this configuration must have the option selected. Selecting
this option will adjust registry settings for your KGPSA host bus adapter to operate in an
"Extended Configuration" environment
• All configurations require the Connection Name Operating System parameter set to “WINNT”
• When configuring HSG80 controllers for access from Windows NT servers only, the controll er
SCSI mode should be set to “SCSI-2” and the Command Console LUN (CCL) must be disabled.
When configuring controllers for access from Windows NT and other platforms that requi re
“SCSI-3” mode, set the controllers to SCSI-3 mode. When set in SCSI-3 mode the CCL will be
fixed at LUN 0. This will reduce the total number of LUNs available from 8 to 7 for all server
connections.
AA–RHH6B–TE Page 7
SAN Configurations for RA8000/ESA12000 on Windows NT - Intel
• Each active con t ro l l er h os t po rt p res en t s one SCS I Targ et ID wi th u p to eigh t LUNs to each
server (FC HBA) that has a connected path through the Fabric to the port.
• The maximum number of LUNs per HSG80 controller pair is 128
• Dual redundant controllers can be configured for Transparent Failover Mode or Multiple-Bus
Failover Mode. Multiple-bus Failover mode requires Compaq Secure Path for Windows NT on
the server
• Each storage s y s t em can be configured u s i n g SWCC or t h e Comm an d L ine In t erface (C L I)
through either the HSG80 controller maintenance serial port or a TCP/IP interface
• All configurations utilize ACS V8.5F (Array Controller Software) configured for FABRIC
topology
• ACS V8.5 provides for up to 64 connection name entries. For dual controllers in Transparent
Failover M o d e, up to 2 co nn ect i o n ent ri es are creat ed fo r each s erv er HB A, thus a total of 32
server HBAs are visible to a storage system at any given time. In Multiple-Bus Failover Mode, up
to 4 connectio n en t ri es are creat ed fo r each serv er HB A, so a total of 16 server HBAs are vi s i b l e.
Note: The maximum number of server HBAs supported on a storage system is 16. Each storage
system in the SAN will automatically add connection name entries upon initial power up. As a
result, in certain SAN configurat i ons it is possible to exceed the number of available connect i on
name entries i n a st o rage s y stem prior to the additi o n of ent ri es fo r the s p eci fi c s erv er HB As th at
you want access to. This could result in the inability of a storage system to properly connect to a
desired server HB A i n the SAN. To avoid this, you can configure s p eci fi c s erv er HB As an d
specific st o rag e s y s t ems into FC Switch Zon es o f up to 32 S erv ers wh en in T ran s p aren t F ail o v er
and up to 16 Servers when in Multiple-bus Failover.
• Rename connection names. By default new connection names are automatically added to the
controller connection name table by the controllers when they detect a path to an adapter from
each active controller host port. The default connect i on name assigned by the controllers will be
“!NEWCONnn”, where nn is a num b er fro m 1 to 64. After t h e co ntrol l ers d et ect al l pat h s,
rename each connection name to be more meaningful to the configurati on i. e., SERVER1P1,
SERVER 1P 2, etc. (connectio n names can be a maximum of 9 characters ) .
• Selective Storage Presentation - for configurations with more than 1 FC HBA in a server or more
than one server accessing through the same HSG80 controller host port,
Use the Connection Name Unit Offset value to set the range of unit numbers to be visible
from each server or HB A (u sing the CLI, “SET connection-name UNIT_OFFSET = n”)
Use the Unit Connection Name parameter to allow exclusiv e acces s t o un i ts fro m each s erver
or HBA (using the CLI, “SET unit-number DISABLE_ACCESS_PATH = ALL
ENABLE_ACCESS_PATH = connection-name”)
Shared access between different servers to the same
storage unit (LUN) requires specific application software
(i.e., MSCS) to ensure proper data preservation.
• LUNs (logical units) can consist of single disks (JBOD), a storageset of multiple disks or a
partition (up to 8), configured for a specific RAID level of 0, 1, 0 + 1, or 3/5
• The maxim u m storag e cap aci t y s t at ed (2 .6 TB) assumes 36 GB dis k s, increas es in disk driv e
capacity will increase overall storage capacity for the same enclosure footprint
Enterprise Backup Solution Integration Rules
• Supports Legato NetWorker backup for up to 7 Windows NT or Tru64 UNIX Servers
• Supports all Compaq FC 16-port and 8-port switch models inter-mixed
• Connected to the SAN using FC switch Fabric topology connection
• Up to 10 km between FC switch and a tape controller using long-wavelength laser GBICs and 9u
single-mode fiber optic cable
Page 8 AA–RHH6B–TE
NOTE
Application Note
• Up to 2 Fibre Channel Tape Controllers
• Maximum of 1 TL895 library
• Up to 7 tape drives
• Maximum of 1 switch hop between servers and a tape controller
• Supports backup for up to 2 RA8000 or 1 ESA12000 (4 controller pairs) storage systems
Compaq Secure Path for Windows NT Ru les
• All High Availability Multiple-Bus Failover configurations require Compaq Secure Path for
Windows NT (minimum revision v2.2) and two physical paths to a pair of controller ports on the
storage system
• One instance of the Secure Path Manager can control two paths on up to 4 storage systems. Two
Paths equates to 2 FC HBA’s connected to 2 controller host ports (i.e., Port 1 of Controller A and
Port 1 of Controller B). Another instance is required to control two paths to the other pair of ports
on a storage system (Port 2 of Controller A and Port 2 of Controller B)
• For standalone servers (non-cluster), a separate instance of the Secure Path Manager is run on
each server. For Windows NT Clusters, one instance of t h e Secure Path Manager can be run for
both cluster servers. For centralized management, multiple instances of the Secure Path Manager
can be run on a sin gle Client provided n et w or k acces s is available bet w een the Client and al l HA
servers.
• For servers acces s i ng co n t ro l l er ho st p o rt 1 use LUN n umber an d offs et valu es i n th e rang e o f 0 –
99, for controller host port 2 use number and offset values in the range 100 -199
• Use the Unit Preferred Path parameter to assign units to a specific controller. Distribute units
equally across both controllers for proper load balancing
• SSP - Storages et s (LUNs ) mus t be enab l ed for acces s fro m b oth p at hs us ing the Un i t Connection
Name parameter feature
• Secure Path configurations require a minimum of two FC HBAs in a server. Each FC HBA must
be connected to a different FC switch, and for cable symmetry, should be on the same switch port
number, for example:
Server 1 FC HBA 1 to FC Switch 1 - Switch Port 1
Server 1 FC HBA 2 to FC Switch 2 - Switch Port 1
Server 2 FC HBA 1 to FC Switch 1 - Switch Port 2
Server 2 FC HBA 2 to FC Switch 2 - Switch Port 2
It is highl y recom m en d ed t h at th e cab l i n g sch em e s h o wn in each S ecu re Path multipl e-b us
configurations be followed as depicted.
• For version 2.2, the Secure Path Manager uses the controller serial number to determine which
path (controller) will be displayed first in the GUI, thus the lower value serial number will be
displayed first in the display as the top path labeled “CONTROLLER A”. The higher controller
serial number will be displayed as the bottom path and labeled as “CONTROLLER B”.
AA–RHH6B–TE Page 9
Loading...