HP XP48, XP128, XP512, XP1024, XP12000 User guide

HP StorageWorks

Cluster Extension XP

user guide

XP48

XP128

XP512

XP1024

XP12000

Product Version: 2.05.00 sixth edition (August 2004)

T1609-96004

This guide explains how to use the HP StorageWorks Cluster Extension XP software.

© Copyright 2003-2004 Hewlett-Packard Development Company, L.P., all rights reserved

Hewlett-Packard Company makes no warranty of any kind with regard to this material, including, but not limited to, the implied warranties of merchantability and fitness for a particular purpose. Hewlett-Packard shall not be liable for errors contained herein or for incidental or consequential damages in connection with the furnishing, performance, or use of this material.

This document contains proprietary information, which is protected by copyright. No part of this document may be photocopied, reproduced, or translated into another language without the prior written consent of Hewlett-Packard. The information contained in this document is subject to change without notice.

All product names mentioned herein may be trademarks of their respective companies.

Hewlett-Packard Company shall not be liable for technical or editorial errors or omissions contained herein. The information is provided “as is” without warranty of any kind and is subject to change without notice. The warranties for Hewlett-Packard Company products are set forth in the express limited warranty statements accompanying such products. Nothing herein should be construed as constituting an additional warranty.

Printed in the U.S.A.

HP StorageWorks Cluster Extension XP User Guide

product version: 2.05.00 sixth edition (August 2004) part number: T1609-96004

2	HP StorageWorks Cluster Extension XP user guide

Contents

About this guide	9
Intended Audience			10
Disk array firmware and software dependencies								10
Related information			12
Terminology	13
Conventions	14
HP storage website			15
HP authorized reseller				15
Revision history		15
Warranty statement		17
1 Cluster Extension XP features					19
Integration into cluster software					20
Disaster tolerance through geographical dispersion								21
Disaster tolerance considerations						21
Disaster-tolerant architectures					22
Automated redirection and monitoring of mirrored Continuous Access XP
pairs 29
Rolling disaster protection				30
What is a rolling disaster? 30
Recovering the disaster tolerant environment							31
Command-line interface for easy integration 33
Graphical user interface			34
Quorum service (Microsoft Cluster service only)							35
2 Cluster Extension XP processes and components								37
Cluster Extension XP environments						38
Cluster Extension XP execution					40

Contents

3

	Continuous Access XP and RAID Manager XP 41
	RAID Manager XP instances						43
	RAID Manager XP device groups							44
	Rolling disaster protection and Business Copy XP										45
	Integration with RAID Manager XP								46
	Integration with automatic recovery								47
	Integration with the pair/resync monitor									47
	Restoring server operation					48
	Example	48
	User configuration file			50
	Pair/resync monitor		51
	Force flag	53
	Pre-execution and post-execution programs									54
	Cluster Extension XP log facility						57
	Error return codes		58
	Quorum service for Microsoft Cluster service									59
	Using the quorum service in a Microsoft Cluster service
	environment		59
	Quorum processes			60
3	User configuration file and Cluster Extension XP objects 65
	The user configuration file				66
	File structure 67
	Specifying object values				68
	COMMON section objects				71
	APPLICATION section objects					73
	Basic configuration example				87
4 RAID Manager XP dependencies								89
	RAID Manager XP configuration						90
	RAID Manager XP configuration file								90
	Network considerations				93
	Command device considerations							93
	Start and stop the RAID Manager XP instances										94
	Takeover basic functionality test						95
5	Integration with HACMP				97
	Configuring resources			98
	Procedure for HACMP			99
	User configuration file for HACMP							101

4	HP StorageWorks Cluster Extension XP User Guide

Bringing a resource group online				104
Taking a resource group offline				105
Deleting Cluster Extension XP			106
Pair/resync monitor integration				107
Timing considerations		110
Failure behavior	112
Restrictions for IBM HACMP with Cluster Extension XP									113
6 Integration with Microsoft Cluster service						115
Configuring the quorum service				116
Configuring Cluster Extension XP resources						117
Resource group and resource names 118
Cluster Extension XP resource-specific parameters								118
Setting non-Cluster Extension XP resource-specific parameters										119
Adding a Cluster Extension XP resource					123
Changing Cluster Extension XP resource properties								125
Advanced properties		127
Changing a resource name			128
Adding dependencies on a Cluster Extension XP resource									129
Bringing a Cluster Extension XP resource online							131
Taking a Cluster Extension XP resource offline							133
Deleting a Cluster Extension XP resource					134
Pair/resync monitor integration				135
Timing considerations for Microsoft Cluster service								137
Failure behavior with Microsoft Cluster service							139
Bouncing Resource Groups			139
Unexpected offline conditions				139
Restrictions for Microsoft Cluster service with Cluster Extension XP										141
Disaster-tolerant configuration example using a file share									142
Administration	147
7 Integration with VCS		149
Configuration of the Cluster Extension XP agent							150
Configuring the Cluster Extension XP resource							154
Cluster Extension resource types 154
Resource type definition 155
Adding a Cluster Extension XP resource					156
Changing Cluster Extension XP attributes					158
Linking a Cluster Extension XP resource					160
Bringing a Cluster Extension XP resource online							161

Contents

5

	Taking a Cluster Extension XP resource offline							163
	Deleting a Cluster Extension XP resource						164
	Pair/resync monitor integration					165
	Timing considerations for VCS					166
	Enable/disable service groups					168
	Restrictions for VCS with Cluster Extension XP							169
	Unexpected offline conditions					171
8	Integration with Serviceguard for Linux						173
	Configuration of the Cluster Extension XP environment								174
	Adding a Cluster Extension XP integration to an existing Serviceguard
	package		180
	Starting a Serviceguard package with Cluster Extension XP									181
	Halting a Serviceguard package with Cluster Extension XP									182
	Deleting Cluster Extension XP from a Serviceguard package									183
	Pair/resync monitor integration					185
	Timing considerations for Serviceguard						188
9	Command-line interface (CLI)					191
	Configuring the CLI 193
	Creating the Continuous Access environment and configuring
	RAID Manager				193
	Timing considerations 193
	Restrictions for customized Cluster Extension XP
	implementations				195
	Creating and configuring the user configuration file								195
	CLI Commands			196
	clxrun	196
	clxchkmon		198
	clxqr	201
10	Troubleshooting			203
	Start errors 205
	Failover error handling				206
	HACMP-specific error handling					207
	Start errors		207
	Failover errors			207

6	HP StorageWorks Cluster Extension XP User Guide

	Microsoft Cluster service-specific error handling						211
	Solving quorum service problems					211
	Resource start errors			213
	Failover errors		214
	VCS-specific error handling				216
	Start errors	216
	Failover errors		217
	Serviceguard (SG-LX)-specific error handling 220
	Start errors	220
	Failover errors		220
	Pair/resync monitor messages in syslog/errorlog/messages/Event
	Log 222
A	Recovery procedures			225
	XP disk pair states		226
	Recovery sequence		228
	Quorum service recovery (Microsoft Cluster service only)							230
	Single site failure recovery				230
	Failure recovery if both sites have failed 232
	Procedure for quorum service system cleanup						233
B	Cluster Extension XP resource message catalog						235
C	Cluster Extension XP quorum service message catalog							261
	Quorum service Event Log messages					266

Glossary 269

Index 271

Contents

7

8	HP StorageWorks Cluster Extension XP User Guide

About this guide

This guide provides information about using and configuring HP StorageWorks Cluster Extension XP in an environment where clustered systems are connected to a disaster recovery array-based mirroring solution. Cluster Extension XP allows creation of dispersed multiplatform cluster configurations with the XP disk array. Cluster Extension XP enables cluster software to automatically failover applications where data is stored and continuously mirrored from a local to a remote disk array using HP StorageWorks Continuous Access XP. This guide describes the options you have to make your disaster tolerant environment as robust as possible to keep your data available at all times.

Because the XP family disk arrays supports a broad range of operating systems and cluster software, Cluster Extension XP can be integrated with almost any disk array-supported cluster software. This guide provides you with the information you need to create a two or more data center disaster tolerant environment utilizing the XP disk array and its Continuous Access XP remote mirroring feature.

Unless otherwise noted, the term disk array refers to these disk arrays:

HP Surestore Disk Array XP512

HP Surestore Disk Array XP48

HP StorageWorks Disk Array XP128

HP StorageWorks Disk Array XP1024

HP StorageWorks XP12000 Disk Array

About this guide

9

Intended Audience

This guide is intended for system administrators who maintain the cluster environment and storage subsystems and have the following knowledge:

•A background in data processing and direct-access storage device subsystems and their basic functions.

•Familiarity with disk arrays and RAID technology.

•Familiarity with the operating system, including commands and utilities.

•A general understanding of cluster concepts and the cluster software used in the data center environment.

•Familiarity with related disk array software programs:

HP StorageWorks Continuous Access XP HP StorageWorks RAID Manager XP

Disk array firmware and software dependencies

The features and behavior of failover operations depend on the XP firmware and RAID Manager XP versions. This guide describes Cluster Extension XP behavior based on features implemented in the latest XP firmware and RAID Manager XP versions.

10	HP StorageWorks Cluster Extension XP User Guide

Related information

For information about the disk arrays, please refer to the owner’s manuals.

For related product documentation, see the HP web site (www.hp.com):

HP StorageWorks RAID Manager XP: User’s Guide

HP StorageWorks Continuous Access XP: User’s Guide

HP StorageWorks Business Copy XP: User’s Guide

HP StorageWorks Command View XP: User’s Guide

HP StorageWorks Disk Array XP Operating System Configuration

Guide: IBM AIX

HP StorageWorks Disk Array XP Operating System Configuration

Guide: Sun Solaris

HP StorageWorks Disk Array XP Operating System Configuration

Guide: Windows 2000/2003

•HP StorageWorks Disk Array XP Operating System Configuration Guide: Linux

For information about Serviceguard for Linux, see the HP High Availability web site:

docs.hp.com/hpux/ha/

For information about RS/6000 and HACMP, see the IBM web site: www.rs6000.ibm.com/aix/library

For VERITAS Cluster Server information, see the VERITAS web site: support.veritas.com

About this guide

11

For Microsoft Cluster service information, see the Microsoft web site:

Windows 2000: www.microsoft.com/windows2000/library/technologies/ cluster/default.asp

Windows 2003: www.microsoft.com/windowsserver2003/library/technologies/ clustering/default.mspx

Terminology

This guide uses terminology to describe cluster-specific and disaster recovery-specific processes. Vendors of cluster software use different terms for the components of their cluster software. To standardize the usage among vendors, this guide uses the following terms:

application service This is the unit of granularity for a failover or failback operation. It includes all necessary resources that must be present and which the application depends on. For example, a file share must have a disk, a mount point (or drive letter) and an IP address to be considered an application service. A disk is a necessary resource for the application service. Depending on the cluster software, application services can depend on each other and run in parallel on the same system or on different systems.

Vendor equivalent terms

VCS: service group HACMP: resource group

Microsoft Cluster service: resource group SG-LX (Serviceguard): package

resource	The smallest unit in an application service. It describes
	the necessary parts to build an application service. The
	implementation of such resources in cluster software is
	vendor-specific. Some vendors (such as IBM or HP) do
	not allow accessing the chains between dependent
	resources.
	Vendor equivalent terms
	VCS: resource
	HACMP: resource group

12	HP StorageWorks Cluster Extension XP User Guide

	Microsoft Cluster service: resource
	SG-LX (Serviceguard): package
startup
shutdown	Startup and shutdown are also known as “bringing
	online” and “taking offline,” or “start” and “stop,” or
	“run” and “halt” in regards to an application service or
	resource. Only a few cluster software vendors (such as
	Veritas or Microsoft) offer starting and stopping of
	single resources.

Conventions

This guide uses the following text conventions.

Figure 1	Blue text represents a cross-reference. For the online
	version of this guide, the reference is linked to the
	target.
www.hp.com	Underlined, blue text represents a website on the
	Internet. For the online version of this guide, the
	reference is linked to the target.
literal	Bold text represents literal values that you type exactly
	as shown, as well as key and field names, menu items,
	buttons, file names, application names, and dialog box
	titles.
variable	Italic type indicates that you must supply a value. Italic
	type is also used for manual titles.
input/output	Monospace font denotes user input and system
	responses, such as output and messages.
Example	Denotes an example of input or output. The display
	shown in this guide may not match your configuration
	exactly.
[ ]	Indicates an optional parameter.

About this guide

13

{ }	Indicates that you must specify at least one of the listed
	options.
|	Separates alternatives in a list of options.

HP storage website

For the most current information about HP StorageWorks XP products, visit the support website. Select the appropriate product or solution from this website:

http://h18006.www1.hp.com/storage/arraysystems.html

For information about product availability, configuration, and connectivity, consult your HP account representative.

HP authorized reseller

For the name of your nearest HP authorized reseller, you can obtain information by telephone:

United States	1-800-345-1518
Canada	1-800-263-5868
Or contact:	www.hp.com

Revision history

February 2001

March 2001

July 2001

November 2001

May 2002

First release.

Added command-line interface chapter.

Added MSCS support.

Added quorum filter-service for MSCS on XP512/XP48.

Updated content for version 1.03 of all Cluster Extension products.

14	HP StorageWorks Cluster Extension XP User Guide

	Updated content for version 1.04.00 of Cluster
	Extension for MSCS.
	Added support for Serviceguard on Linux.
	Updated content for version 1.1 of Cluster Extension
	XP quorum service with external arbitrator.
September 2002	Updated content for version 2.00.
	Changed product terminology from MSCS to Microsoft
	Cluster service.
	Added arguments for clxchkmon.
	Changed LogLevel values.
	Changed Windows log file directory location.
	Added message catalog.
December 2002	Updated content for version 2.01 for VCS and
	Serviceguard.
	Added rolling disaster protection features.
	Added GUI features.
January 2003	Updated content for version 2.01 for Windows GUI.
April 2003	Updated content for version 2.02.
	Added “Cluster Extension XP quorum service message
	catalog” (page 261).
November 2003	Updated for versions 2.02 and 2.03. Added SUSE
	Linux and Windows 2003 support. Removed XP256.
	Changed MC/ServiceGuard to Serviceguard.
March 2004	Modified document for version 2.04.00.
August 2004	New format applied. Modified document for version
	2.05.00

About this guide

15

Warranty statement

HP warrants that for a period of ninety calendar days from the date of purchase, as evidenced by a copy of the invoice, the media on which the Software is furnished (if any) will be free of defects in materials and workmanship under normal use.

DISCLAIMER. EXCEPT FOR THE FOREGOING AND TO THE EXTENT ALLOWED BY LOCAL LAW, THIS SOFTWARE IS PROVIDED TO YOU “AS IS” WITHOUT WARRANTIES OF ANY KIND, WHETHER ORAL OR WRITTEN, EXPRESS OR IMPLIED. HP SPECIFICALLY DISCLAIMS ANY IMPLIED WARRANTIES OR CONDITIONS OF MERCHANTABILITY, SATISFACTORY QUALITY, NON-INFRINGEMENT, TITLE, ACCURACY OF INFORMATIONAL CONTENT, AND FITNESS FOR A PARTICULAR PURPOSE. Some jurisdictions do not allow exclusions of implied warranties or conditions, so the above exclusion may not apply to you to the extent prohibited by such local laws. You may have other rights that vary from country to country, state to state, or province to province.

WARNING! YOU EXPRESSLY ACKNOWLEDGE AND AGREE

THAT USE OF THE SOFTWARE IS AT YOUR SOLE RISK. HP

DOES NOT WARRANT THAT THE FUNCTIONS CONTAINED IN

THE SOFTWARE WILL MEET YOUR REQUIREMENTS, OR THAT

THE OPERATION OF THE SOFTWARE WILL BE UNINTERRUPTED,

VIRUS-FREE OR ERROR-FREE, OR THAT DEFECTS IN THE

SOFTWARE WILL BE CORRECTED. THE ENTIRE RISK AS TO THE

RESULTS AND PERFORMANCE OF THE SOFTWARE IS ASSUMED

BY YOU. HP DOES NOT WARRANT OR MAKE ANY

REPRESENTATIONS REGARDING THE USE OR THE RESULTS OF

THE USE OF THE SOFTWARE OR RELATED DOCUMENTATION IN

TERMS OF THEIR CORRECTNESS, ACCURACY, RELIABILITY,

CURRENTNESS, OR OTHERWISE. NO ORAL OR WRITTEN

INFORMATION OR ADVICE GIVEN BY HP OR HP’S AUTHORIZED

REPRESENTATIVES SHALL CREATE A WARRANTY.

16	HP StorageWorks Cluster Extension XP User Guide

LIMITATION OF LIABILITY. EXCEPT TO THE EXTENT PROHIBITED BY LOCAL LAW, IN NO EVENT INCLUDING NEGLIGENCE WILL HP OR ITS SUBSIDIARIES, AFFILIATES, DIRECTORS, OFFICERS, EMPLOYEES, AGENTS OR SUPPLIERS BE LIABLE FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, CONSEQUENTIAL, PUNITIVE OR OTHER DAMAGES (INCLUDING LOST PROFIT, LOST DATA, OR DOWNTIME COSTS), ARISING OUT OF THE USE, INABILITY TO USE, OR THE RESULTS OF USE OF THE SOFTWARE, WHETHER BASED IN WARRANTY, CONTRACT, TORT OR OTHER LEGAL THEORY, AND WHETHER OR NOT ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. Your use of the Software is entirely at your own risk. Should the Software prove defective, you assume the entire cost of all service, repair or correction. Some jurisdictions do not allow the exclusion or limitation of liability for incidental or consequential damages, so the above limitation may not apply to you to the extent prohibited by such local laws.

NOTE. EXCEPT TO THE EXTENT ALLOWED BY LOCAL LAW, THESE WARRANTY TERMS DO NOT EXCLUDE, RESTRICT OR MODIFY, AND ARE IN ADDITION TO, THE MANDATORY STATUTORY RIGHTS APPLICABLE TO THE LICENSE OF THE SOFTWARE TO YOU; PROVIDED, HOWEVER, THAT THE CONVENTION ON CONTRACTS FOR THE INTERNATIONAL SALE OF GOODS IS SPECIFICALLY DISCLAIMED AND SHALL NOT GOVERN OR APPLY TO THE SOFTWARE PROVIDED IN CONNECTION WITH THIS WARRANTY STATEMENT.

About this guide

17

18	HP StorageWorks Cluster Extension XP User Guide

1

Cluster Extension XP features

The quest to extend high availability over geographically dispersed locations has driven today’s IT personnel to demand cluster solutions capable of recovering from even the most extensive disasters. HP StorageWorks Cluster Extension XP enables you to monitor HP StorageWorks Continuous Access XP-mirrored disk pairs and allows access to the remote data copy if the application becomes unavailable on the local site. If the application service is restarted on the remote site, after the local (primary) application service has been shut down, Cluster Extension XP uses its internal database to check whether the current disk states allow automatic access to your data based on consistency and concurrency considerations. Integrated in the cluster software or available as command-line interface for your own integration, Cluster Extension XP ensures that the data can be accessed if necessary.

Cluster Extension XP software provides these key features:

•integration into cluster software

•disaster tolerance through geographical dispersion

•automated redirection and monitoring of mirrored Continuous Access XP pairs

•command-line interface for easy integration

Cluster Extension XP features

19

Integration into cluster software

The value of Cluster Extension XP is to provide tight integration into the cluster software, wherever possible. Cluster Extension XP is a resource to the clustered application service (like the disk or volume group) and must therefore be managed as such. The architecture of Cluster Extension XP allows integration into many cluster software products, including these:

•VERITAS Cluster Server (VCS)

•IBM HACMP

•Windows 2000 Advanced Server and Datacenter Server Cluster service

•Windows Server 2003 Enterprise Edition and Datacenter Edition

•Serviceguard for Linux (SG-LX)

For the current list of supported cluster software, contact your HP representative.

20	HP StorageWorks Cluster Extension XP User Guide

Disaster tolerance through geographical dispersion

Using two or more disk arrays with Continuous Access XP allows you to copy your most valuable data to a remote data center. Cluster Extension XP provides the cluster software with a mechanism to check and allow data access (in case the local application service must be transferred to a remote cluster system). The distance to your remote location is only limited by the technology your cluster software uses to communicate with each system in the cluster, the technology you use for physical data replication, and the degree of failover automation.

Disaster tolerance considerations

Application availability is essential for today’s businesses. The capability to restore the application service after a failure of the server, storage or the whole data center in a timely fashion is a must and is considered as disaster tolerance. Complete data center failures can be caused by earthquakes or hurricanes but more often they are caused by power outages or fires.

To protect against such disasters, a single data center is not sufficient. Systems (storage as well as the servers) must be geographically distributed in order to build a disaster tolerant architecture which protects against planned and unplanned downtimes.

Of course, redundant network cards and storage host bus adapters are a basic requirement. The same applies for the power supplies of both the storage and the server. With this hardware in place, the external power service and the network must also be designed to provide no single point of failure (SPOF).

Today, data is the most valuable asset in your enterprise. The XP family of disk arrays provides a fully redundant architecture, and the flexibility to upgrade firmware online reduces the risk of unplanned and planned downtime. The disk array also provides the feature of remotely mirroring your data to a second disk array.

Cluster Extension XP features

21

To have this expensive hardware in place must be compared to the risk of a true disaster. The costs pay off in a real disaster to ensure that the business critical applications are still accessible from another location.

Guidelines The following considerations, applied to the cluster environment, can ensure an application service survives a disaster with minimal downtime.

•geographical dispersion of hardware and applications

•redundant paths to access the network and storage

•alternative power sources

•redundant networks

•data replication

Several ways of implementing such disaster tolerant architectures are possible. All of those solutions can be covered by a clustered solution using the XP family of disk arrays and Continuous Access XP. Cluster Extension XP is needed to enable access to your critical data.

Disaster-tolerant architectures

With Cluster Extension XP, you can extend your cluster solution beyond the limitations of existing data center and campus-wide distances. Cluster Extension XP enables metropolitan-wide failover capabilities, and beyond.

Having a local disk array in each data center also means that the server does not have to write twice because Continuous Access XP mirrors each write-IO to the remote site and therefore relieves the server of the burden, preventing performance bottlenecks.

Disaster tolerant architectures using data replication over the network

Data replication over the network is a way to achieve disaster tolerance and is considered logical replication.

22	HP StorageWorks Cluster Extension XP User Guide

WAN

(IP ,AT M,

T 3,

Write-IO DWDM)

Replication

data center	data center
in	in
San Francisco	New York

Figure 1. Logical replication over networks

Logical replication uses specific host-based software to write data to local disks and also to replicate that data to a remote system connected to an attached storage device. Because data is replicated over the network, there is no distance limitation for such solutions.

Logical replication techniques imply that the failover process is mainly manual. This means each site belongs to a different cluster, or only the primary site is clustered, while the secondary site acts as a standby system. It is also possible that no cluster software is involved and that only one system is available at each site.

Data replicated over the network can be at the granularity of a single volume, a file system, or a transaction.

All logical replication techniques have some significant disadvantages: The remote system is a standby system. That is, it must perform the same task as the primary system and cannot be used for any other purpose. If the standby system is activated, it must replay redo logs first and cannot automatically serve as a replication source (for example, Oracle’s standby database implementation).

Another significant disadvantage of such architectures is that the server must write every IO twice, once to the attached storage device and once to

Cluster Extension XP features

23

the remote system over the network. These replication techniques can only be implemented asynchronously; otherwise, the application experiences noticeable performance degradation.

Because of the nature of replication products, additional CPU power is necessary to mirror write requests.

Logical replication implies that all logs, which have not been shipped (or which are in transit) are lost in case of a disaster.

Disaster-tolerant architectures using Fibre Channel networks

Disaster tolerant architectures using Fibre Channel networks can be achieved by the use of physical replication.

Write-IO

Replication

data center	data center
in	in
building 1	building 2

Figure 2. Physical replication using Fibre Channel

As with logical replication products, physical replication often uses host-based software to replicate data. Here, data is written to server-attached storage devices twice. Most of today’s logical volume management products offer this feature.

Using Fibre Channel, you could use dual-attached storage devices, where one port is connected to the local server and one is connected to the remote server. To be able to access your data at the remote location in case of a disaster, each server must have a local and a remote storage device. The

24	HP StorageWorks Cluster Extension XP User Guide

volume management software, then, must be set up to mirror each write request to both the local and the remote storage device. With the XP disk arrays, several servers can be connected to each disk array.

This solution is called campus cluster. A single cluster can be used and the failover process can be automated. With campus clusters, both sites can be active.

Data replicated via volume mirroring is based on the granularity of a single volume.

Campus cluster solutions are limited to the distance Fibre Channel supports today. While storage systems must be in a range of 500 meters (direct connect) or up to 10 kilometers (connected via Fibre Channel switches or Fibre Channel hubs), campus cluster solutions can only offer limited protection against natural disasters.

Another limiting factor is the cluster heartbeat protocol or the communications protocol used for cluster reformation processes. Those protocols are vendor-specific implementations and require private networks. This means, those protocols are not routable. The distance limitations of a private network depends on the supplied network infrastructure and latency issues of the heartbeat or cluster reformation protocol.

Another significant disadvantage of such architectures is that the server must write every IO twice: once to the locally attached storage device and once to the remote attached storage device. These replication techniques are implemented as software running on the server, which reduces the available compute power and degrades server performance.

Because of the nature of volume mirroring products, additional CPU power is necessary to mirror write requests across two host bus adapters.

Most of these products have another significant disadvantage. In case of a path failure, the whole volume must be copied to resynchronize the second volume with the current state of the first volume. If the storage device must be replaced, all volumes must be copied. This significantly affects server performance.

Cluster Extension XP features

25

Disaster-tolerant architectures using disk array-based mirroring

Using Continuous Access XP-based mirroring is also considered physical replication. Continuous Access XP is disk array-based mirroring. As with campus clusters, such solutions require two or more disk arrays.

The key difference from the above-mentioned solutions is that the disk array keeps track of the data integrity of the mirrored disks. XP disk arrays offer RAID-1 or RAID-5 protection as a standard feature and allow online addition and replacement of disks, IO adapter cards, and memory. To provide copies of data, internal and external mirroring features are available. For disaster tolerant solutions, Continuous Access XP can mirror your data with no distance limitation.

ESCON or Fibre Channel protocol is used to transfer data between two disk arrays. Using converters, ESCON and Fibre Channel can be routed over IP networks and T3 to allow unlimited distance between the disk arrays. To replicate data over more than 0.5 km (Fibre Channel) or 3km (ESCON), special extenders or switches must be purchased.

The cluster solutions using Continuous Access XP-based disk mirroring are called metropolitan clusters or geographically dispersed clusters. Servers are members of the same cluster dispersed over two or more sites. Since the disk array controls the replication process, the server is relieved from writing any IO-request to the disk more than once.

Continuous Access XP-mirrored disks typically have a read/write-enabled primary disk and a read-only secondary disk. This leads to problems because current cluster software products cannot distinguish between write-protected and write-enabled disks.

Cluster software assumes that the application service has access to read/write-enabled data disks on any system that the application service has been configured to run. Since the secondary volume of a disk pair is not normally accessible, the failover process would typically involve manual intervention.

26	HP StorageWorks Cluster Extension XP User Guide

	Write-IO	MAN
		(ESCON,IP
	Replication
		ATM,
		DWDM)

data center		data center
in	continuous access xp	in
Manhattan		Brooklyn
	(extension)

Figure 3. Physical replication using HP StorageWorks Continuous Access XP

Cluster Extension XP provides the software to enable automated failover and failback procedures integrated as a resource of the application service. Cluster Extension XP uses an internal database to decide whether the data on the failover site is safe to be accessed or not. Manual intervention is required if the current disk array states and the user settings conflict with the rules stored in the Cluster Extension XP internal database.

The limiting factor of metropolitan or geographically dispersed clusters is the cluster heartbeat protocol or the communications protocol used for the cluster reformation processes. Those protocols are vendor-specific implementations and typically require private networks. These protocols are not routable; a router cannot be used. The distance limitations of networks supporting a private network are dependent on the supplied network infrastructure and latency issues of the heartbeat or cluster reformation protocol.

To address these issues, cluster manager software can be used. This software offers disaster tolerance by managing two or more clusters from a single console or server and is considered a continental cluster. Depending on the implementation, automated or semiautomated failover processes between clusters are possible.

Cluster Extension XP features

27

As mentioned above, metropolitan or geographically dispersed clusters as well as continental clusters require metropolitan area networks or wide area networks. In most cases, those network connections involve common carriers and special network equipment which can be very expensive. The reliability of a direct connection or a campus network can be degraded and involves more planning to deploy and maintain a disaster tolerant environment.

Using Continuous Access XP, data is accessible and consistent in every failover case and the resynchronization of a completely failed disk array can be done while the application is running with almost no impact to the server performance. This allows reestablishing disaster tolerance without application downtime.

28	HP StorageWorks Cluster Extension XP User Guide

Automated redirection and monitoring of mirrored Continuous Access XP pairs

Disk arrays with Continuous Access XP provide a unique feature that allows the redirection of the mirroring destination. This means Continuous Access XP almost instantaneously swaps the primary/secondary relationship of disk pairs if the application must access the secondary disk. This feature ensures that the disk pairs are always synchronized, ensuring that the failback process is as fast as the failover process. If the links between your disk arrays are broken, each array maintains a bitmap table to synchronize the changed, delta data if the links become available again. In a failover case, Cluster Extension XP takes the appropriate action for each link/array status and makes sure that your application service has the latest data.

Cluster Extension XP includes a pair/resync monitor to monitor the health of the links between your arrays. Furthermore, it detects a lost and later reestablished link and automatically resynchronizes the suspended disk pairs, ensuring the most current data is available on either site.

Cluster Extension XP features

29

Rolling disaster protection

Rolling disaster protection minimizes the impact of downtime and ensures data integrity during recovery operations. Rolling disaster protection combines Continuous Access XP remotely mirrored disk pairs and internal Business Copy XP disk copies to protect data locally as well as remotely. In combination, these features support the highest data protection levels to prevent disastrous loss of data.

What is a rolling disaster?

A rolling disaster refers to catastrophic events or outages that affect the data stored on remote mirrored disk pairs. In a rolling disaster, data stored on remote mirrored disk pairs can be entirely lost during a recovery attempt.

In a rolling disaster, the mirrored disk pairs typically experience the following sequence of events:

1.The primary data center failed.

The cluster software successfully transferred application execution to the remote data center.

2.The Continuous Access XP link failed.

3.The secondary volume of the disk pair is used to continue operation after failover while the CA link is not functional.

The secondary volume represents the latest state of data, whereas the data on the primary volume is now out of date.

4.The primary data center is recovered and the Continuous Access XP link is restored.

30	HP StorageWorks Cluster Extension XP User Guide