EMC RecoverPoint/Cluster Enabler Plug-in...EMC Corporation Corporate Headquarters: Hopkinton, MA...

EMC RecoverPoint/Cluster Enabler Plug-inVersion 4.0

Product GuideP/N 300-010-474

REV A02

EMC CorporationCorporate Headquarters:

Hopkinton, MA 01748-9103

1-508-435-1000www.EMC.com

2

Copyright © 2010 EMC Corporation. All rights reserved.

Published April, 2010

EMC believes the information in this publication is accurate as of its publication date. The information issubject to change without notice.

THE INFORMATION IN THIS PUBLICATION IS PROVIDED “AS IS.” EMC CORPORATION MAKES NOREPRESENTATIONS OR WARRANTIES OF ANY KIND WITH RESPECT TO THE INFORMATION IN THISPUBLICATION, AND SPECIFICALLY DISCLAIMS IMPLIED WARRANTIES OF MERCHANTABILITY ORFITNESS FOR A PARTICULAR PURPOSE.

Use, copying, and distribution of any EMC software described in this publication requires an applicablesoftware license.

For the most up-to-date regulatory document for your product line, go to the Document/Whitepaper Libraryon EMC Powerlink.

For the most up-to-date listing of EMC product names, see EMC Corporation Trademarks on EMC.com.

All other trademarks used herein are the property of their respective owners.

EMC RecoverPoint/Cluster Enabler Plug-in Version 4.0 Product Guide

Contents

Chapter 1 About Cluster EnablerCluster Enabler overview ................................................................ 16

Cluster Enabler plug-in architecture........................................18Cluster Enabler components .....................................................19

Cluster Enabler Manager interface................................................. 20The Cluster Enabler Manager window ...................................20The Cluster Enabler Manager wizards....................................21

Cluster Enabler logging ................................................................... 22Disk space requirements............................................................23Extracting logs.............................................................................23Changing the logging level .......................................................24Changing the logging directory................................................24Changing logging retention period..........................................25Changing the maximum log file size .......................................25Windows event log messages ...................................................26

Microsoft Windows Server support ............................................... 28Quorum model support.............................................................28Multiple CE cluster management.............................................30Setting up devices on Windows Server 2008..........................30

Virtualization support...................................................................... 32Hyper-V support.........................................................................32

Supported functionality ................................................................... 34Delay Failback .............................................................................34Mount point support..................................................................37Multiple storage array support.................................................38Delegating CE administration ..................................................38

EMC RecoverPoint/Cluster Enabler Plug-in Version 4.0 Product Guide 3

Contents

Chapter 2 About RecoverPoint/ Cluster EnablerRecoverPoint/Cluster Enabler plug-in overview........................ 42RecoverPoint overview.................................................................... 44

Command line interface (CLI) .................................................. 44RecoverPoint Management Application ................................. 44

Chapter 3 Clustering ConceptsMicrosoft Failover Clusters ............................................................. 46

Microsoft Failover Cluster concepts ........................................ 48Microsoft Failover Cluster modes of operation ..................... 50

CE geographic cluster system......................................................... 51Cluster Enabler modes of operation ........................................ 53Cluster behavior and Failover operations .............................. 55

Application software in a cluster environment............................ 56

Chapter 4 Cluster BehaviorCluster failover operation................................................................ 58

Disaster scenarios and recovery actions.................................. 60Complete site failure and recovery.......................................... 63

Response to complete site failure ................................................... 65Failure behavior when using MNS with File Share Witness...... 67

Chapter 5 RecoverPoint/CE InstallationInstallation overview........................................................................ 70

Before you begin ......................................................................... 71Getting started with RecoverPoint/CE................................... 72

Installing the RecoverPoint/CE plug-in module......................... 74Uninstalling the RecoverPoint/CE plug-in module.................... 75

Uninstalling the plug-in from some cluster nodes ................ 75Uninstalling the plug-in from all cluster nodes/deconfiguring the cluster........................................................... 76Uninstalling the plug-in from all cluster nodes/destroying the cluster................................................................. 76

Chapter 6 Using Cluster Enabler ManagerGetting started using the CE Manager .......................................... 78

The Cluster Enabler Manager window ................................... 78Cluster Enabler wizards ............................................................ 79

EMC RecoverPoint/Cluster Enabler Plug-in Version 4.0 Product Guide4

Contents

Using the CE Configuration Wizard .............................................. 80Adding nodes ..............................................................................83

Managing a CE cluster...................................................................... 84Storage Discover Wizard ...........................................................84Update Mirrored Pairs Wizard .................................................85Change Quorum Model Wizard...............................................86

Managing a CE cluster group.......................................................... 89Create Group Wizard .................................................................89Modify Group Wizard................................................................91Deconfigure a CE group ............................................................93Delete a CE group .......................................................................93

Storage component ........................................................................... 94Adding and removing devices from a group .........................96

Viewing information......................................................................... 97Displaying group information ..................................................97Displaying node information ..................................................100Displaying site information.....................................................103

Restore and recovery operations .................................................. 106

Appendix A Base Component Installation and UpgradeInstallation overview ...................................................................... 108Before you begin.............................................................................. 109Installing the Cluster Enabler Base component.......................... 110

Installing the Base component separate from the plug-ins(clean install)..............................................................................110Installing the Base component together with the plug-ins(clean install)..............................................................................110Installing the Base component together with plug-ins whileupgrading...................................................................................111

Uninstalling the Cluster Enabler Base component..................... 112Uninstalling the Base component from some cluster nodes.....112Uninstalling the base component from all clusternodes/deconfigure the cluster................................................113Uninstalling the base component from all clusternodes/destroy the cluster........................................................113

Configuring CE cluster on R2 Server Core.................................. 114Requirements and considerations ..........................................114R2 Server Core configuration ..................................................114

Upgrading Windows Server 2008 to Windows Server 2008 R2 116

Glossary

5EMC RecoverPoint/Cluster Enabler Plug-in Version 4.0 Product Guide

Contents


Title Page

Figures

1 Overview example of a typical CE cluster configuration......................... 172 Cluster Enabler Manager window............................................................... 203 Lateral and peer nodes .................................................................................. 354 RecoverPoint/CE cluster configuration...................................................... 435 A typical two-node Microsoft failover cluster ........................................... 476 A typical four-node Microsoft failover cluster .......................................... 487 A geographically distributed two-node CE cluster................................... 518 A geographically distributed four-node CE cluster.................................. 529 Two-node two-cluster CE configuration .................................................... 5410 A typical RecoverPoint/CE configuration ................................................. 5911 Types of complete site failure....................................................................... 6312 Lateral and peer nodes .................................................................................. 6613 MNS clusters with file share witness .......................................................... 6714 Cluster Enabler Manager window............................................................... 7815 CE Manager expanded navigation tree....................................................... 8316 Windows Server 2003 supported quorum models.................................... 8717 Change quorum model.................................................................................. 8718 Cluster Type changed to Node Majority .................................................... 8819 Create Group Wizard, Select Group Policy................................................ 9020 Modify Group Wizard, Select Devices........................................................ 9221 Example of RecoverPoint/CE storage array view .................................... 9422 CE Manager storage actions ......................................................................... 9623 CE Manager Groups component ................................................................. 9724 CE Manager groups information ................................................................. 9825 CE Manager Nodes component ................................................................. 10026 CE Manager node information................................................................... 10127 CE Manager Sites component..................................................................... 10328 CE Manager RecoverPoint site information............................................. 104


Figures


Preface

As part of its effort to continuously improve and enhance the performanceand capabilities of the EMC product line, EMC periodically releasesrevisions of its hardware and software. Therefore, some functions describedin this document may not be supported by all revisions of the software andhardware currently in use. For the most up-to-date information on productfeatures, refer to your product release notes.

If a Cluster Enabler feature does not function properly or does not functionas described in this guide, please contact the EMC Customer Support Centerfor assistance.

Audience This guide is part of the EMC Cluster Enabler for Microsoft FailoverClusters documentation set and is intended for use by systemadministrators during installation, system setup, and routineoperations.

System administrators working with Cluster Enabler must beproficient in the use of the following products:

◆ Microsoft products:

• Windows Server 2003 or 2008 Enterprise and DatacenterEditions, as installed

• Windows Server 2008 R2 or Windows Server 2008 R2 ServerCore Enterprise and Datacenter Editions, as installed

• Microsoft Failover Clusters or server clusters as used inMicrosoft Cluster Server (MSCS)


10

Preface

◆ EMC RecoverPoint, as per your Cluster Enabler product versionand the following applicable software:

• RecoverPoint Management Application

• RecoverPoint CLI

• EMC PowerPath, if installed

Requireddocumentation

The following documentation is part of the EMC Cluster Enabler forMicrosoft Failover Clusters documentation set, and is required forRecoverPoint/Cluster Enabler:

◆ EMC RecoverPoint CLI Reference Guide◆ EMC RecoverPoint Administrator’s Guide

Related third-partydocumentation

The following Microsoft documentation available at microsoft.comcontains information about or related to the products discussed inthis guide:

◆ Windows Server 2008 Clustering Whitepapers, containing variouswhitepapers and datasheets overviewing Windows Server 2008Clustering.

Relateddocumentation

The following documentation from EMC Corporation containsinformation that may be helpful in a Cluster Enabler environment.

EMC PowerPath

◆ EMC PowerPath Product Guide


Preface

Conventions used inthis document

EMC uses the following conventions for special notices.

Note: A note presents information that is important, but not hazard-related.

CAUTION!A caution contains information essential to avoid data loss ordamage to the system or equipment. The caution may apply tohardware or software.

IMPORTANT!An important notice contains information essential to operation ofthe software. The important notice applies only to software.


12

Preface

Typographical conventionsEMC uses the following type style conventions in this document:

Normal Used in running (nonprocedural) text for:• Names of interface elements (such as names of windows,

dialog boxes, buttons, fields, and menus)• Names of resources, attributes, pools, Boolean expressions,

buttons, DQL statements, keywords, clauses, environmentvariables, filenames, functions, utilities

• URLs, pathnames, filenames, directory names, computernames, links, groups, service keys, file systems, notifications

Bold: Used in running (nonprocedural) text for:• Names of commands, daemons, options, programs,

processes, services, applications, utilities, kernels,notifications, system call, man pages

Used in procedures for:• Names of interface elements (such as names of windows,

dialog boxes, buttons, fields, and menus)• What user specifically selects, clicks, presses, or types

Italic: Used in all text (including procedures) for:• Full titles of publications referenced in text• Emphasis (for example a new term)• Variables

Courier: Used for:• System output, such as an error message or script• URLs, complete paths, filenames, prompts, and command

syntax.

Courier bold: Used for:• Specific user input (such as commands)

Courier italic: Used in procedures for:• Variables on command line• User input variables

< > Angle brackets enclose parameter or variable values supplied bythe user

[ ] Square brackets enclose optional values

| Vertical bar indicates alternate selections - the bar means “or”

{ } Braces indicate content that you must specify (that is, x or y or z)

... Ellipses indicate nonessential information omitted from theexample


Preface

Where to get help EMC support, product, and licensing information can be obtained asfollows.

Product information — For documentation, release notes, softwareupdates, or for information about EMC products, licensing, andservice, go to the EMC Powerlink website (registration required) at:

http://Powerlink.EMC.com

Technical support — For technical support, go to EMC CustomerService on Powerlink. To open a service request through Powerlink,you must have a valid support agreement. Please contact your EMCsales representative for details about obtaining a valid supportagreement or to answer any questions about your account.

Your comments Your suggestions will help us continue to improve the accuracy,organization, and overall quality of the user publications. Please sendyour opinion of this document to:

[email protected]


http://powerlink.emc.com

14

Preface


1Invisible Body Tag

This chapter provides a high-level overview of clustering andexplains how EMC Cluster Enabler provides disaster recoveryprotection in geographically distributed Microsoft Failover Clusters.

Important: EMC recommends reading this chapter in its entirety before installingand configuring Cluster Enabler for Microsoft Failover Clusters.

◆ Cluster Enabler overview ................................................................. 16◆ Cluster Enabler Manager interface.................................................. 20◆ Cluster Enabler logging .................................................................... 22◆ Microsoft Windows Server support ................................................ 28◆ Virtualization support ....................................................................... 32◆ Supported functionality .................................................................... 34

About Cluster Enabler

About Cluster Enabler 15

16


Cluster Enabler overviewCluster Enabler (CE) for Microsoft Failover Clusters is a softwareextension of failover clusters functionality. Cluster Enabler allowsWindows Server 2003 and 2008 (including R2) Enterprise andDatacenter editions running Microsoft Failover Clusters to operateacross multiple connected storage arrays in geographicallydistributed clusters. In Windows Server 2003, the failover clusters arecalled server clusters and use Microsoft Cluster Server (MSCS). Eachcluster node is connected through a storage network to the supportedstorage arrays. The method of automatic failover for mirrored pairsduring a node failure depends on the storage environment.

CE software supports the following replication technologies:

◆ SRDF®/Cluster Enabler for Microsoft Failover Clusters (forSymmetrix® storage arrays)

◆ MirrorView™/Cluster Enabler for Microsoft Failover Clusters(for CLARiiON® storage arrays)

◆ RecoverPoint/Cluster Enabler for Microsoft Failover Clusters (formultiple RecoverPoint-supported storage arrays)

Note: Refer to the EMC Powerlink website for CE version 4.0 plug-insoftware module availability for your replication technology or check withyour EMC sales representative.

Once configured using the EMC Cluster Enabler Manager graphicuser interface (GUI), Microsoft Failover Clusters are referred to as CEclusters.

Cluster Enabler expands the range of cluster storage andmanagement capabilities while ensuring full business continuanceprotection. An iSCSI or Fibre Channel connection from each clusternode is made to its own storage array. Two connected storage arraysprovide automatic failover of mirrored volumes during a Microsoftfailover cluster node failover.



This connection effectively extends the distance between clusternodes (depending on network latency) and forms a geographicallydistributed cluster (stretch cluster) with disaster-tolerantcapabilities.1

Figure 1 provides an example of a typical Cluster Enablerconfiguration. There are two hardware sites. Primary Site A has astorage array connected to Microsoft Cluster Servers and SecondarySite B has a storage array connected to another set of MicrosoftCluster Servers. The Microsoft Cluster Servers are connected via aLocal Area Network (LAN) connection and the Storage arrays areconnected via the storage array’s links.

Cluster Enabler protects data from the following types of failures, 24hour a day, 7 days a week, and 365 days per year:

◆ Storage failures◆ System failures◆ Site failures

Figure 1 Overview example of a typical CE cluster configuration

1. The EMC Networked Storage Topology Guide provides additionalinformation regarding distance restrictions for specific configurations.

Cluster Enabler overview 17

18


Cluster Enabler plug-in architectureEMC Cluster Enabler version 4.0 for Microsoft Failover Clustersprovides a new plug-in architecture consisting of a CE base moduleand separately available plug-in modules. Each CE plug-in modulesupports a different storage replication technology as follows:

◆ SRDF®/Cluster Enabler for Microsoft Failover Clusters (forSymmetrix storage arrays)

◆ MirrorView™/Cluster Enabler for Microsoft Failover Clusters(for CLARiiON® storage arrays)

◆ RecoverPoint/Cluster Enabler for Microsoft Failover Clusters (formultiple RecoverPoint supported storage arrays)

Select your plug-in module based on your storage environment’srequirements. The new Cluster Enabler architecture supports thecoexistence of multiple plug-ins, which can be installed on the samecluster node.

Note: You cannot mix replication technologies and storage configurationswithin the same cluster group. For example, Symmetrix® SRDF andCLARiiON MirrorView devices cannot be part of the same CE cluster group.

The base module must be installed prior to installing a plug-inmodule. Appendix , “Base Component Installation and Upgrade,”provides detailed installation instructions for the CE BaseComponent.




Cluster Enabler componentsCluster Enabler integrates Microsoft Failover Cluster software withreplication technology software and supported storage hardware,allowing the seamless use of disks to function as a single SCSI disk.Cluster Enabler achieves this using several components:

◆ CE Manager — An MMC-based (Microsoft ManagementConsole) user interface allows you to configure operationalparameters and perform cluster tasks.

◆ CE Resource DLL— A Dynamic Link Library (DLL) is used byMicrosoft failover cluster to perform group failover/failbackoperations for all group resources.

◆ CE WMI provider — A Windows Management Instrumentationcomponent that interfaces with the underlying storage array andperforms various operations such as failover, group creation, andso on, on the storage array.

◆ CE Service — A plug-in module used for quorum device failoverand to manage the preferred owners list.

◆ Quorum Filter Driver — A component that performs arbitrationor ownership protocol for the Microsoft Failover Cluster databasequorum.

Cluster Enabler documentationEMC Cluster Enabler version 4.0 product documentation consists ofan integrated online help system and the following documents:

EMC Cluster Enabler Base Component:◆ EMC Cluster Enabler Base Component Release Notes

EMC SRDF/Cluster Enabler Plug-in:◆ EMC SRDF/Cluster Enabler Plug-in Release Notes

◆ EMC SRDF/Cluster Enabler Plug-in Product Guide

EMC RecoverPoint/Cluster Enabler Plug-in:◆ EMC RecoverPoint/Cluster Enabler Plug-in Release Notes

◆ EMC RecoverPoint/Cluster Enabler Plug-in Product Guide

Note: Additional related documentation is provided with each replicationtechnology plug-in module.

Cluster Enabler overview 19

20


Cluster Enabler Manager interfaceVersion 4.0 of Cluster Enabler for Microsoft Failover Clustersprovides a graphic user interface called Cluster Enabler Manager. TheCE Manager provides several wizards to streamline cluster tasks andreduce the complexity of typical cluster management.

The CE Manager allows you to configure your Microsoft FailoverClusters (or MSCS clusters) for disaster recovery protection. The CEManager allows you to set up and configure disk-based resources toautomatically move geographically dispersed resource groups backand forth.

The Cluster Enabler Manager windowThe CE Manager window shown in Figure 2 contains a menu bar,two views, and a navigation tree. After cluster configuration, thenavigation tree can be expanded to show four separate components:Groups, Storage, Sites, and Nodes.

Figure 2 Cluster Enabler Manager window



The Cluster Enabler Manager wizardsThe CE Manager provides several wizards to assist you in completingvarious cluster tasks. Wizards are a series of dialog boxes that stepyou through the completion of a complex task. The first step towardsmanaging disaster recovery for distributed failover clusters is to runthe Configuration Wizard to configure a CE cluster.

Table 1 lists the various wizards that are included in the CE Manager.

Table 1 Cluster Enabler Manager wizards

Wizard Functionality

Configuration Wizard Used to configure a CE cluster. The configuration process is the firststep towards managing disaster recovery for distributed failoverclusters. The Configuration Wizard steps you through the process ofconfiguring your failover cluster for management with CE.

Create Group Wizard Used to create a CE Group, add devices, and select a group policy.

Modify Group Wizard The Modify Group Wizard steps you through the process of adding orremoving devices in a CE group.

Recover CE Cluster Wizard Used to recover a disk-based quorum cluster.

Note: The Recover CE Cluster Wizard feature is for shared quorumcluster recovery only and is not supported in RecoverPoint/ClusterEnabler version 4.0.

Change Quorum Wizard Used to change a cluster's quorum model type.

Update Mirror Pairs Wizard Used to discover storage, update the storage configuration, validatethe storage groups, and sett up the storage group definitions in thecluster properties database to update the mirrored pairs in a cluster.

Storage Discovery Wizard Used to automatically discover and set up the attached storage. Thestorage discovery process should be performed after any changeshave been made to the storage configuration.

Cluster Enabler Manager interface 21

22


Cluster Enabler loggingCluster Enabler V4.0 provides detailed logging features andimplements a simplified extraction process for extracting log filedata. If there is a problem with Cluster Enabler, detailed logs provideEMC Customer Support with the technical information necessary tohelp diagnose the problem and help Cluster Enabler engineers withdebugging.

Cluster Enabler incorporates various logging capabilities to createapplication detail logs. The amount of detail that these logs contain iscontrolled by the logging level. You can adjust the logging level tosuit your needs. Refer to “Changing the logging level” on page 24 formore information. Under normal operating conditions, error,warning, and information entries will be written to the applicationdetail log. When verbose logging is enabled, these logs containenough information to help developers diagnose various applicationfailures.

By default, logs are stored in the C:\ProgramFiles\EMC\Cluster-Enabler\Logs directory. The latest log file isnamed ce_event_trace_current.txt.

The stored logs are saved as text files and can be viewed using anytext editor. Note that the current log file is an active file and thereforemay not contain a complete set of log entries, as some may still be inprocess. Some text editors may not be able to access the current logfile. To obtain a complete copy of the current log file, you can use theCE_EventTraceDump.exe program. “Extracting logs” on page 23provides more detail and some extraction examples.

By default, when the log file exceeds 100 MB in size, it will be closedout and renamed from ce_event_trace_current.txt toce_event_trace_yyyymmddhhmmss.txt, where yyyymmddhhmmss isthe current date and time. The maximum file size of the log iscontrolled by a registry key entry and can be changed. “Changing themaximum log file size” on page 25 provides more information.

To help manage logging disk space, older log files are automaticallypurged. By default, 7 logs are saved. You can control how many logsare saved. “Changing logging retention period” on page 25 providesmore information.



Disk space requirementsThe amount of disk space required depends on the logging level andthe amount of cluster activity taking place. As a general guide, youmight expect 50 KB per day for a logging level of normal. If thelogging level is set to verbose, and cluster activity is greater thannormal, you might expect 200 MB or more per day.

Extracting logsTo extract a current log file, type CE_EventTraceDump.exe from thecommand line of the working directory. This will extract the dumpfile to the designated log directory and name itce_event_tract_yyymmddhhmmss.txt, where yyyymmddhhmmss isthe current date and time. You can use the -o filename option tochange the name of the output file to your desired name.

Examples Each of the following examples assume that the current workingdirectory is C:\Program Files\EMC\Cluster-Enabler and that thedefault log directory is C:\ProgramFiles\EMC\Cluster-Enabler\Logs.

Example 1To extract the dump file to the log directory and name itce_event_trace_yyyymmddhhmmss.txt, enter:

CE_EventTraceDump.exe

Example 2To extract a named dump file to a particular location, use thefollowing format:

CE_EventTraceDump.exe -o C:\filename.txt

Note: Use the -h option to display usage information.

Cluster Enabler logging 23

24


Changing the logging levelThe logging level is controlled by a registry key. To change thelogging level, follow these steps:

1. Open a command prompt and enter:

regedit

2. Edit the registry key value for:

HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CE\Configuration\EventTraceLevel

By default the level is set to 4. At this level error, warning, andinformational messages will appear in the log file. To createverbose logs, you can change the value to a 5. At this level error,warning, informational, and verbose messages will be sent to thelog file. Be aware that changing this level to a 5 will dramaticallyincrease the amount of data that is sent to the log file. Thece_eventrace service does not need to be restarted. The newvalue will take effect almost immediately.

Changing the logging directoryThe logging directory is controlled by a registry key. To change thelogging directory, follow these steps:


net stop ce_eventrace

2. Then enter:

regedit


HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CE\Configuration\EventTraceDirectory

Your edited path must have a trailing backslash and must existbefore you make this change.

4. Then enter:

net start ce_eventtrace



Changing logging retention periodThe log retention period is controlled by a registry key. To change thelog retention period, follow these steps:


regedit


HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CE\Configuration\EventTraceLogRetention

The DWORD value represents the number of logs to keep. Thece_eventrace service does not need to be restarted. The new valuewill take effect almost immediately.

Changing the maximum log file sizeThe maximum log file size is controlled by a registry key. To changethe maximum log file size, follow these steps:


regedit


HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CE\Configuration\EventTraceFileMaxSize

The DWORD value represents the file size in MBs. The ce_eventraceservice does not need to be restarted. The new value will takeeffect almost immediately.


26


Windows event log messagesThe Windows event log displays descriptive event messages for someof the more common events encountered when using Cluster Enabler.Table 2 lists the event log messages by event ID, Event type,description, and the action that should be taken when the specificevent has been encountered.

Note: Event types are described as an error, warning or information.

Table 2 Windows event log messages (page 1 of 2)

EventID Event type Description Action

1 Informational Generic ID used to report informationalmessages.

Action will vary based on description text.

2 Warning Generic ID used to report warning messages. Action will vary based on description text.

3 Error Generic ID used to report error messages. Action will vary based on description text.

4 Informational Informational message generated when a groupcomes on line successfully.

No action necessary.

5 Error Error message generated when a group fails tocome online.

The description text will indicate the name of thegroup that failed to come online. Look at theprevious event log messages and application logsto find the root cause of the failure.

6 Error An unexpected application error occurred. 1. Attempt the action again.2. Turn on verbose logging (logging level 5) and

attempt again.3. If failure occurs again, save the Windows

event log and the CE application log, andcontact EMC support.

7 Error The link between the storage arrays is down forstorage group (GroupName).

Use storage array CLI interfaces to determine theroot cause of the problem.

8 Informational The link between the storage arrays is replicatingdata to the remote storage array.

No action necessary.

9 Error Communication or data access to the WMI(Windows Management Instrumentationcomponent) service failed.

1. Read the event log messages and applicationlogs to find the root cause of the problem.

2. If failure occurs again, save the Windowsevent log and the CE application log, andcontact EMC support.



10 Error A failure occurred while reading or writing storagegroup information.

1. Attempt the action again.2. Turn on verbose logging (logging level 5) and



11 Error A failure occurred while reading or writing storagegroup information to the cluster registry.

1. Attempt the action again.2. Turn on verbose logging (logging level 5) and



12 Error A failure occurred while deleting a mirror group. Read the event log messages and applicationlogs to find the root cause of the problem.

13 Error A failure occurred while creating a mirror group. Read the event log messages and applicationlogs to find the root cause of the problem.

Table 2 Windows event log messages (page 2 of 2)

EventID Event type Description Action


28


Microsoft Windows Server supportEMC Cluster Enabler for Microsoft Failover Clusters version 4.0 issupported on Microsoft Windows Server 2003 or 2008 systems,including Windows Server 2008 R2 and Core editions.

There are three Windows processor architectures that are supported:

• x86• x64 (AMD64 and Intel EM64T)• IA64 Itanium

Note: Microsoft does not support mixed architecture clusters. All nodesmust be the same Windows architecture.

Appendix , “Base Component Installation and Upgrade,” providesinstallation instructions and lists the prerequisites and requirementsfor supported Microsoft Windows Servers.

Quorum model supportQuroum model support and changing the cluster model type of acluster depends on your chosen CE plug-in module and storagereplication technology. Cluster Enabler provides a wizard forchanging the quorum model of a cluster. The Change QuorumWizard will step you through the process of changing a cluster’squorum model type. “Supported model type descriptions” providesdetailed descriptions of each model.

Supported model type descriptionsMicrosoft has introduced the following new terminology andquorum options for Windows Server 2008:

◆ Shared Quorum is now referred to as No Majority: Disk Only

◆ Majority Node Set (MNS) is now referred to as Node Majority

◆ MNS with File Share Witness is now referred to as Node and FileShare Majority

◆ A new mode on 2008 is Node and Disk Majority



Shared Quorum or No Majority: Disk Only — This quorum modelcan sustain failures of all nodes except one (if the disk is online). Inthe event of a quorum disk failure using this quorum model, theentire cluster would shut down if the quorum disk is lost.

Note: This quorum model is not supported in RecoverPoint/Cluster Enablerversion 4.0.

Majority Node Set or Node Majority — This cluster model isrecommended for clusters with an odd number of nodes. The clustercan sustain failures of half the nodes (rounding up) minus one. Forexample, a seven node cluster can sustain three node failures.

MNS with File Share Witness or Node and File Share Majority —This cluster model is recommended for clusters with specialconfigurations. It works in a similar way to Node and Disk Majority,but instead of a witness disk, this cluster uses a witness file share.Note that if you use Node and File Share Majority, at least one of theavailable cluster nodes must contain a current copy of the clusterconfiguration before you can start the cluster. Otherwise, you mustforce the starting of the cluster through a particular node.

Note: Windows Server 2003 only supports 2-node clusters with this quorummodel.

Node and Disk Majority — This cluster model is recommended forclusters with an even number of nodes. It can sustain failures of halfthe nodes (rounding up) if the witness disk remains online. Forexample, a six node cluster in which the witness disk is online couldsustain three node failures. It can sustain failures of half the nodes(rounding up) minus one if the witness disk goes offline or fails. Forexample, a six node cluster with a failed witness disk could sustaintwo (3-1=2) node failures.

Note: This quorum model is not supported in RecoverPoint/Cluster Enablerversion 4.0.

Microsoft Windows Server support 29

30


Using the Change Quorum WizardOnce your Microsoft cluster has been configured as a CE cluster, youmust use this wizard for all quorum model changes. If yourconfigured CE clusters are Shared Quorum or No Majority: DiskOnly model type, you can use this wizard to change the selectedquorum disk. You can also use this wizard to change the file share forconfigured CE clusters of Majority Node Set with File Share Witnessor Node and File Share Majority model types.

Note: To change the quorum model to "Node and File Share Majority" inWindows Server 2008, you must first update the FileShare permissions to addthe Cluster Name and allow "Change" and "Read" permissions for the fileshare. Your windows documentation provides instructions on changingpermissions for FileShare.

Multiple CE cluster managementThe Cluster Enabler V4.0 CE Manager lets you manage multiple CEclusters simultaneously, as long as all of the clusters are eitherWindows Server 2003 or 2008 clusters and are in the same domain. Tomanage the cluster, CE Manager runs under a domain administratoraccount. This account is part of local administrator group of everynode of the cluster it manages.

Note: Mixing both Windows Server 2003 and 2008 clusters in one CEManager session is not supported.

Setting up devices on Windows Server 2008On Windows Server 2008, all disks must first be added to FailoverCluster Management before they can be configured for ClusterEnabler. By default, Failover Cluster assigns all disks to a groupcalled Available Storage. You must ensure that Failover Cluster canbring these disks online before using them in Cluster Enabler.

Follow these steps to correctly set up devices on the Windows Server2008:

1. Choose the appropriate instructions from the following threescenarios, as listed below for disks shown in Available Storage:



a. If there are no disks in Available Storage, ensure that all disksto be added are write enabled on the same site (for example,site A).

b. If there are already disks in Available Storage and you want toadd more disks, ensure that all disks to be added are writeenabled on the same site where Available Storage is online.

c. If some existing disks in Available Storage are not online,move them to the site where the Available Storage is online. Ifthis does not solve the problem, then you need to do thefollowing:

– Remove those disks from Available Storage.– Move all groups and devices to the same node in Failover

Cluster. Manually move the corresponding devices toensure that devices are write enabled on the node to whichyou are moving the group.

– Evict all peer nodes.2. Ensure that you have access to the disks where they are write

enabled. If not, you must reboot and reformat them.

3. Right-click Storage in Failover Cluster Management, and selectAdd a Disk. All available disks will display. You can select disksto add to the cluster. All added disks will be in the groupAvailable Storage. Verify that all disks are online in AvailableStorage.

4. The devices should now be available for use in Cluster Enabler.

Microsoft Windows Server support 31

32


Virtualization supportCE version 4.0 supports the following virtualization tools andfeatures:

◆ Windows Server 2008 (x64) Hyper-V

◆ Windows Server 2008 R2 (x64) Hyper-V including R2 Server Core

Windows Server 2008 (x64) Hyper-V server virtualization issupported for Symmetrix and CLARiiON arrays. Once configured asa CE group using the CE Configuration wizard, groups with Hyper-Vresources display as regular device groups.

Hyper-V supportCE version 4.0 supports Windows Server 2008 (including R2)Hyper-V server virtualization. Hyper-V is installed and managed as arole under Windows Server 2008 and requires an x64-basedprocessor. CE support for Hyper-V is limited to configurationsemploying Host Clustering. Host clustering allows you to host andfailover virtual machines between nodes or sites, thereby makingthem highly available. Once configured using the CE Configurationwizard, groups with Hyper-V resources display as regular devicegroups.

The following descriptions explain the difference between hostclustering and guest clustering:

Host Clustering — With host clustering, the physical host is thecluster node. If the host stops running, all of its guests (VirtualMachines) are restarted on another physical host. Host clusteringprotects against the failure of a physical host (hardware failure of acomputer).

Guest Clustering — With guest clustering, a guest (Virtual Machine)is a cluster node, and therefore the guest runs applications that aremonitored in some way by the Cluster service, either because they aredesigned to work with clustering (cluster-aware) or because they areconfigured in the cluster as a Generic Service, Generic Application, orGeneric Script resource. With guest clustering, if either the guestoperating system or the clustered application fails, the guest can failover to another guest, either on the same host or on a different host.Guest clustering protects against failure of a cluster-aware



application on a guest as well as failure of an individual instance of aguest.

Note: CE version 4.0 only supports configurations employing host clustering.

The following listed Microsoft documentation should be consultedfor Hyper-V configuration instructions:

◆ The Hyper-V Getting Started Guide is available at:

http://technet.microsoft.com/en-us/library/cc732470(WS.10).aspx#CommunityContent

◆ The Virtualization with Hyper-V: FAQ is available at:

http://www.microsoft.com/windowsserver2008/en/us/hyperv-faq.aspx

The following steps are provided as a guide for getting started withHyper-V and CE version 4.0 for a non-CSV disk:

1. Follow the instructions provided in Microsoft’s Hyper-V GettingStarted Guide to install Hyper-V using the Server Manager.

2. Follow the instructions provided in Microsoft’s Hyper-V GettingStarted Guide to create and setup a virtual machine (guestmachine) using the Hyper-V Manager.

3. Install an operating system on the virtual machine.

4. Install the application that you want to be highly available on theoperating system of the virtual machine.

5. Using Microsoft Failover Cluster Manager, configure a failovercluster for the virtual machine resources that you just created.Consult your Microsoft Failover Cluster documentation forinstructions.

Note: Turn off the virtual machine before adding it to the cluster.

6. Bring the virtual machines online in Failover ClusterManagement.

7. Open the CE Manager and configure a CE cluster using the CEConfiguration wizard.

8. On the Current Nodes wizard page, add a second node to thecluster.

Virtualization support 33

34


Supported functionalityIn addition to the wizards noted in Table 1, “Cluster Enabler Managerwizards,” on page 21, the CE Manager provides various features thatmanage and monitor cluster, group, storage device, site, and nodeinformation. Chapter 6, “Using Cluster Enabler Manager,” providesinformation on how to use the Cluster Enabler Manager GUI tocomplete cluster management tasks.

The following sections explains some of the Base functionalityavailable with Cluster Enabler.

Delay FailbackDelay Failback capability is implemented as part of Cluster Enabler’sdefault functionality. Delay Failback automatically modifies thePreferred Owner list for each failover cluster group so that a failoverwill occur to a lateral node first, and if the lateral node is unavailable,to a peer node. Lateral nodes are defined as nodes connected to thesame storage array. Peer nodes are defined as nodes connected todifferent storage arrays, located across the link from each other, asshown in Figure 3 on page 35.

Cluster Enabler manipulates the Microsoft failover cluster(previously referred to as MSCS) Preferred Owners list whenever agroup is brought online. CE then examines the group PreferredOwners list and determines which node is the lateral node. It canthen modify the Preferred Owner list so the current node and itslateral partner are the first two in the list.

Therefore, no matter which side a group is moved to, the PreferredOwner list is modified to allow a group to fail over to a lateral node,and not fail back or fail over across the link as a first option. MicrosoftFailover Clusters only moves a group across the link as a last resort.This prevents the failover clusters from arbitrarily performing whatamounts to a failback/failover across the link in an automaticfashion. This feature delays the actual failback of a group to aprimary node, and is therefore termed delay failback.



Figure 3 Lateral and peer nodes

Delay Failback runs simultaneously on all nodes. Therefore, when agroup comes online on any node, the Preferred Owner list is updated,regardless of whether it is a primary or secondary device. Forexample, the sequence for Group x on Node 1 will be the following:

1. Delay Failback first determines if it knows the other nodes in thefour-node cluster. This information is gathered by CE duringnormal operations. If not, the default feature is bypassed becauseit is unable to differentiate between a lateral node, peer node, andso on.

2. If Delay Failback knows the other nodes, then it determines ifGroup x has come online on Node 1.

3. If Group x has come online on Node 1, the Microsoft failovercluster Preferred Owner list is modified so that Node 1 is the firstPreferred Owner, followed by the lateral node and then the peernodes.

Supported functionality 35

36


Enabling and Disabling Delay FailbackThe Delay Failback feature is enabled by default. This feature can beenabled or disabled by setting a cluster private property using thecommand line interface. In the steps shown below, the commandprompt is shown as C:\>.

1. To verify the current Delay Failback setting, issue the followingcommand at the CLI:

C:\> cluster /priv

2. Check the output for the text DelayFailBackEnabled. A value of0 means the feature is disabled. A value of 1 means the feature isenabled. If the DelayFailBackEnabled property has not beenadjusted, it will not be visible in the cluster /priv commandoutput and is assumed to be the default value of 1 (enabled).

3. To disable the Delay Failback setting, issue the followingcommand at the CLI:

C:\> cluster /priv DelayFailbackEnabled=0:DWORD

To enable the Delay Failback setting, issue the followingcommand at the CLI:

C:\> cluster /priv DelayFailbackEnabled=1:DWORD



Mount point supportCluster Enabler supports mount points. By using mount points, youcan overcome the limitation on drive letters, which makes it possiblefor a cluster to support more than 26 volumes.

For mount points to work correctly, all related disks must belong tothe same cluster group. If related disks are spread across multiplecluster groups, volumes cannot be brought online because clustergroups can be online on different nodes. To avoid this scenario,Cluster Enabler first groups all related disks by identifying the mountpoints on a given disk and any disks upon which the given disk ismounted. Cluster Enabler then creates a parent/child relationshipbetween the disks.

When a user chooses a disk to create a group (or adds a disk to anexisting group), Cluster Enabler finds all related disks by traversingits parent/child relationships and adding every related disk to thegroup. It then adds appropriate dependencies between the disks sothat the resources can be brought online in an orderly fashion.

Table 3 on page 38 illustrates a cluster example consisting of driveletters and mount points for six volumes. Using this configuration,you can see various parent/child relationships among the disks.

For example, the user chooses E:\MNT1. Therefore:

◆ E:\MNT1 is a mount point with E:\ as its parent.

◆ E:\ is a child of F:\. Thus, disk F:\will be included in the group.

◆ F:\ has additional children F:\MNT2 and F:\MNT2\MNT3.Thus, the group will include these disks too.

The result of these parent/child relationships is that the group willinclude volumes 0BCE, 0BCF, 0BD0, 0BD1, and 0BD2. Each disk isdependent on its parent to come online. In this example, 0BCF isdependent on 0BCE, and 0BD0 is dependent on 0BCE, and so forth.

Of course, each group is also dependent on the Cluster Enablerresource.


38


Table 3 Cluster mount point example

When you delete a device, Cluster Enabler finds all related disks anddeletes them too. For example, if the current mount points are F: andF:\MNT2 and F:\MNT2\MNT3, and if the device that correspondsto F:\MNT2 is deleted from the group, all three devicescorresponding to F:\, F:\MNT2, and F:\MNT2\MNT3 are deleted.

However, if you were to first delete mount point F:\MNT2 from theoperating system and then delete its corresponding device from thegroup, Cluster Enabler would delete only the devices that correspondto F:\MNT2 and F:\MNT2\MNT3. The device corresponding to F:\would be left in the group because, after the mount point deletion, itis no longer related to F:\MNT2.

Multiple storage array supportCluster Enabler for Microsoft Failover Clusters V4.0 supports the useof multiple storage arrays per cluster. This feature provides greaterflexibility to you and your storage provisioning.

Delegating CE administrationThe CE Manager lets you manage multiple CE clusterssimultaneously, as long as all of the clusters are either WindowsServer 2003 or 2008 clusters and are in the same domain. To managethe cluster, CE Manager and Cluster Administrator are used with adomain account, which is part of local administrator group on everycluster node. This effectively grants full control of every cluster nodeto the domain account used to manage the cluster.

Drive letter andmount point Symmetrix volume ID

F:\ 0BCE

F:\MNT1, E:\ 0BCF

F:\MNT2 0BD0

F:\MNT2\MNT3 0BD1

D:\ 0BCD

E:\MNT1 0BD2



CE provides a utility that allows the most common CE and clustermanagement tasks to be delegated to a non-local administrator. Tosupport this option, a command-line utility, called cesec.exe is usedon each cluster node after the cluster has been fully configured.

Using the cesec.exe utilityThe cesec.exe command-line utility allows the local administrator todelegate the most common CE cluster tasks to non-localadministrators by adding a domain group (recommended) or adomain user.

A CE cluster must have already been configured by a localadministrator using the CE Configuration Wizard. The cesec.exeutility is located in the CE install directory (typically C:\Program

Files\EMC\Cluster-Enabler) and must be run on all nodes in thecluster by a local administrator. On Windows Server 2008, the utilitymust be run from an elevated command prompt.

Note: Due to a Microsoft limitation, the Windows Server 2008 FailoverCluster Manager can not be used by a non-local administrator account even ifthat account has been granted full control of the cluster. Use thecluster.exe command-line utility instead.

System security changesRunning the cesec.exe command-line utility allows you to changethe following security administration privileges:

◆ Allows a non-local administrator to manage the cluster.

◆ Allows a user to make remote DCOM connections.

◆ Opens the Windows Firewall for the Windows ManagementInstrumentation (WMI) rule group on Windows Server 2008.

◆ Allows remote write access to the following WMI namespaces:Root/CIMV2, Root/EMC, and Root/MSCluster.

◆ Allows a user to query the Service Control Manager and tocontrol the following CE related services: Cluster Service(clussvc), CE Event Trace service (ce_eventtrace), and CEService (cesvc).

◆ Allows remote access to the CE portion of the registry(HKLM\SOFTWARE\EMC\CE).


40


◆ Allows the user to export CE log files by granting write access tothe CE log directory (typically C:\Program

Files\EMC\Cluster-Enabler\Logs).

RestrictionsCertain CE configuration operations are not allowed. The followingCE configuration changes will be blocked:

◆ CE install/uninstall

◆ Using the Configuration Wizard to convert MS clusters to CEclusters

◆ Adding and deleting nodes for an existing cluster

◆ De-configuring a CE cluster

Command syntaxThe following is the cesec.exe command syntax:

cesec.exe -ce <action> <principal>

action set | remove | list

principal domain\user | domain\group | [email protected] |[email protected]

Options:

-ce <action> <principal>

Used to modify the CE security settings for a principal.

-ce list

Used to list the security settings relevant to CE.

Usage examplesThe following examples assume that the domain users who will bemanaging CE have been added to a domain group calledDomainName\CE Admins.

To allow the domain group to manage CE, type:

cesec.exe -ce set “DomainName\CE Admins”

To remove the domain group, type:

cesec.exe -ce remove “DomainName\CE Admins”

To list your current security settings, type:

cesec.exe -ce list


2Invisible Body Tag

This chapter provides an introduction to the RecoverPoint/ClusterEnabler plug-in module and explains how EMC Cluster Enablerprovides disaster recovery protection in geographically distributedMicrosoft Failover Clusters using RecoverPoint.

◆ RecoverPoint/Cluster Enabler plug-in overview ......................... 42◆ RecoverPoint overview ..................................................................... 44

About RecoverPoint/Cluster Enabler

About RecoverPoint/ Cluster Enabler 41

42

About RecoverPoint/ Cluster Enabler

RecoverPoint/Cluster Enabler plug-in overviewRecoverPoint/Cluster Enabler (RecoverPoint/CE) version 4.0 is asoftware plug-in module to EMC Cluster Enabler for MicrosoftFailover Clusters software. Cluster Enabler (CE) version 4.0 plug-inarchitecture consists of a CE base module component and separatelyavailable plug-in modules, which support your chosen storagereplication technology. The CE base component must be installedprior to installing a plug-in module.


The RecoverPoint/CE plug-in module provides a software extensionof failover clusters functionality that allows Windows Server 2003and 2008 (including R2) Enterprise and Datacenter editions runningMicrosoft Failover Clusters to operate across multiple connectedstorage arrays in geographically distributed clusters. In WindowsServer 2003, the failover clusters are called server clusters and useMicrosoft Cluster Server (MSCS).

Each cluster node is connected through a storage network to thesupported storage arrays. Once configured using the EMC ClusterEnabler Manager graphic user interface (GUI), Microsoft FailoverClusters are referred to as CE clusters.

Important: Mixed replication technologies/storage configurations are notsupported. For example, Symmetrix® SRDF and CLARiiONMirrorView devices cannot be part of the same CE cluster group.

Cluster Enabler expands the range of cluster storage andmanagement capabilities while ensuring full business continuanceprotection. An iSCSI or Fibre Channel connection from each clusternode is made to its own storage array. Two connected storage arraysprovide automatic failover of mirrored volumes during a Microsoftfailover cluster node failover.



This connection effectively extends the distance between clusternodes (depending on network latency) and forms a geographicallydistributed cluster with disaster-tolerant capabilities. 1

Figure 4 on page 43 shows Cluster Enabler in a RecoverPointenvironment.

Cluster Enabler protects data from the following types of failures, 24hour a day, 7 days a week, and 365 days per year:

◆ Storage failures◆ System failures◆ Site failures

Figure 4 RecoverPoint/CE cluster configuration

1. The EMC Networked Storage Topology Guide provides additionalinformation regarding distance restrictions for your specificconfiguration.

RecoverPoint/Cluster Enabler plug-in overview 43

44


RecoverPoint overviewEMC RecoverPoint replicates data over any distance: within the samesite (continuous data protection, CDP), or to another site halfwayaround the globe (continuous remote replication, CRR), or both(concurrent local and remote data protection, or CLR). RecoverPointsupports replication of data over Fibre Channel to localSAN-attached storage; and over WAN/Fibre Channel to remote sites.RecoverPoint allows you to fail over to a secondary site and continueoperations in the event of a disaster at the primary site.

The RecoverPoint system automatically switches between thefollowing modes, according to load conditions: continuoussynchronous, continuous asynchronous, and snapshot. The EMCRecoverPoint Administrator's Guide provides detailed descriptions ofthe RecoverPoint modes. RecoverPoint automatically uses thereplication mode that is most effective for the current conditions,including the application load, throughput capacity, and replicationpolicy. Regardless of the replication mode, RecoverPoint is unique inits ability to guarantee a consistent replica at the target side under allcircumstances, and in its ability to retain write order fidelity inmulti-host heterogeneous SAN environments.

RecoverPoint provides both a management application and acommand line interface (CLI) that allow you to manage theRecoverPoint system.

Command line interface (CLI)Management activities can be carried out interactively or by scriptsusing the command-line interface. For information about thecommand-line interface, refer to the EMC RecoverPoint CLI ReferenceGuide.

RecoverPoint Management ApplicationManagement activities can also be carried out using any standardweb browser to access the RecoverPoint Management Application.The RecoverPoint Management Application can be used to manageand monitor RecoverPoint replication. For information about theRecoverPoint Management Application, refer to the EMCRecoverPoint Administrator's Guide.


3Invisible Body Tag

This chapter describes the various clustering concepts for MicrosoftFailover Clusters using a Cluster Enabler cluster solution and themodes of operation:

◆ Microsoft Failover Clusters .............................................................. 46◆ CE geographic cluster system .......................................................... 51◆ Application software in a cluster environment ............................. 56

Clustering Concepts

Clustering Concepts 45

46

Clustering Concepts

Microsoft Failover ClustersMicrosoft Failover Clusters is the clustering extension to WindowsServer 2008 Enterprise and Datacenter editions. This is also referredto as Microsoft Cluster Server or MSCS in Windows Server 2003Enterprise and Datacenter editions.

Microsoft Failover Clusters enables up to eight servers running thesame Windows operating system in the same domain to be connectedto a shared storage system. Typically, this is a RAID array on a sharedSCSI storage bus. With iSCSI connections, all servers must be within40 m of one other (each less than 20 m from the storage). With FibreChannel connections, you can increase the distance between the twoservers.

Note: Consult with EMC Customer Support if you need more detail.

Microsoft Failover Clusters protect against failure of productionserver hardware or network connections. For data protection,Microsoft Failover Clusters use a protected storage subsystem. Thestandard failover cluster relies on RAID 1 or RAID 5 array storage toguarantee data protection.

In a typical failover cluster containing one to eight nodes, servernodes share the application workload. Typically, in a node clusterenvironment with n nodes, each node serves one-nth of the totalnumber of disks and clients connected by a common SCSI bus. If oneserver node fails, one or several of the remaining nodes takeownership of all the disks and assume all the application workload.

Note: The number of cluster nodes supported differs between WindowsServer 2003 and 2008 Enterprise and Datacenter editions. Refer toyour Microsoft Windows Server documentation for your specificoperating system.

EMC RecoverPoint/Cluster Enabler Version 4.0 Product Guide

Clustering Concepts

Figure 5 presents a typical two-node failover cluster on WindowsServer 2003 or 2008 Enterprise and Datacenter editions.

Figure 5 A typical two-node Microsoft failover cluster

Microsoft Failover Clusters 47

48

Clustering Concepts

Figure 6 presents a typical four-node Windows Server 2003 cluster.

Figure 6 A typical four-node Microsoft failover cluster

Microsoft Failover Cluster conceptsMicrosoft Failover Cluster is a loosely coupled cluster system. It isnot a fault-tolerant, closely coupled system. The concept of a cluster isto take two or more independent computers and set them up to worktogether to provide higher availability and scalability than what youcan obtain using a single system. When failure occurs in the cluster,control of a disk, or resource, moves to another cluster node. Thisprocess is called a failover. Failovers can be initiated by a number ofevents, including the following:


Clustering Concepts

◆ Manual failover — The moving of resources from one server toanother. Done for system load balancing or for servermaintenance.

◆ Failover due to hardware failure — The surviving node takes overwhen a server, iSCSI or Fibre Channel host bus adapter (HBA)card, or network interface card (NIC) fails.

◆ Failover due to application failure — The failure of a virtual server orIP resource can initiate the failover.

By contrast, a fault-tolerant system uses special-purpose hardware torun multiple computers in lock step, which provides nonstopcomputing with no data loss when a component failure occurs.

There are benefits and limitations to using a cluster architecture.

Benefits Clustering provides:

◆ Improved availability by continuing to provide a service evenduring hardware or software failure.

◆ Increased scalability by allowing new components to be added asthe system load increases.

◆ Simplified management of groups of systems and their applications byenabling multiple applications on multiple servers to be managedas a single system.

Limitations Clustering cannot protect against:

◆ Software corruption

◆ Human-induced failures

Note: Protection of user data through backup, EMC businesscontinuance volumes (BCVs), or other forms of offline data redundancy,remains vitally important to the reliable operation of mission-criticalapplications.

Microsoft Failover Clusters 49

50

Clustering Concepts

Microsoft Failover Cluster modes of operationMicrosoft Failover Cluster supports 16 node clusters for WindowsServer 2003 or 2008 Enterprise and Datacenter Editions.

Similar to the modes of operation generally discussed for ClusterEnabler, the configuration for a failover multinode cluster in ageographically distributed cluster environment is eitheractive/passive or active/active. “Cluster Enabler modes ofoperation” on page 53 provides an example.

Availability Failover clusters allows active/active application operation. Duringnormal operation, software applications can be running on bothnodes. If either node fails, the applications are restarted on theremaining cluster node. This provides high availability byminimizing application downtime. Usually, it takes one to 10 minutesto fail over and restart an application on a Microsoft Failover Cluster.Restart time is highly application dependent.

Scalability In addition to availability protection, cluster technology is scalable.You can add new components to the system and run the sameapplication (accessing the same database) on multiple nodes of acluster to deliver increased processing power. To provide scalability,data sharing is needed.


Clustering Concepts

CE geographic cluster systemCluster Enabler provides disaster-tolerant capabilities that enable thecluster servers to be geographically separated1. Figure 7 illustrates atypical hardware configuration of a two-node CE cluster solution.

Figure 7 A geographically distributed two-node CE cluster

Network connections can provide a guaranteed maximum round-triplatency between nodes of up to 300 ms. Since many servers canconnect to one storage array, it is possible to implement many clustersacross this distance.

1. The EMC Networked Storage Topology Guide provides additionalinformation regarding distance restrictions for your specificconfiguration.

CE geographic cluster system 51

52

Clustering Concepts

Figure 8 illustrates a typical hardware configuration of a four-nodecluster solution.

Figure 8 A geographically distributed four-node CE cluster

Cluster Enabler provides disaster-tolerant capabilities by exploitingmirroring and failover capabilities. CE allows two stoprage arrays tobe attached using direct-connect fiber.


Clustering Concepts

Cluster Enabler modes of operationDifferent cluster designs support different modes of operation anddata sharing mechanisms. The configuration for a CE two-node ormultinode cluster in a geographically distributed clusterenvironment is either active/passive or active/active. EMC definesactive/passive and active/active configurations as follows:

◆ Active/Passive: A cluster of two or more nodes where allprocessing is done on one node during normal operation and thework is picked up by a remaining passive node (or nodes) onlywhen a failure occurs on the active node. In a two-nodeconfiguration, half of the hardware is normally idle. Whenfailover occurs, the application restarts with full performance.

Note: Active/passive multinode clustering provides greater flexibilitythan the standard active/passive Microsoft failover cluster two-nodecluster by providing more options in resolving failures and loaddistribution after server failures. For example, in a multinode cluster,your configuration may include one or more passive (idle) servers to takeover the load from other servers during a site failure, or you maydistribute the load among the surviving active nodes.

◆ Active/Active: A cluster of two or more nodes where all nodes arerunning application software during normal operation. When afailure occurs on a node (or nodes), the work is transferred to aremaining node (or nodes) and restarted. The node(s) that picksup the work must then handle the processing load of bothsystems, and performance is usually degraded. However, all thecomputer hardware is used during normal operation.

Note: The terms active/active and active/passive apply to the clusterand to the applications running on the cluster. Both the cluster softwareand the application software must be designed for active/activeoperation.

Figure 9 on page 54 presents a typical CE two-node two-clusterconfiguration.


54

Clustering Concepts

Figure 9 Two-node two-cluster CE configuration


Clustering Concepts

Cluster behavior and Failover operationsClusters are designed to overcome failures. There are several possiblefailure modes in a cluster configuration. Cluster Enabler protectsagainst more failure scenarios than local clusters can. Cluster Enablerprotects Microsoft Failover Clusters against disasters by providinggeographically dispersed (stretched) cluster capabilities.

Cluster behavior and recovery failover operations depend on thespecific scenario of failure, storage configuration, and version ofCluster Enabler plug-in module deployed.

Explanations of specific Cluster Enabler failover and recoverybehavior, as well as instructions for site failure and recovery actionsare provided in each EMC Cluster Enabler plug-in module productguide.


56

Clustering Concepts

Application software in a cluster environmentSoftware running on a cluster may, or may not, be cluster aware.When software is cluster aware, it provides a restart mechanism thatinvokes whenever the application resource is moved to another nodein the cluster.

Application failover requires a restart of the application wheneverfailover occurs. Restart is not instantaneous. Unlike a fault-tolerantcomputer, a distributed cluster does not provide nonstop computing.The time that restart takes, and the completeness of the recovery, isapplication dependent.

◆ For a transaction-oriented application (such as SQL or Exchangethat contain both a database and transaction log files), theapplication provides a restart mechanism to recover work inprogress. Usually a transaction log is used to record all work inprogress. When a node fails, the information in host memory islost, but the work can be reconstructed by applying thetransaction log to the database to restart. This mechanismrecovers all transactions completed before the failure.Transactions partially complete are lost and must be reentered.

◆ Applications such as Microsoft Word or Microsoft Excel provide acheckpoint capability. If the application experiences a failover, allwork since the last disk checkpoint is lost.

◆ If an application has neither a database nor checkpoint capability,and also retains no information (or state) between client requests(such as a Web Browser or a Microsoft Outlook client), then it canfail over by reissuing the outstanding request. In this scenario, nowork is lost, and no restart is needed on the server.

◆ If the application has neither a checkpoint nor restart capability,and it retains the state between client requests to the server, thenit must be rerun from the beginning when the node it is runningon fails.


4Invisible Body Tag

This chapter describes RecoverPoint/Cluster Enabler behavior invarious operational modes. Unless otherwise noted, Cluster Enablerbehavior is described for a standard two-node cluster:

◆ Cluster failover operation ................................................................. 58◆ Response to complete site failure .................................................... 65◆ Failure behavior when using MNS with File Share Witness....... 67

Cluster Behavior

Cluster Behavior 57

58

Cluster Behavior

Cluster failover operationClusters are designed to overcome failures. There are several possiblefailure modes in a cluster configuration. Cluster Enabler protectsagainst more failure scenarios than local clusters can. Failure of anindividual client affects only one user and is not discussed in thischapter.

RecoverPoint/CE protects Microsoft Failover Clusters againstdisasters by providing geographically dispersed (stretched) clustercapabilities. This chapter discusses a variety of disaster scenarios andhow to use RecoverPoint/CE stretch cluster capabilities to recoverfrom them.

Note: Microsoft Windows Server 2008 Failover Cluster replaces MicrosoftWindows Server 2003 Cluster Server. In this document, Failover Clusterrefers to both Windows 2003 Cluster Server and to Windows 2008 FailoverCluster unless the Windows version is explicitly specified.

Figure 10 on page 59 shows a typical RecoverPoint/CE configuration.

This section describes the following:

◆ “Disaster scenarios and recovery actions” on page 60

◆ “Complete site failure and recovery” on page 63


Cluster Behavior

Figure 10 A typical RecoverPoint/CE configuration

Clients

Switch

Local site Remote site

File Share Witness at 3rd Location

HeartbeatConnector

Enterprise LAN

Clients

Enterprise LAN

MicrosoftFailoverCluster

MicrosoftFailoverCluster

StorageArray

StorageArray

Switch

RecoverPoint Replication Link

RecoverPoint/CEStretch Cluster

Cluster failover operation 59

60

Cluster Behavior

The disaster and failover scenarios that follow all assume that:

◆ Both sites are fully operational (cluster nodes, RecoverPointreplication, switches, and communication links).

◆ The Microsoft Failover Cluster is configured for Majority NodeSet (MNS) mode with a file share witness at a third site.

◆ Both the WAN link and the heartbeat link are configured in theMicrosoft Failover Clusters as enabled for all network access.

Simply stated, the failover and recovery operations Cluster Enablerprovides can be divided into situations where:

◆ The behavior of Cluster Enabler is the same as Microsoft failoverlocal clusters.

◆ The geographic separation and disaster tolerance of ClusterEnabler causes unique behavior and provides recoveryalternatives.

Disaster scenarios and recovery actionsThe following scenarios describe various disasters and howRecoverPoint/CE responds.

Replication link failureIf the replication link between sites fails but the heartbeat link isfunctioning, no failover will occur. The clusters at both sites areavailable and communicate with each other over the heartbeat link.Replication between sites will stop, and at the production site,RecoverPoint will go into marking mode until the replication link isrestored. When the replication link is restored, changes to productionwill be transferred to the replication site.

Heartbeat link failureIf the heartbeat link fails, but the replication link continues tofunction, operations continue normally. Heartbeat messages will besent over the public cluster link until the heartbeat link is restored.

Replication link and heartbeat link failureIf both the replication link and the heartbeat link fail (for instance, ifthey are both transported over the same WAN, resource groups at themajority node set will continue to operate normally in RecoverPointmarking-on-host mode. Resource groups and applications at theother site will fail. The failed resource groups can be restarted by


Cluster Behavior

manually putting RecoverPoint into CE mainenance mode andfailing those groups over to the majority node set.

Cluster node failureIf a cluster node crashes or otherwise fails, its heartbeat will stop.Other nodes in the cluster will detect the lack of heartbeat and failover the resources of that node to another node. The recovery nodewill bring the resources of the failed node on-line and restore theapplications of the failed node. The failover is automatic and no dataloss occurs.

Application software failureIf an application fails, the cluster node will fail over to another clusternode. In case the failover is to a recovery node at the second site,RecoverPoint/CE resource monitor will make storage resourcesavailable for the failed application to allow application failover.Failover is automatic and no data loss occurs.

Host bus adapter failureIf a host bus adapter fails on a cluster node, other nodes in the clusterwill fail over the resources of that node to another cluster node. Therecovery node will bring the resources of the failed node on-line andrestore the applications of the failed node. The failover is automaticand no data loss occurs.

Storage failure at one siteIf the storage at one site fails, resource groups automatically fail overto the second site, but remain off-line until distribution of the latestdata to the replica is completed. Then the failed-over resources willcome on-line automatically. No data loss will occur.

RPA cluster failureIf an RPA cluster fails (all RPAs at one site fail), cluster groups at bothsites will remain operational and active. At the site where theRecoverPoint cluster failed:

◆ If any active cluster group is failed over to any other node, it willfail, and not restart.

◆ If you stop any cluster group, you will not be able to restart it.

◆ You can use RecoverPoint CE maintenance mode to manually failover any resource group to the other site.


62

Cluster Behavior

At the other site, where the RecoverPoint cluster is still operating:

◆ Resource groups will remain operational.

◆ Resource groups can be failed over to another node at this site.

RecoverPoint manual failoverIf an entire RPA cluster fails and either hosts or storage fails at thatsite, the resources cannot fail over automatically to the other site. Tofail them over manually:

1. At the RecoverPoint Management Application, in the Navigationpane, select the consistency group containing the resourcesmanaged by RecoverPoint/CE. In the Component pane, select thePolicy tab.

2. In the Stretch Cluster Support area, select Group in maintenancemode. It is managed by RecoverPoint, CE can only monitor.

3. In the Component pane, select the Status tab. From the ImageAccess menu of the operating site, select Enable Image Accessselect the latest image, and select Logged Access. Activate theFailover menu for the second site and select Failover to <secondsite>.

4. After failover is complete, select the consistency group in theNavigation pane and then select the Policy tab in the Componentpane. In the Stretch Cluster Support area, select Group ismanaged by RecoverPoint CE. RecoverPoint can only monitor.

Note: Data loss may occur if the last writes from the production site were nottransferred before the failure occurred.

File share witness failureIf the cluster nodes cannot communicate with the file share witness,no change occurs and no action is necessary. The system responds thesame way as when any other cluster node fails.


Cluster Behavior

Complete site failure and recovery

Local MS failovercluster

In a local MS failover cluster, if an entire site fails (such as from aflood, fire, and so forth) the entire cluster fails. By contrast, with a CEcluster, each site contains only one of the two nodes in the cluster (oronly one of the n nodes in a multinode cluster).

CE cluster A complete site failure can be caused by either a site failure or a totalcommunication failure. Figure 11 illustrates the two types of completesite failure.

Figure 11 Types of complete site failure

Site (server, storage and RecoverPoint) failuresSite failure occurs when the hosts, storage array, and theRecoverPoint cluster fail (such as from a natural disaster or humanerror).


64

Cluster Behavior

Total communication failureA total communication failure can occur while the host and storagearray remain operational (such as a backhoe digs up the cable conduitwhere all communications cables leave a building).

A total communication failure, while both nodes remain operational,is referred to as a split-brain condition and is a potential cause oflogical data corruption. For example, if both sides assume the other isdead and begin processing new transactions against their copy of thedata, two separate and unreconcilable copies of the data can becreated.

Both nodes are isolated from each other, but not from local clients. Itis impossible to determine if the other node is alive. No remote clientprocessing is possible, but running processes continue.

Note: There is no way for the surviving node to determine which of these twotypes of failures caused the site failure.


Cluster Behavior

Response to complete site failureIn Cluster Enabler, the site failure modes determine the behavior of acluster when a failure occurs, separating the two storage arrays andsuspending remote data mirroring protection.

If a complete site failure occurs, MS failover cluster on the survivingnode first notices that heartbeat messages are no longer beingreceived. MS failover cluster attempts to communicate with the othernode using the LAN communication path, to see if communication isstill possible.

MS failover cluster then queries the status of the disk resource anddecides whether to bring the disk resources on the local node onlineor to set them offline. The commands to perform this query from MSfailover cluster to Cluster Enabler are:

◆ Is Alive? — Determines whether a currently online resource isstill healthy and can continue to be used, or whether it and alldependent cluster resources must be taken offline.

◆ Online Request — Changes the state of an offline resource toonline for a failover.

Each group’s failover option setting determines how Cluster Enablerresponds to queries from Cluster Service. This setting must bemanually configured to select the desired failover and recoverybehavior.

Inappropriate user actions that cause groups to bounce back actdifferently. If you attempt to move the quorum group when thestorage link is down, the MS failover cluster destination nodeterminates, and the group bounces back. Active/activeconfigurations are obviously affected because any applications on thedestination node now move. This behavior is a result of the precedingbehavior.

Response to complete site failure 65

66

Cluster Behavior

Currently, the RecoverPoint/CE site failure mode setting can only beset to:

Restrict Group Movement — In a storage link failure, this settingwill only attempt to move disks laterally. Figure 12 shows lateral andpeer nodes. If the storage link is up, this setting has no impact.

Whenever a failure occurs such that mirrored data protectionbetween sites is lost (for example, the replication link is down or theRecoverPoint cluster is down), Cluster Enabler responds to the failureby not allowing any new disk groups to be brought online untilcommunication with the other node has been reestablished.

Figure 12 Lateral and peer nodes


Cluster Behavior

Failure behavior when using MNS with File Share Witness

Failure behavior andrecovery

In general, Cluster Enabler behaves similarly to a two-node clusterusing a quorum disk.

The following example explains a four-node cluster for MajorityNode Set with File Share Witness. Figure 13 provides an illustratedexample. The production nodes, Nodes 1 and 2 are at the Left site. Theremote nodes, Nodes 3 and 4, are at the Right site, and the file sharenode or server is at a third location. The cluster is configured with allthe described settings.

Figure 13 MNS clusters with file share witness

ProdProd

ProdProd

ProdProd

Failure behavior when using MNS with File Share Witness 67

68

Cluster Behavior

In the following examples, groups are cluster groups that contain oneor more Cluster Enabler managed physical disk resources. Thefailover policy has been set to Restrict Group Movement.

Storage failure at primary site

◆ Groups on Nodes 3 and 4 remain online but cannot failover.

◆ Groups on Nodes 1 and 2 automatically failover to Nodes 3 and 4but stay offline and must be brought online manually byfollowing the procedure for “RecoverPoint manual failover” onpage 62.

Replication link failure



◆ Resource groups can freely be moved within each site.

Site failure (server, storage, and replication) at primary site


◆ Groups on Nodes 1 and 2 automatically failover to Nodes 3 and 4but stay offline and must be brought online manually byfollowing the procedure for “RecoverPoint manual failover” onpage 62.

Total communication failure

◆ If all nodes have connectivity to the file share witness, the clusterwill take two of the nodes at one site offline.

◆ If only one node has connectivity to the file share witness, thecluster will take the other nodes offline.

◆ If no nodes have connectivity to the file share witness, the entirecluster will go offline. (See Microsoft procedures for forcing anMNS cluster node online.)

◆ If Nodes 3 and 4 are the surviving nodes:

• Groups on Nodes 3 and 4 remain online but cannot failover tonodes 1 or 2.

• Groups on Nodes 1 and 2 automatically failover to Nodes 3and 4 but stay offline and can be brought online by followingthe procedure for “RecoverPoint manual failover” on page 62.


5Invisible Body Tag

This chapter explains how to install and uninstall theRecoverPoint/Cluster Enabler Plug-in module.

◆ Installation overview......................................................................... 70◆ Installing the RecoverPoint/CE plug-in module .......................... 74◆ Uninstalling the RecoverPoint/CE plug-in module ..................... 75

RecoverPoint/CEInstallation

RecoverPoint/CE Installation 69

70

RecoverPoint/CE Installation

Installation overviewThis chapter describes how to install the RecoverPoint/CE forMicrosoft Failover Clusters Version 4.0 plug-in module on thesupported Microsoft Windows Server 2003 or 2008 systems. It alsodescribes how to uninstall the Cluster Enabler Version 4.0 software.

Note: RecoverPoint/CE version 4.0 supports software upgrades from Version3.x and later. Appendix , “Base Component Installation and Upgrade,”provides instructions on how to upgrade your existing supportedRecoverPoint/CE software from Version 3.x and later to Version 4.0.

It is recommended that you contact EMC Customer Support forassistance if any of the following issues are applicable:

◆ You have applications already layered with dependencies.

◆ You need other devices online.

◆ You are not confident about installing and configuring newsoftware within the context of Windows Server 2003 or 2008,Microsoft Failover Clusters, and RecoverPoint supported storagearrays.



Before you beginBefore you begin to install RecoverPoint/Cluster Enabler, you shouldread the following installation requirements and considerations:

◆ Starting with Cluster Enabler V4.0, the Cluster Enabler Basecomponent is a prerequisite for the Cluster Enabler plug-ins, andtherefore must be installed prior to or with the plug-ins. Forinstructions on installing the Base component, refer to Appendix ,“Base Component Installation and Upgrade.”

◆ The supported versions of CE that may be upgraded to ClusterEnabler Version 4.0 using the InstallShield wizard include onlyCluster Enabler for Microsoft Failover Clusters Versions 3.x andlater.

Note: For a clean install, all existing clusters will have to be reconfiguredand any unique settings in CE will be lost.

◆ There are three Windows processor architectures that aresupported:


Note: Microsoft does not support mixed architecture clusters. All nodesmust have the same Windows architecture.

◆ Installation on Windows Server 2003 and 2008 requires thatMicrosoft Windows Installer version 4.5 first be installed.

◆ Installation on Windows Server 2003 systems requires that aminimum of SP2 (Service Pack) first be installed.

◆ Installation on Windows Server 2003 requires that Microsoft .NETFramework version 2.0 first be installed.

◆ Installation on Windows Server 2008 requires that all nodes firstbe installed with the Failover Cluster feature.

◆ For Failover Cluster on Windows Server 2008, Microsoft ClusterValidate must pass all tests except storage.

◆ Installation of Cluster Enabler requires a reboot of the system forthe configuration to take effect. After installation, you may choosethe restart option or postpone your restart until a moreconvenient time.

Installation overview 71

72


◆ Upgrade scenarios where the storage is being replaced is notsupported.

◆ Configurations where the cluster node is zoned to both local andremote storage arrays are not supported.

◆ For upgrade scenarios, the cluster quorum type can only bechanged before or after the upgrade.

Note: For information on converting existing clusters to CE clusters, referto “Using the CE Configuration Wizard” on page 80.

Getting started with RecoverPoint/CEThe following configuration and installation instructions areprovided as a high-level overview to getting started withRecoverPoint/Cluster Enabler.

Note: RecoverPoint/CE only supports existing RecoverPoint consistencygroups configured as CRR with exactly two-copies.

Note: If you are using fabric based splitters (such as SANTap or Brocade) orstorage array based splitters (such as a CLARiiON splitter) in a site, ensurethat no host based splitters are installed on the nodes at that site.

Configuration and installation instructionsFollow these step to get started with RecoverPoint/Cluster Enabler:

1. Before installing EMC RecoverPoint/Cluster Enabler on all hostnodes, Microsoft host clusters must be operating at both sites.

2. RecoverPoint version 3.1 or later must be installed and replicatingall volumes to be included in the cluster group between the twosites.

3. Before specifying the cluster groups in Microsoft Cluster Services,enable image access at the remote copy of each of the consistencygroups that contain volumes to be included in cluster groupsusing the RecoverPoint Management Application. To bereplicated by RecoverPoint/Cluster Enabler, each consistencygroup must have exactly the same name as the correspndingcluster resource group and must contain exactly the same



volumes. Names of consistency groups and the correspondingcluster groups may contain only alphanumeric characters andunderscores.

If image access of the consistency groups is not enabled,Microsoft Cluster Services will not allow the volumes to be addedto the cluster resource groups. Ensure that the consistency grouphas finished synchronizing before continuing.

4. Define all cluster groups and resources using Microsoft ClusterServices.

Note: On Windows Server 2008, all disks must first be added to FailoverCluster Management before they can be configured for Cluster Enabler.“Setting up devices on Windows Server 2008” on page 30 providesinstructions.

5. Disable image access of the RecoverPoint consistency groups thatwere enabled in step 3.

6. In the RecoverPoint Management Application, select theconsistency group. In the Components pane, select the Policy tab.In the Stretch Cluster Support area, check Use RecoverPoint/CE.Ensure that Group is managed by CE, RecoverPoint can onlymonitor is selected. Repeat for each consistency group to bereplicated by RecoverPoint/CE.

7. Install RecoverPoint/CE on all host nodes in the cluster.“Installing the Cluster Enabler Base component” on page 110describes how to install the base component, either separately orwith the RecoverPoint plug-in. “Installing the RecoverPoint/CEplug-in module” on page 74 describes how to install theRecoverPoint plug-in.

8. In the RecoverPoint/CE Manager GUI, use the CE ConfigurationWizard to configure a CE cluster. The configuration wizard willautomatically validate credentials, discover the storage volumesand perform the required configuration.

Note: You will be prompted to set up the authorization list for the clusterthe first time that you run the CE Configuration Wizard.

Installation overview 73

74


Installing the RecoverPoint/CE plug-in moduleThere are two methods for installing the RecoverPoint/CE plug-in: asa standalone module, or together with the Base component.

This section describes how to install the plug-in as a standalonemodule. For information on installing the plug-in together with theBase component, refer to “Installing the Cluster Enabler Basecomponent” on page 110.

1. Review “Before you begin” on page 71.

2. Verify that the Base component is installed, as described in“Installing the Cluster Enabler Base component” on page 110.

3. Run the plug-in installation program from Powerlink andcomplete the steps in the installation wizard, being sure to selectthe same installation directory as the Base component.

4. When prompted to restart your system, click Yes to restart thesystem, or No to restart it at a later time.



Uninstalling the RecoverPoint/CE plug-in moduleThis section explains the methods for uninstalling theRecoverPoint/CE plug-in module from a configured cluster.

As an alternative method, you can uninstall the Base component,which will also uninstall the plug-in at the same time. Forinstructions, refer to “Uninstalling the Cluster Enabler Basecomponent” on page 112.

Uninstalling the plug-in from some cluster nodesTo remove some cluster nodes and leave the plug-in on the remainingcluster nodes:

1. Open Microsoft Failover Cluster Manager.

2. Ensure no cluster resource groups are owned by the nodes youwill remove. Move any owned resource groups to a differentnode.

3. Right-click the nodes to remove and choose Stop Cluster Service.Wait for the cluster service to stop on the nodes as indicated by ared X.

4. Right-click the nodes you want to remove and choose Evict.Evicting a node will uninstall the cluster service on that node andremove that node from the cluster.

Note: If CE Manager is already open, perform a refresh before runningthe Storage Discover Wizard.

5. After evicting nodes, open CE Manager and right-click the clustername. Choose Storage Discover and follow through theprocedure steps to complete the Storage Discover Wizard.

6. Use the Add/Remove Programs utility in the Control Panel toremove EMC Cluster Enabler RecoverPoint Plug-in from theevicted nodes .

Uninstalling the RecoverPoint/CE plug-in module 75

76


Uninstalling the plug-in from all cluster nodes/deconfiguring the clusterTo uninstall the plug-in from all nodes of the cluster and deconfigurethe CE cluster, while maintaining the Microsoft Failover Cluster:

1. Move all resource groups to the nodes on one site (i.e., Site A).

2. Right-click only the nodes on the remote site (i.e., Site B) andchoose Evict.


3. After evicting the nodes on the remote site, open CE Manager ona node at Site A and right-click the cluster name. Choose StorageDiscover and follow through the procedure steps to complete theStorage Discover Wizard.

4. From the CE Manager, select Deconfigure CE.

5. Use the Add or Remove Programs utility in the Control Panel toremove EMC Cluster Enabler RecoverPoint Plug-in from allnodes.

Uninstalling the plug-in from all cluster nodes/destroying the clusterTo uninstall the plug-in from all nodes of the cluster and destroy thecluster:

1. Deconfigure the cluster according to steps 1 through 4 in“Uninstalling the plug-in from all cluster nodes/deconfigure thecluster” on page 76.

2. Destroy the cluster using Microsoft Failover Cluster Manager.

3. Use the Add or Remove Programs utility in the Control Panel toremove EMC Cluster Enabler RecoverPoint Plug-in from allnodes.


6Invisible Body Tag

This chapter provides instructions for using theRecoverPoint/Cluster Enabler Manager graphical user interface.

◆ Getting started using the CE Manager............................................ 78◆ Using the CE Configuration Wizard............................................... 80◆ Managing a CE cluster ...................................................................... 84◆ Managing a CE cluster group........................................................... 89◆ Storage component ............................................................................ 94◆ Viewing information ......................................................................... 97◆ Restore and recovery operations ................................................... 106

Using Cluster EnablerManager

Using Cluster Enabler Manager 77

78

Using Cluster Enabler Manager

Getting started using the CE ManagerThe Cluster Enabler (CE) Manager graphic user interface (GUI)allows you to configure your Microsoft Failover Clusters (or MSCSclusters) for disaster recovery protection. The CE Manager allowsyou to set up and configure disk-based resources to automaticallymove geographically dispersed resource groups back and forth.

The CE Manager provides several wizards to assist you in completingvarious cluster tasks. The first step towards managing disasterrecovery for distributed failover clusters is to run the ConfigurationWizard to configure a CE cluster.

The Cluster Enabler Manager windowThe CE Manager window shown in Figure 14 contains a menu bar,two views, and a navigation tree. After cluster configuration, thenavigation tree can be expanded to show four separate components:Groups, Storage, Sites, and Nodes.

Figure 14 Cluster Enabler Manager window



Cluster Enabler wizardsWizards are a series of dialog boxes that step you through thecompletion of a complex task. The Cluster Enabler Manager providesseveral wizards, as follows:

Configuration Wizard The Configuration Wizard is used to configure a CE cluster. Theconfiguration process is the first step towards managing disasterrecovery for distributed failover clusters. The Configuration Wizardwill step you through the process of configuring your failover clusterfor management with CE. “Using the CE Configuration Wizard” onpage 80 provides detailed instructions for using the wizard.

Storage DiscoverWizard

The Storage Discover Wizard automatically discovers and sets up theattached storage. The storage discovery process should be performedafter any changes have been made to the storage configuration.“Storage Discover Wizard” on page 84 provides detailed instructionsfor using the wizard.

Update Mirror PairsWizard

The Update Mirror Pairs Wizard steps you through the process ofdiscovering storage, updating the storage configuration, validatingthe storage groups, and setting up the storage group definitions inthe cluster properties database to update the mirrored pairs in acluster. “Update Mirrored Pairs Wizard” on page 85 providesdetailed instructions for using the wizard.

Change QuorumWizard

The Change Quorum Wizard steps you through the process ofchanging a cluster's quorum model type. “Change Quorum ModelWizard” on page 86 provides detailed instructions for using thewizard.

Create Group Wizard The Create Group Wizard steps you through the process of creating aCE Group, adding devices and selecting a group policy. “CreateGroup Wizard” on page 89 provides detailed instructions for usingthe wizard.

Modify Group Wizard The Modify Group Wizard steps you through the process of addingor removing devices in a CE group. “Modify Group Wizard” onpage 91 provides detailed instructions for using the wizard.

Recover CE ClusterWizard

The Recover CE Cluster Wizard feature is for shared quorum clusterrecovery only and is not supported in RecoverPoint/Cluster Enablerversion 4.0.

Getting started using the CE Manager 79

80


Using the CE Configuration WizardCluster Enabler provides a wizard for configuring a CE cluster. Theconfiguration process is the first step towards managing disasterrecovery for distributed failover clusters. The Configuration Wizardwill step you through the process of configuring your failover clusterfor management with CE.

If any of the steps in wizard configuration process fail, the wizardwill display a list of the specific errors for each node on a Summarypage. Note each error to be corrected and click Finish to exit thewizard. After the listed summary problems have been fixed, launchthe configuration wizard again to configure the CE cluster.

Note: Whether running Windows Server 2003 or 2008, the applicableMicrosoft Failover Clusters (MSCS clusters) or Failover Clusters must beinstalled on at least one node prior to configuring a cluster.

Follow these steps to configure a RecoverPoint/CE cluster using theConfiguration Wizard:

1. All RecoverPoint consistency groups must be configured beforerunning the configuration wizard. The existing RecoverPointconsistency groups must have been previously enabled for imageaccess and configured as a stretch cluster in the RecoverPointManagement Application. For instructions on setting up theRecoverPoint consistency groups for Cluster Enablermanagement, refer to “Getting started with RecoverPoint/CE” onpage 72.

2. Select the EMC Cluster Enabler icon from the NavigationConsole Tree and click the Configure CE Cluster link in thecenter pane. This opens the first page of the ConfigurationWizard.

Note: The Configuration Wizard can also be launched using theright-click or Action menus.

3. The Enter cluster name page appears. Enter a Cluster Name orNode Name in the space provided and click Configure. If you donot enter a name and click Configure, the wizard willautomatically detect the current cluster on the server andcontinue.



4. The Current Nodes page appears listing the current nodes in thecluster. To add a node, enter the node name and click Add. If youdo not know the node name, you can click Browse to browse anactive directory of computers. Select a computer name from thelist and click OK, then click Add. Click Next.

5. The Validating System Setup process begins. This automated stepvalidates the system configuration by checking that theappropriate versions of Cluster Enabler and Microsoft FailoverClusters (MSCS) are installed and configured. Upon theValidation Complete notification, click Next.

Note: If the system validation process fails, the wizard will list thevalidation errors for each node on the Summary page. Note each error tobe corrected and click Finish to exit the wizard.

6. The Storage Discovery process begins. This automated stepperforms a storage discovery for each cluster node to identify thelocally-attached and remotely-attached storage. Upon theDiscover Completed notification, click Next.

If the storage discovery process fails because the authorization listhas not been set up for the cluster, the following error displays:

[node_name]: An error occurred when discovering storageon node [node_name]. The Authorization list is notcompletely set up.

Click Next to set up the authorization list. On the GetAuthorization List page, click Next again.

On the Set Authorization List page, supply the authorizationcredentials. Select the check box next to a cluster node and enterthe following credentials:

Table 4 Authorization list credentials

Click Update after entering the credentials for a node.

Parameter Value and description

RecoverPointManagement IP

Floating management IP address of the RPA with which this host communicates.

User plugin

Password default = plugin

Using the CE Configuration Wizard 81

82


Repeat this process for each listed node. When done, click Next.

The Configuration Wizard sets up the authorization list. Upon“Setting Authorization List Completed,” click Next.

Note: EMC recommends that you change the default password for theplugin user. To change the password, log in to the RecoverPoint CLI withUser = plugin and Password = plugin. Run the set_password command.You can then run the CE Configuration Wizard again and enter the newpassword on the Set Authorization List page. Alternatively, you can usethe RecoverPoint Access Settings facility to update the password(without having to re-run the configuration wizard). Navigate to Start >Programs > EMC > Cluster Enabler > RecoverPoint Access Settings andenter the new password.

Note: If the storage discovery process fails for any other reason, thewizard will list the storage discovery errors for each node on theSummary page. Note each error to be corrected and click Finish to exitthe wizard.

7. The Storage Setup process begins. This automated step performsa storage setup for each cluster node. Upon Setup of StorageConfiguration Completed, click Next.

Note: If the storage setup process fails, the wizard will list the storagesetup errors for each node on the Summary page. Note each error to becorrected and click Finish to exit the wizard.

8. The Validating Groups process begins. This automated stepperforms a group validation for each converted failover clustergroup. Upon Validated Groups, click Next.

Note: If the validating groups process fails, the wizard will list thevalidation errors for each node on the Summary page. Note each error tobe corrected and click Finish to exit the wizard.

9. The Summary page appears. Upon Configuration WizardCompleted Successfully, click Finish to exit the wizard.

10. After exiting the CE Configuration Wizard, Cluster Enabler willconnect to the newly configured cluster. Once connected to thecluster, you will notice that the configured cluster node is nowvisible in the navigation tree, located in the left pane.



11. Double-click the cluster icon to expand the cluster and view thefollowing folders: Groups, Storage, Sites, and Nodes. You are nowready to begin managing your cluster. Figure 15 on page 83shows an example view of the expanded CE Manager navigationtree.

Figure 15 CE Manager expanded navigation tree

Adding nodesAdding new nodes is also accomplished through using the CEConfiguration Wizard. The CE Configuration Wizard steps youthrough the process of adding a cluster node for management withCE. New nodes must be added using CE Manager and not MicrosoftFailover Cluster. Step 4 in the “Using the CE Configuration Wizard”on page 80 provides instructions.

Using the CE Configuration Wizard 83

84


Managing a CE clusterOnce your CE cluster has been configured using the CEConfiguration Wizard, you are ready to begin managing the cluster.Even after you exit Cluster Enabler and close the application, yourcluster will remain configured unless you perform a deconfigure ordelete action on the cluster.

Follow these instructions to begin managing your cluster.

1. Select the EMC Cluster Enabler icon from the NavigationConsole Tree and click the Manage CE Cluster link in the centerpane. The Manage CE Cluster option can also be launched byusing the right-click or Action menus.

2. Enter the cluster name in the selection box. If you do not enter aname, the default will automatically connect to a clusteraccessible on the server. Click OK.

3. Once connected to the cluster, you will notice that the configuredcluster node is now visible in the navigation tree located in theleft pane. Double-click the cluster icon to expand the cluster andview the following folders: Groups, Storage, Sites and Nodes. Youare now ready to begin managing your cluster.

Cluster Enabler Manager allows you to perform the following clustermanagement actions on configured clusters. When a particular actionis accessed, the appropriate wizard process is launched to assist youin completing the task.

Storage Discover WizardThe Storage Discover Wizard helps you to automatically discoverand setup the attached storage. The storage discovery process shouldbe performed after any changes have been made to the storageconfiguration.

Follow these steps to automatically discover and set up your attachedstorage using the Storage Discover Wizard:

1. Select the Cluster icon from the Navigation Console Tree andselect Action and Storage Discover from the menu bar. Thisopens the first page of the Storage Discover Wizard. The StorageDiscover Wizard can also be launched by using the right-click orAction menus.



2. The Storage Discovery page appears. Upon the DiscoverCompleted notification, click Next.

Note: If any storage discovery process fails, the wizard will list thediscovery errors for each node on the Summary page. Note each error tobe corrected and click Finish to exit the wizard.

3. The Storage Setup page appears. Upon the Set up of StorageConfiguration Completed notification, click Next.

4. The Summary page appears. Upon the Discovered all Nodesnotification, click Finish to exit the wizard.

5. Cluster Enabler will then refresh the CE cluster to reflect anystorage changes.

Update Mirrored Pairs WizardThe Update Mirrored Pairs Wizard helps you update the mirroredpairs in a cluster. This wizard steps you through the variousprocesses of discovering storage, updating the storage configuration,validating the storage groups, and setting up the storage groupdefinitions in the cluster properties database to update the mirroredpairs in a cluster.

Follow these steps to update the mirrored pairs in a cluster using theUpdate Mirror Pairs Wizard:

1. Select the Cluster icon in the navigation tree and select Action,More Actions... and Update Mirror Pairs from the menu bar. TheUpdate Mirror Pairs Wizard can also be launched using theright-click or Action menus.

2. The first page of the Update Mirror Pairs Wizard opens andbegins the Storage discovery process. Upon the DiscoverComplete notification, click Next.

3. The Storage setup process begins setting up the storageconfiguration. Upon the Setup of Storage ConfigurationCompleted notification, click Next.

4. The Validating Groups process begins validating each group inthe cluster. Upon the Validated Groups notification, click Next.

Managing a CE cluster 85

86


5. The Updating Storage Mirror Relationships process beginsupdating the mirrored pairs in the groups. Upon the Updatemirror pairs for groups notification, click Next.

6. The Summary page appears. Upon Update Mirror PairsCompleted Successfully, click Finish to exit the wizard.

Change Quorum Model WizardThe Change Quorum Wizard changes the quorum model of a cluster.This wizard will step you through the various processes of changinga cluster's quorum model type.

◆ For Windows Server 2003, RecoverPoint/CE allows you tochange the cluster model type between Majority Node Set (MNS)and MNS with File Share Witness.

◆ For Windows Server 2008, RecoverPoint/CE allows you tochange the cluster model type between Node Majority and Nodeand File Share Majority.

A descriptive list of all quorum model types is provided in“Supported model type descriptions” on page 28.

Using the Change Quorum WizardOnce your Microsoft cluster has been configured as a CE cluster, youmust use this wizard for all quorum model changes.RecoverPoint/CE version 4.0 supports only Windows Server 2003Majority Node Set (MNS) and MNS with File Share Witness orWindows Server 2008 Node Majority and Node and File ShareMajority.

Note: To change the quorum model to "Node and File Share Majority" inWindows Server 2008, you must first update the FileShare permissions to addthe Cluster Name and allow "Change" and "Read" permissions for the fileshare. Your Windows documentation provides instructions on changingpermissions for FileShare.

Figure 16 on page 87 shows the first page of the Change QuorumWizard for Windows Server 2003.



Figure 16 Windows Server 2003 supported quorum models

Change Quorum Model exampleThe following example steps through the processes of changing aNode and File Share Majority cluster to a Node Majority cluster forWindows Server 2008.

Follow these steps to change the quorum model type of a clusterusing the Change Quorum Wizard:

1. Select the Cluster icon in the navigation Console tree and selectAction, More Actions... and Change Quorum Model from themenu bar. The Change Quorum Model Wizard can also belaunched by using the right-click or Action menu. Figure 17shows how to access the Change Quorum Wizard. Note that thecurrent Cluster Type is Node and File Share Majority.

Figure 17 Change quorum model

Managing a CE cluster 87

88


2. Cluster Enabler automatically begins by reading the storageconfiguration.

3. Once the storage configuration has been read, the first page of theChange Quorum Wizard opens. Select the New Cluster Modelradio button for the model you want to convert the cluster to. Inthis example, we are changing the cluster model from a Node andFile Share Majority cluster to a Node Majority cluster. Click Next.

4. The Validate Cluster Model process automatically beginsvalidating the chosen cluster model. Upon the Validation ofCluster Model Successfully notification, click Next.

5. The Change Cluster Model process automatically beginschanging the cluster settings to the new model. Upon ChangeCluster Model Successfully, click Next.

6. The Summary page appears. Upon the Changed Cluster ModelSuccessfully notification, click Finish to exit the wizard.

7. In the Cluster Enabler Manager, select the Cluster icon, noticethat the Cluster Type is now Node Majority. Figure 18 shows theCluster Type has been changed to Node Majority.

Figure 18 Cluster Type changed to Node Majority



Managing a CE cluster groupCluster Enabler Manager provides several group actions formanaging CE cluster groups. There are two automation wizardsavailable for groups, the Create Group Wizard and the Modify GroupWizard. The Create Group Wizard steps you through the process ofcreating a CE cluster group. The Modify Group Wizard allows you toedit an existing cluster group by adding or removing devices to andfrom the group. The group action features also allow you todeconfigure a CE group to convert it back to a cluster group or todelete a group.

Create Group WizardThe Create Group Wizard will step you through the process ofconverting your existing RecoverPoint consistency group to a CEGroup and selecting a group policy.

Follow these steps to create a CE Group using the Create GroupWizard:

1. Select the Groups icon from the Navigation Console Tree andselect Action and Create Group from the menu bar. This beginsthe process of reading the storage configuration. After the storagehas been read, the first page of the Create Group Wizard opens.The Create Group Wizard can also be launched using theright-click or Action menus.

Note: A mirror needs to be present on the array before attempting tocreate a group. Run the Storage Discover Wizard to detect a newlycreated mirror.

2. The Enter Group Name page appears. Enter the name of anexisting RecoverPoint consistency group in the space providedand click Create. You must enter the exact name of an existingRecoverPoint consistency group to continue. To abort theoperation and close the wizard, click Cancel.

Note: The existing RecoverPoint consistency group must have beenpreviously enabled for image access and configured as a stretch cluster inthe RecoverPoint Management Application. For instructions on settingup the RecoverPoint consistency group for Cluster Enabler management,refer to “Getting started with RecoverPoint/CE” on page 53.

Managing a CE cluster group 89

90


3. The Select Group Policy page appears containing the existingRecoverPoint consistency group device information. Currentlyfor RecoverPoint/CE, you can only select Restrict GroupMovement. Once selected, click Next.

The Restrict Group Movement selection restricts the group fromfailing over to a peer node. In a link failure, this setting will onlyattempt to move disk laterally. If the link is up, this setting has noimpact..

Figure 19 Create Group Wizard, Select Group Policy

4. The Summary page appears. Upon Group Created Successfully,click Finish to exit the wizard.

Cluster Enabler automatically begins refreshing the CE Cluster. Uponcompletion of the refresh, you should see the group that you createdlisted under Groups. If you do not see the newly created group, selectAction and Refresh from the menu bar. The Refresh action can alsobe accessed from the right-click or Action menus.



Modify Group WizardThe Modify Group Wizard will step you through the process ofadding or removing devices in a CE group. Changes can only bemade to a CE group to reflect the existing RecoverPointconfiguration.

Before using the Modify Group Wizard to add or remove devices,you must add or remove the devices in the RecoverPoint consistencygroup. If you have previously removed devices from theRecoverPoint consistency group using the RecoverPointManagement application and refreshed the storage/group views,you will not be able to remove the selected devices using the ModifyGroup wizard and must use Microsoft Cluster Administrator todelete the devices.

Follow these steps to add or remove devices from a CE Group usingthe Modify Group Wizard:

1. Select the Groups icon in the navigation tree and select Actionand Modify Group from the menu bar. This begins the StorageSynchronization process. After the storage has finishedsynchronizing, the first page of the Modify Group Wizard opens.The Modify Group Wizard can also be launched using theright-click or Action menus.

2. From the Select Devices page, select the Action from thepull-down menu for either Add Devices or Delete Devices.Depending on your selection, a list of available devices that canbe added or removed will display. Select the desired devices fromthe list shown by clicking in the select boxes. Selected devices areidentified by the checked box. Click Next. Figure 20 on page 92shows this wizard page.


92


Figure 20 Modify Group Wizard, Select Devices

3. The Validate Selection page appears, click Next to validate yourselection or click Cancel to abort the action.

4. The Summary page appears. Upon Group Modified Successfully,click Finish to exit the wizard.

Cluster Enabler automatically begins refreshing the CE cluster. Uponcompletion of the refresh, you should see the updated groupinformation reflecting the devices added or deleted. If you do not seethe updated group information, select Action and Refresh from themenu bar. The Refresh action can also be accessed from the right-clickor Action menus.



Deconfigure a CE groupTo deconfigure a CE group, select a group listed under the Groupsicon located in the Navigation Console Tree and select Action andDeconfigure CE Group from the menu bar. The deconfigure optioncan also be accessed by using the Right-click or Action menus. Adialog box pop-up will appear asking you to confirm the action. ClickYes to convert the group or No to abort the action.

CAUTION!If the group is deconfigured, CE failover support to the remotenodes will no longer be operational. To make group failoveroperational again, you will need to reconfigure the cluster groupusing the CE Configuration Wizard in the CE Manager.

Delete a CE groupTo delete a CE group, select a group listed under the Groups iconlocated in the Navigation Console Tree and select Action and DeleteGroup from the menu bar. The delete group option can also beaccessed using the Right-click or Action menus. A dialog box pop-upwill appear asking you to confirm the action. Click Yes to delete thegroup or No to abort the action.

Note: Deleting a CE group deconfigures the group and then removes it fromthe cluster.


94


Storage componentSelecting the Storage icon from the navigation tree allows you viewthe attached storage device information for each storage array. Selecta storage array to view the summary information columns in thecenter pane. Figure 21 shows the CE Manager Storage componentview for a RecoverPoint supported storage array.

Figure 21 Example of RecoverPoint/CE storage array view

Table 5 on page 94 lists the center pane heading information that isdisplayed for RecoverPoint supported storage arrays.

Table 5 Storage component displayed information

Column heading Description

Name The RecoverPoint array name.

Version Displays the RecoverPoint version.

Device ID Shows all RecoverPoint device IDs that are mapped to any cluster member node.

Cluster Group Name Indicates the CE Group name to which the device belongs.

Consistency Group Name Indicates the consistency group name to which the remote mirror belongs; derivedfrom Cluster Group name.



Owning Node If a device belongs to a cluster group, the owning node information is obtained directlyfrom Microsoft Failover Cluster (or MSCS). Otherwise, the owning node is a nodewhere the device is write-enabled.

Mount Points Indicates the mount point of the physical drive on the owning node.

Sync State Indicates the consistency group state.

Device Status Indicates the replication set device status. The possible device status states areReady, Not Ready, and No access. No access means that the host cannot access thestorage.

Capacity MB Shows the device capacity in megabytes.

Swap Capable Shows whether the device is swap capable (True or False).

Async Capable Shows whether the device is asynchronous capable (True or False).

Volume Name The storage volume name.

Vendor The storage vendor name.

Product Product information returned through the RecoverPoint SCSI reply.

Cluster Mode The RecoverPoint consistency group cluster mode setting.

Copy The RecoverPoint copy inside the storage group.

Replication Set Indicates the RecoverPoint replication set name that contains the volume and itscorresponding replicated volume.

Table 5 Storage component displayed information (continued)


Storage component 95

96


Adding and removing devices from a groupFrom the storage view, when you select a device listed in the centerpane, you can add or remove it from the group by clicking the Add toGroup or Remove from Group icons displayed in the right Actionpane. Figure 22 displays the storage actions located in the right actionpane.

As with the Modify Group wizard, changes can only be made to a CEgroup to reflect the existing RecoverPoint configuration. Beforeadding or removing devices, you must add or remove the devices inthe RecoverPoint consistency group. If you have previously removeddevices from the RecoverPoint consistency group using theRecoverPoint Management application and refreshed thestorage/group views, you will not be able to remove the selecteddevices using the Modify Group wizard and must use MicrosoftCluster Administrator to delete the devices.

Figure 22 CE Manager storage actions

When you select the action, Cluster Enabler opens the “ModifyGroup Wizard” at the validation step. Click Next to add or removeyour selection. The Summary dialog box appears. Upon GroupModified Successfully, click Finish to exit the wizard.



Viewing informationCluster Enabler allows you to view certain summary informationabout CE Groups, Storage, Sites and Nodes. This information isdisplayed in the center pane as you select each of the icons located inthe navigation tree. “Storage component” on page 94 describes thestorage information and available actions. Summary information forGroups, Nodes, and Sites are described below.

Displaying group informationSelecting the Groups icon from the navigation tree allows you viewgroup information for all configured CE cluster groups in the centerpane. Double-clicking on a specific group displays summaryinformation for that group. Figure 23 displays the CE Manager groupcomponent.

Figure 23 CE Manager Groups component

Selecting a specific group icon from the navigation tree allows youview the group summary information columns for each configuredCE cluster group in the center pane. Figure 24 on page 98 displays theCE Manager group information for a specific group.

Viewing information 97

98


Figure 24 CE Manager groups information

Table 6 lists the summary and column heading information thatdisplays for CE Groups.

Table 6 Groups component displayed information

Summary and columnheading information Description


Device Group Name Indicates the consistency group name to which the remote mirror belongs;derived from Cluster Group name.

Owning Node Shows the failover cluster node name that owns the particular group. Thisinformation is obtained directly from MS Failover Cluster (or MSCS). Only groupsthat are part of the cluster will be displayed.


Devices Listed by disk resource name in the cluster.

Cluster Resource Name The RecoverPoint consistency group cluster mode setting.

Owner Device ID Listed by physical disk resource name.

Owner Storage ID Indicates which node owns the cluster.

Mount Points The device ID mapped to the owning node.



Clicking the Display Events icon option in the action pane displaysevent information in the lower tier of the center pane. Table 7 lists theheading information that displays for CE Groups events.

Table 7 Groups event information


Date/Time Shows the date and time that the recorded event occurred.

Computer Name Indicates the computer name on which the event occurred.

Group Name Indicates the group name to which the event occurred.

Message Displays a detailed message of the event type.


100


Displaying node informationSelecting the Nodes icon from the navigation tree allows you viewnode information for all nodes in the center pane. Double-clicking ona specific node displays summary information for that node.Figure 25 displays the CE Manager nodes component.

Figure 25 CE Manager Nodes component



Selecting a specific node icon from the navigation tree allows youview the node summary information columns for each node in thecenter pane. Figure 26 displays the CE Manager node information fora specific node.

Figure 26 CE Manager node information

Table 8 lists the heading information that displays for Nodes.

Table 8 Nodes component displayed information (page 1 of 2)


Name Displays the node name.

OS Name Displays the Windows operating system (such as, 2003 SP2).

OS Version Displays the Windows operating system version (such as, 5.2.3790).

System Type Displays the Windows system type (such as, X86).


Consistency Group Name Indicates the consistency group name to which the remote mirror belongs;derived from Cluster Group name.


102


Clicking the Display Events icon option in the action pane displaysevent information in the lower tier of the center pane. Table 9 lists theheading information that displays for CE Node events.

Owning Node Shows the failover cluster node name that owns the particular group. Thisinformation is obtained directly from MS Failover Cluster (MSCS). Only groupsthat are part of the cluster will display.


Devices Listed by cluster resource name.

Table 8 Nodes component displayed information (page 2 of 2)


Table 9 Node event information


Date/Time Shows the date and time that the recorded event occurred.

Computer Name Indicates the computer name on which the event occurred.

Group Name Indicates the group name to in which the event occurred.

Message Displays a detailed message of the event type.



Displaying site informationSelecting the Sites icon from the navigation tree allows you view siteinformation for all sites in the center pane. Double clicking on aspecific site displays summary information for that site. Figure 27 onpage 103 displays the CE Manager Site component.

Figure 27 CE Manager Sites component

Selecting a specific site icon from the navigation tree allows you viewthe site summary information columns for each site in the centerpane. Figure 28 on page 104 displays the CE Manager siteinformation for a Symmetrix site.

You can change the name of a site by using the right-click menu andselecting Rename. The site name will then become editable. The sitename can also be changed by using the right-click menu and selectingProperties. You may then change the site name in the Properties tab.


104


Figure 28 CE Manager RecoverPoint site information

Table 10 lists the summary and column heading information thatdisplays for Symmetrix sites.

Table 10 Site component displayed information (page 1 of 2)


Name Displays the Site name.

Number of Storage Units Displays the number of storage units for this site.

Number of Nodes Displays the number of nodes for this site.

Storage Unit Name & ID The RecoverPoint supported storage array ID.

Version Information Displays the RecoverPoint version.

Node Name Displays the node name.

OS Name Displays the Windows operating system (such as, 2003 SP2).

OS Version Displays the Windows operating system version (such as, 5.2.3790).

System Type Displays the Windows system type (such as, X86).




Device Group Name Indicates the consistency group name to which the remote mirror belongs;derived from Cluster Group name.

Owning Node Shows the failover cluster node name that owns the particular group. Thisinformation is obtained directly from MS Failover Cluster (MSCS). Only groupsthat are part of MS Failover Cluster (MSCS) will display.


Devices Listed by cluster resource name.

Table 10 Site component displayed information (page 2 of 2)



106


Restore and recovery operationsThis section details some of the restore and recovery operations thatshould be performed for different types of failures.

Restrict group movement and recoveryIf a CE group has the "Restrict Group Movement" policy set and thestorage link is down, it may take a long time for the resource to comeonline if the user manually tries to move the group to a node that isconnected to a different storage array. For example, if the user tries tomove group G1 from the source side to the remote side when thestorage link is down, then Microsoft’s preferred owner logic willattempt to bring the group online on the remote side as expected.

But since the restrict group movement policy is set for the CE group,Microsoft will fail the resource on the remote side nodes. This iscorrect behavior and is expected, but it may take a long time for theresource to fail on all the remote nodes before coming back online onone of the source side nodes. This is because by default Microsoft willtry to bring the group online 3 times on each node. The more nodesyou have in the cluster the longer it will take for Microsoft tocomplete the preferred owner logic. To minimize this undesirableeffect you can change the property of the resources to "Do notRestart". This will minimize number of retries and reduce the timerequired to bring the group online.

Recover CE Cluster WizardThe Recover CE Cluster Wizard feature is for shared quorum clusterrecovery only and is not currently supported in version 4.0 of theRecoverPoint/Cluster Enabler product.

Server or storage failure with failure of the RecoverPoint clusterIf an entire RecoverPoint cluster fails and either hosts or storage failsat that site, the resources cannot fail over automatically to the othersite. “RecoverPoint manual failover” on page 62 providesinstructions on failing them over manually.


Appendix AInvisible Body Tag

This chapter explains how to install and uninstall the Cluster EnablerBase component and provides configuration instructions forWindows Server 2003 and 2008 (including R2).

◆ Installation overview....................................................................... 108◆ Before you begin .............................................................................. 109◆ Installing the Cluster Enabler Base component........................... 110◆ Uninstalling the Cluster Enabler Base component ..................... 112◆ Configuring CE cluster on R2 Server Core................................... 114◆ Upgrading Windows Server 2008 to Windows Server 2008 R2 116

Base ComponentInstallation and

Upgrade

Base Component Installation and Upgrade 107

108

Base Component Installation and Upgrade

Installation overviewThe Base component InstallShield wizard allows you to install theBase component by itself or with one or more plug-ins.

Note: Starting with Cluster Enabler V4.0, the Cluster Enabler Basecomponent is a prerequisite for the Cluster Enabler plug-ins, and thereforemust be installed together with or prior to the plug-ins.

It is recommended that you contact EMC Customer Support forassistance if any of the following issues apply:

◆ You have applications already layered with dependencies.

◆ You need other devices online.

◆ You are not confident about installing and configuring newsoftware within the context of Windows Server 2003 or 2008, orMicrosoft Failover Clusters.

IMPORTANT!Upgrading to Cluster Enabler version 4.0 is supported for versions3.x and later. Refer to “Before you begin” on page 109 for additionalrequirements before upgrading.



Before you beginBefore you begin to install the Cluster Enabler Base component, youshould read the following requirements and considerations:

◆ The following Windows processor architectures are supported:


Note: Microsoft does not support mixed architecture clusters. All nodesmust be the same Windows architecture.

◆ Installation on Windows Server 2003 systems requires that aminimum of SP2 (Service Pack) first be installed.

◆ Installation on Windows Server 2003 requires that Microsoft .NETFramework Version 2.0 first be installed.

◆ Installation on Windows Server 2003 requires using the ConsoleMode in Remote Desktop to connect to a host for installationchaining to work properly. This is a Microsoft limitation. To useRemote desktop in console Mode, follow these steps:

1. Click start and select Run.

2. Enter:

mstsc/console/v:<host name>

3. Click OK.

◆ Installation on Windows Server 2003 and 2008 requires thatMicrosoft Windows Installer version 4.5 first be installed.

◆ Installation on Windows Server 2008 requires that all nodes firstbe installed with the Failover Cluster feature.

◆ For Failover Cluster on Windows Server 2008, Microsoft ClusterValidate must pass all tests except storage.

◆ Upgrade scenarios where the storage is being replaced is notsupported.

◆ Configurations where the cluster node is zoned to both local andremote storage arrays are not supported.

◆ Installation on Windows Server 2008 R2 Core requires additionalconfiguration steps. “Configuring CE cluster on R2 Server Core”on page 114 provides configuration instructions.

109

110


Installing the Cluster Enabler Base componentThis section explains the methods for installing the Cluster EnablerBase component:

◆ Installing the Base component separate from the plug-ins (cleaninstall).

◆ Installing the Base component together with the plug-ins (cleaninstall).

◆ Installing the Base component while upgrading.

Note: Before starting either of the procedures in this section, be sure to review“Before you begin” on page 109 of this guide. In addition, if you are installinga plug-in together with the Base component, be sure to review “Installationoverview” on page 70.

Installing the Base component separate from the plug-ins (clean install)To install the Base component separate from the plug-ins:

1. Run the Base component installation program either fromPowerlink or the installation CD.

If running the installation from the CD, navigate to youroperating system's directory (either x64, x86 or IA64) and run theEMC_CE_Base.msi file to launch the InstallShield wizard.

2. Complete the steps in the InstallShield wizard.

3. When prompted to restart your system, click Yes.

You have now finished installing the Base component.

Installing the Base component together with the plug-ins (clean install)To install the Base component together with the plug-ins:

1. Create a temporary directory and into it either copy the contentsthe Base component installation CD, or download (save) the Basecomponent from Powerlink.

2. Download the plug-in(s) from Powerlink to the temporarydirectory you just created, being sure not to rename it.



3. In the temporary directory, navigate to your operating system'sdirectory (either x64, x86, or IA64) and run the EMC_CE_Base.msifile to launch the InstallShield wizard.

4. Complete the steps in the InstallShield wizard.

5. When prompted to restart your system, click Yes to restart thesystem, or No to restart it at a later time.

Installing the Base component together with plug-ins while upgradingTo install the Base component while upgrading from V3.x or later toV4.0:

1. Move all cluster groups to node A.

2. Do the following on all other cluster nodes:

a. Create a temporary directory and into it either copy thecontents the Base component installation CD, or download(save) the Base component from Powerlink.

b. Download the plug-in(s) from Powerlink to the temporarydirectory you just created, being sure not to rename it.

c. In the temporary directory, navigate to your operatingsystem's directory (either x64 , x86 or IA64) and run theEMC_CE_Base.msi file to launch the InstallShield wizard.

d. Complete the steps in the InstallShield wizard, being sure toselect the Upgrade path.

e. When prompted to restart your system, click Yes to restart thesystem.

f. After the node has finished rebooting, log onto the node tocomplete the upgrade.

Note: After reboot, the location from which the EMC_CE_Base.msifile is launched should be accessible without any login prompt.

g. After the InstallShield wizard completes the upgrade, clickYes to restart the system as prompted.

3. After all other nodes are up, move all groups from node A to oneof the other nodes. Ensure that the quorum group comes onlineon the other node before proceeding.

4. Repeat step 2 on node A.

111

112


Uninstalling the Cluster Enabler Base componentThis section explains the methods for uninstalling the Cluster EnablerBase component from a configured cluster:

◆ Uninstalling the Base component from some cluster nodes

◆ Uninstalling the Base component from all clusternodes/deconfigure the cluster

◆ Uninstalling the Base component from all cluster nodes/destroythe cluster

IMPORTANT!Uninstalling the Base component will also uninstall the ClusterEnabler plug-ins.

Uninstalling the Base component from some cluster nodesTo remove some cluster nodes and leave Cluster Enabler on theremaining cluster nodes:

1. Open Microsoft Cluster Administrator.

2. Ensure no cluster resource groups are owned by the nodes youwill remove. Move any owned resource groups to a differentnode.

3. Right-click the nodes to remove and choose Stop Cluster Service.Wait for the cluster service to stop on the nodes as indicated by ared X.

4. Right-click the nodes you want to remove and choose Evict.Evicting a node will uninstall the cluster service on that node andremove that node from the cluster.

5. After evicting nodes, open CE Manager and right-click the clustername. Choose Storage Discover and follow through theprocedure steps to complete the Storage Discover Wizard.




6. Uninstall CE from the evicted nodes. Use the Add/RemovePrograms utility in the Control Panel to remove EMC ClusterEnabler Base Component. Reboot when prompted to completethe uninstall.

Uninstalling the base component from all cluster nodes/deconfigure the clusterTo uninstall Cluster Enabler from all nodes of the cluster anddeconfigure the CE cluster. The Windows Server failover cluster willbe maintained:

1. Move all resource groups to the nodes on one site (i.e., Site A).

2. Right-click only the nodes on the remote site (i.e., Site B) andchoose Evict.

3. After evicting the nodes on the remote site, open CE Manager ona node at Site A and right-click the cluster name. Choose StorageDiscover and follow through the procedure steps to complete theStorage Discover Wizard.



5. Uninstall CE from all nodes. Use the Add or Remove Programsutility in the Control Panel to remove EMC Cluster Enabler BaseComponent. Reboot when prompted to complete the uninstall.

Uninstalling the base component from all cluster nodes/destroy the clusterTo uninstall Cluster Enabler from all nodes of the cluster and destroythe cluster:

1. Deconfigure the cluster according to steps 1 through 4 in“Uninstalling the base component from all clusternodes/deconfigure the cluster” on page 113.

1. Destroy the cluster using Microsoft Cluster Administrator.

2. Uninstall CE from all nodes. Use the Add or Remove Programsutility in the Control Panel to remove EMC Cluster Enabler BaseComponent. Reboot when prompted to complete the uninstall.

113

114


Configuring CE cluster on R2 Server CoreThe following instructions are provided as guidelines for configuringand managing a CE cluster for a Windows Server 2008 R2 Coreinstallation.

Requirements and considerationsBefore you begin to install the Cluster Enabler Base component on aWindows Server 2008 R2 Core edition, you should read the followingrequirements and considerations:

◆ Managing an R2 Server Core cluster can be done from a WindowsServer 2008 or Windows Server 2008 R2 host.

◆ The remote host where you are managing the R2 Server Corecluster must be on the same domain as the R2 Server Core hosts.

◆ The following website provide useful information for Server Coreinstallations:

http://technet.microsoft.com/en-us/library/cc753802.aspx

R2 Server Core configurationTo configure and install the Base component on a Windows Server2008 R2 Core edition:

1. Create a failover cluster on an R2 Server Core host.

To enable Failover Clustering, type the following from acommand line on a R2 Server Core host:

Ocsetup FailoverCluster-Core

To enable Hyper-V, type the following:

Ocsetup Microsoft-Hyper-V

Create a failover cluster using cluster.exe. You can create acluster from Failover Cluster Manager on the remote host.

Open Failover Cluster Manager on a remote host. If you haven’tcreated a cluster on the R2 Server Core hosts, create it usingFailover Cluster Manager. Once the cluster is created, connect tothis cluster.



2. Install .Net Framework, Solutions Enabler and CE on R2 ServerCore hosts.

To install .Net Framework 2.0, type the following from acommand line on a R2 Server Core host:

Start /w ocsetup NetFx2-ServerCore

Follow the installation instructions for “Installing the Basecomponent together with the plug-ins (clean install)” on page 110.

3. Manage an R2 Server Core cluster from a remote host.

• Install EMC CE Base Component on the remote host andreboot after prompted.

• Run CE Configuration Wizard on the remote host to convertthe R2 Server Core cluster to a CE cluster.

115

116


Upgrading Windows Server 2008 to Windows Server 2008 R2Upgrading Windows Server 2008 SP1 or SP2 to Windows Server 2008R2 while Cluster Enabler is installed on the host is not supported.

IMPORTANT!Attempting to upgrade to Windows Server 2008 R2 with ClusterEnabler version 4.0 installed causes undesirable results.

Follow these steps to prepare your host before upgrading fromWindows Server 2008 SP1 or SP2 to Windows Server 2008 R2:


2. Uninstall the CE Base Component from all nodes. Use the Add orRemove Programs utility in the Control Panel to remove EMCCluster Enabler Base Component. Reboot when prompted tocomplete the uninstall.

3. Follow the Windows Server 2008 R2 upgrade instructions forupgrading your operating system. Refer to the Microsoft Technetarticle titled, Understanding the Process of Migrating to a ClusterRunning Windows Server 2008 R2, available at:

http://technet.microsoft.com/en-us/library/cc731812.aspx

4. Install the Cluster Enabler Base Component and any plug-ins onall nodes.

5. From CE Manager, use the Configuration Wizard to configure theCE cluster.


Glossary

This glossary contains terms related to the Cluster Enabler software.

A

active state The state in which a MirrorView remote mirror is running normally.See also, “remote mirror states”.

agent An installed program designed to control a particular resource type.Each type of resource supported in a cluster is associated with anagent.

Application ProgramInterface (API)

A language and message format used by an application program tocommunicate with another program that provides services for it.APIs are usually implemented by writing function calls. Examples ofAPIs are the calls made by an application program to such programsas an operating system, messaging system, or database managementsystem.

See also, “SYMAPI” and “CLARAPI”.

asynchronous mode See “SRDF Asynchronous (SRDF/A)”.

attention state The MirrorView mirror's secondary image is fractured, and themirror is configured to generate an alert in this case. The mirrorcontinues to accept server I/O in this state.


118

Glossary

Auto recovery A MirrorView option to have synchronization start as soon as asystem-fractured secondary image is determined to be reachable.

availability The ability to continue to provide a service even during hardware orsoftware failure.

BBCV device A Symmetrix business continuance volume (BCV) that functions as a

mirrored media to a standard device for a protected storageenvironment.

BCV mirror A Symmetrix BCV device upon establishing or reestablishing a BCVpair.

BCV pair A standard Symmetrix device and a BCV device that provide aprotected storage environment.

business continuance An SRDF function that ensures business applications continuerunning despite possible disk failures.

Ccache Random access electronic storage used to retain frequently used data

between the CPU and either a hard disk or slower RAM. It speeds upgeneral data flow because a cache can be accessed quickly.

channel director The component in the Symmetrix array that interfaces between thehost channels and data storage. It transfers data between the channeland cache.

CLARAPI CLARiiON Application Program Interface. See “ApplicationProgram Interface (API)”.

client A computer using services or resources provided by a remotemachine, called a server. Often, communications software will have aseparate version for the client or guest, and the server or host.

Clients create a TCP/IP session with a service in the cluster using aknown IP address. This address appears to the cluster software as aresource in the same group as the application providing the service.In a failure, the Cluster Service will move the entire group to anothersystem.


Glossary

client failover The response of a client machine after resource failure on the serverfor the client caused a resource failover. A client will detect a failurein the session and reconnect in exactly the same manner as theoriginal connection. The IP address is now available on anothermachine and the connection is quickly reestablished. In this simplecase, all information related to the original session not committed todisk is lost. This provides higher availability, but no fault tolerancefor the service. Applications can use transactions to guarantee theclient request is committed to the server database to gainfault-tolerant semantics.

cluster A group of two or more independent computers addressed and usedas a single system.

cluster aware software Software that provides a restart mechanism invoked whenever theapplication resource is moved to another node in the cluster.

cluster Service The collection of software on each node that manages allcluster-specific activity.

consistency group A set of MirrorView logical units that are mirrored in a way thatallows a recoverable copy in the event of a disaster.

Also, for RecoverPoint, a consistency group is a data set consisting ofthe production source and its replicas. A consistency group comprisesthe production source volumes and either a local replica, remotereplica, or both. Each consistency group contains as many replicationsets as there are volumes in the production storage to replicate.

consistency groupcondition

Displays more detailed information about the MirrorViewconsistency group, including whether the group is active, inactive,admin fractured, system fractured, waiting on admin, or invalid.

consistency groupstate

Indicates the current state of the MirrorView consistency group:synchronized, consistent, synchronizing, out-of-sync, scrambled,empty, incomplete, or local only.

consistent state (ofimage)

State in which a MirrorView secondary image is identical to either thecurrent primary image or to some previous instance of the primaryimage.

continuousasynchronous

A RecoverPoint replication mode where each write transaction isacknowledged locally at the source side and then sent to the targetside. The primary advantage of continuous asynchronous replication


120

Glossary

is its ability to provide synchronous-like replication withoutdegrading the performance of host applications.

continuoussynchronous

A RecoverPoint replication mode. In continuous synchronousreplication, the host application that initiates the write waits for anacknowledgment from the replica before continuing. Replication insynchronous mode produces a replica that is 100% up to date withthe production source.

create mirror To establish a remote mirror, that is, use the remote mirror software tocreate data structures on one or more LUNs on specific storagesystems, such that one is the primary image and the other is asecondary image.

Ddata center

migrationsA function that reduces application outage to minutes instead ofhours.

dependency The requirement of one resource needing another resource tofunction properly. The Cluster Enabler resource becomes adependency for physical disk resources in the cluster. Therefore, anyoperations performed on the disk resource cannot be completed untilthe Cluster Enabler resource has been invoked.

device A uniquely addressable part of the storage array consisting of a set ofaccess arms, the associated disk surfaces, and the electronic circuitryrequired to locate, read and write data. Also called a LUN (logicalunit number).

device group A grouping of several devices established to provide configuration,status, and performance data on the collective devices within thegroup.

device number The value that logically identifies a disk device in a string. See also“LUN”.

director The component in the Symmetrix array that allows the Symmetrixarray to transfer data between the host channels and disk devices.

See also channel director and disk director.

disaster recovery A function that recovers data at the disaster recovery site in minutesrather than days.


Glossary

discover A discover action performed in the Cluster Enabler ConfigurationWizard scans the storage array connected to the current node andgathers device information.

disk director The component in the Symmetrix array that interfaces between cacheand the disk devices.

Eestablish A BCV process that assigns a BCV device as the next available mirror

of a standard device.

established The BCV pair condition where the BCV device and standard deviceare synchronized and functioning as a Symmetrix mirror. A BCV pairis established by the BCV commands establish, reestablish, restore, orincremental restore.

Ffailback The action of moving a resource back to the cluster member

designated to be the resource's preferred owner. By default, resourcesare owned by their preferred owners, so a failback would only occurif the resource moved from its preferred owner. This is likely theresult of a failover.

failover The process of taking one or more resources offline on one clustermember and bringing them online on another cluster member.

fault-tolerant Continuous operation in case of failure. A fault-tolerant system canbe created using two or more computers that duplicate all processing,or having one system stand by if the other fails. It can also be builtwith redundant processors, control units, and peripherals.Fault-tolerant operation requires backup power in a power failure. Itmay also imply duplication of systems in disparate locations in theevent of natural catastrophe or vandalism.

FDDl An acronym for Fiber Distributed Data Interface.

Fibre Channel A high-speed serial interface capable of data transfer rates of up to400 MB/s.

Fibre Channel Director The Fibre Channel adapter (FA) in the Symmetrix subsystem thatinterfaces between the host Fibre Channel interface and data storage.It transfers data between the channel and cache.


122

Glossary

forced failover A CE feature allowing you to automatically keep a cluster up on aparticular Symmetrix array or arrays in a total site disaster.

forced quorum Software functionality allowing the cluster to be forced up in theevent that total communication is lost between nodes and Microsoftfailover cluster (or MSCS) wants to shut down the cluster to avoid asplit-brain condition.

See also split-brain condition.

fracture A condition in which I/O is not mirrored to the MirrorViewsecondary image and can be caused when you initiate the fracture(Admin Fracture) or when the system determines that the secondaryimage is unreachable (System Fracture). An admin fracture may alsooccur if the MirrorView software detects an error condition thatrequires administrative intervention to correct.

fracture log A bitmap, maintained in SP memory, that indicates which portions ofthe MirrorView primary image might differ from the secondaryimages. The fracture log is used to shorten the synchronizationprocess after fractures. The bitmap is maintained in SP memory, so ifthe mirror is not configured to use the optional write intent log(which stores the data on disk), and the SP that controls the primaryfails while the secondary image is fractured, the fracture log is lost,and full synchronization of the secondary image is required.

Ggraphical userinterface (GUI)

A method that allows users to interact with the computer and itsspecial applications based on graphics instead of text. GUIs use icons,pictures, and menus and use a mouse as well as a keyboard to acceptinput.

group A collection of resources to be managed as a single unit. Usually, agroup contains all elements needed to run a specific application andfor client systems to connect to the service provided by theapplication. Groups allow an administrator to combine resources intolarger logical units and manage them as a unit. Operations performedon a group affect all resources contained within that group.

HHBA See also host bus adapter (HBA).


Glossary

heartbeat A polling communication mechanism used by the cluster processes todetermine whether the other members of the cluster are alive andworking or have failed. If the heartbeat is not functioning, a failover isinitiated, and another node in the cluster will take over the services.

high availability The characteristic of a computer system/computing environmentthat allows it to continue to provide applications and access to data ifa single component or resource fails. Service is interrupted for only abrief time, and may or may not be apparent to the end users.

host bus adapter(HBA)

A device circuit board that provides an interface between the SCSIbus and the computer I/O bus (for example, PCI, EISA,microchannel).

hyper-volume The term used by the Symmetrix array to make a physical diskappear as multiple physical disks. Each hypervolume has its ownunique SCSI address.

II/O Input/output.

identifier (ID) A sequence of bits or characters that identifies a program, device,controller, or system.

image condition The condition of a MirrorView secondary image provides additionalinformation about the status of updates for the image. Values includenormal, administratively fractured, system fractured, queued to besynchronized, synchronizing, or waiting-on-admin.

image state Indication of the relationship between a MirrorView secondary imageand the primary image of a mirror. The image states are:synchronized, consistent, synchronizing, and out-of-sync.


124

Glossary

incremental establish A TimeFinder BCV or SRDF control operation.

For BCV control operations, an incremental establish causes the BCVdevice to be incrementally synchronized and functioning as aSymmetrix mirrored device. (The devices must have been previouslypaired.) This is the same as an establish operation except anincremental establish is much faster: It copies only the differences ornew storage data from the standard device to the BCV device. Anychanged tracks on the BCV device are overwritten by the data on thecorresponding tracks from the standard device.

For SRDF control operations, an incremental establish causes the target(R2) device to be incrementally synchronized and established as aSymmetrix mirrored device. (The devices must have been previouslypaired.) This is the same as an establish operation except that anincremental establish is much faster: It copies only the differences ornew storage data from the source (R1) device to the target (R2)device. Any changed tracks on the device are overwritten by the dataon the corresponding tracks from the source (R1) device.

incremental restore A TimeFinder BCV or SRDF control operation.

In BCV control operations, an incremental restore is a control operationthat reassigns a BCV device as the next available mirror of thestandard device in the pair. However, the standard devices areupdated with only the data written to the BCV device during the timeof the original pair split. The data written to the standard deviceduring the split is overwritten with data from the BCV mirror.

In SRDF control operations, an incremental restore is a controloperation that reassigns a target (R2) device as the next availablemirror of the source (R1) device in the pair. The source (R1) devicesare updated with only the data written to the target (R2) deviceduring the time of the original pair split. The data written to thesource (R1) device during the split is overwritten with data from thetarget (R2) mirror.

Llateral node Nodes connected to the same Symmetrix array.

LUN A logical unit number (LUN) is a unique identifier used on a SCSIbus that enables it to differentiate between up to eight separatestorage devices (each of which is a logical unit). See also, “devicenumber”.


Glossary

MMajority Node Set

(MNS)A quorum-capable resource based on replicating data to local disksassociated with a majority of cluster nodes. MNS enables you tocreate a server cluster without shared disk for the quorum resource.Cluster Enabler allows you to configure an MNS cluster on WindowsServer 2003 Enterprise and Datacenter Editions.

MicrosoftManagement

Console (MMC)

A Microsoft user interface (UI) framework for use in administratingdifferent components of the Microsoft Windows operating platform.This framework is used to host specific UI/control extensions calledsnap-ins. Use snap-ins to administer both local and remote computers.Third-party snap-ins can be written for use with MMC.

mirrored pair A device comprising two hypervolumes with all data recordedtwice—once on each disk drive.

mirroring A device comprising two hypervolumes with all data recordedtwice—once on each disk drive. The Symmetrix array maintains twoor more identical copies of a set of data on separate disks. Each copyautomatically updates during a write operation. If one disk devicefails, the Symmetrix array automatically uses one of the other copiesfrom another disk drive.

Also, a MirrorView feature that provides disaster recovery bymaintaining one or more mirrors of LUNs on other storage systems.MirrorView can work in conjunction with, but is independent of, theother major CLARiiON software options such as PowerPath softwareand SnapView software. MirrorView works with LUNs in SANstorage systems, and thus can be used to mirror one or more LUNsthat may compose a SAN storage group.

MMC See “Microsoft Management Console (MMC)”.

MSCS Microsoft Cluster Service. A shared-nothing cluster solution forWindows Server Family. In Windows Server 2008, this is now calledMicrosoft Failover Cluster.

Nnetwork interface

card (NIC)A device that provides network communication capabilities to andfrom a computer system.


126

Glossary

nodes Members of a cluster. Also referred to as systems. A node contains aCPU, disk, and network resource.

Ooffline The state of a resource or group that classifies it as unavailable. When

used in context with a cluster member, offline implies the clustermember may not be booted, or the cluster service on the node inquestion may not be functioning properly.

online The state of a resource or group that classifies it as available. Whenused in context with a cluster member, online implies the othercluster members are receiving heartbeats from the cluster member inquestion. See also resource.

out-of-sync state In MirrorView, a remote mirror state in which the software does notknow how the primary and secondary images differ; therefore, a fullsynchronization is required to make the secondary images usable forrecovery. See also, “image state”.

Ppeer node Nodes connected to different Symmetrix arrays located across the

link from each other.

primary image The LUN on the MirrorView production storage system that containsuser data and is the source for data copied to the secondary image.For MirrorView/CE there is one primary image and one secondaryimage. A remote mirror is ineffective for recovery unless it has at leastone secondary image. This manual also refers to primary image asprimary or primary mirror image.

promote (to primary) The operation by which the administrator changes a MirrorViewimage’s role from secondary to primary. As part of this operation, theprevious primary image becomes a secondary image. If the previousprimary image is unavailable when you promote the secondaryimage (perhaps because the primary site suffered a disaster), thesoftware does not include it as a secondary image in the new mirror.A secondary image can be promoted if it is in either the synchronizedstate or the consistent state. An image cannot be promoted if it isout-of-sync or synchronizing.


Glossary

Qquery A command reporting the state of all the BCV devices in the system,

as well as the status of SRDF states.

quiesce threshold The time period after which, without I/O from the server, anyMirrorView secondary image in the consistent state and not fracturedis marked as being in the synchronized state (the default is 60seconds).

quorum disk An ordinary disk volume used as a special communicationmechanism between server systems. In a Microsoft failover cluster (orMSCS), a small amount of cluster system data (a few megabytes) isstored on this volume. The SCSI-3 Reserve and Reset commands areused to move quorum-disk ownership back and forth between nodes.If the heartbeat mechanism fails, the quorum disk is used for eachnode to verify whether the other node is still functioning. Because notall disk products implement these multihost SCSI-3 commands, notall disk products will work in a failover cluster environment. Thus,Microsoft is very rigorous in providing the Cluster/RAID category oftests to qualify disks (refer to Microsoft’s Hardware CompatibilityList) capable of running with Microsoft failover cluster software).

RR1 device See “source (R1) device”.

R2 device See “target (R2) device”.

RA Remote adapter. An RA provides the link connection and fiber opticprotocol support between the local and remote Symmetrix arrays.The RA cable connection is ESCON fibre (ESCON protocol).

RAID Redundant array of independent disks. Data is stored on multiplemagnetic or optical disk drives to increase output performance andstorage capacities and to provide varying degrees of redundancy andfault tolerance. Instead of storing valuable data on a single hard diskthat could fail at any time, RAID ensures a backup copy of allinformation always exists by spreading data among multiple harddisks.

RDF1/RDF2 A type of SRDF device group. Only RDF devices can be assigned toan RDF group. An RDF1 group type contains source (R1) devices andan RDF2 group type contains target (R2) devices.


128

Glossary

RDF State The SRDF state information displayed in a column in the middlepane.

The possible RDF states are the following:

Invalid — The device and link are in an unrecognized combination.

SyncInProg — Synchronizing in progress.

Synchronized — The source and target have identical data.

Split — The source is split from the target and the target is enabled.

Suspended — The link is suspended.

Failed Over — The target is write-enabled, the source iswrite-disabled and the link is suspended.

Partitioned — The communication link to the remote Symmetrixarray is down and the device is write-enabled.

R1 Updated — The target is write-enabled, the source iswrite-disabled and the link is up.

R1 UpdInProg — The target is write-enabled, the source iswrite-disabled, the link is up, but there are invalid tracks between thetarget and the source.

Mixed — This state is only set for a SymDgShow() call when the RDFstates of the devices in the group are different from each other,thereby making the RDF state of the group mixed.

N/A — Not applicable.

Consistent — R2 data is consistent.

recovery policy In MirrorView, the policy for recovering the secondary mirror imageafter a system fracture. If the recovery policy is set to Auto, then thesecondary starts re-synchronizing as soon as the primary imagedetermines that the secondary mirror image is once again accessible.If the policy is set to Manual, then an administrator must explicitlystart a synchronization operation to recover the secondary mirrorimage.

reestablish A business continuance process that reassigns a BCV device as thenext available mirror of the standard device with which it waspreviously paired. The BCV mirror is updated with the data writtento the standard device during the period the BCV pair was split. Thedata written to the BCV device during the split is overwritten by datafrom the standard device.


Glossary

Remote Link Director(RLD)

RLDs create the data link paths between two data storage units. EachSymmetrix array requires a minimum of two, up to a maximum ofeight RLDs, depending on the Symmetrix model in use. Each RLDmanages two ESCON fibre link connections. Each RLD can perform asingle I/O at a time to its paired RLD in the remote Symmetrix array.

RLDs have either an RA1 or RA2 designation. RA1s reside in thesource Symmetrix array. RA2s reside in the target Symmetrix array.These RLDs can also be assigned to an RA group.

See also “RA”.

remote mirror For Symmetrix, the remote mirror refers to a target (R2) devicelocated in a remote Symmetrix array. When a source (R1) device isparticipating in SRDF operations with a target (R2) device, all writesto the R1 device are mirrored to a target (R2) device in a remoteSymmetrix array.

For MirrorView, a remote mirror is the combination of a LUN on onestorage system, called the primary image, and another LUN on adifferent storage system, called the secondary image. The softwaremaintains the secondary image as an exact copy of the primary imageat some (possibly previous) point in time. If the server and/or storagesystem at the primary site fails, you can promote the secondary imageto take over the role of the primary, thus allowing continued access toyour production data.

remote mirroring A feature that provides the means for disaster recovery bymaintaining one or more copies (images) of LUNs at separatelocations.

remote mirror states There are three types of MirrorView mirror states. The mirror statesare active, inactive, and attention.

Active — The remote mirror is running normally.

Inactive — I/O is rejected. This can be a temporary state during someconsistency group operations or a result of an error during aconsistency group operation.

Attention — The state to alert you that the minimum number ofimages required is not currently met. A fracture or the removal of animage can cause this. The mirror will continue to accept I/O in thisstate.

Replication set A RecoverPoint term. A storage volume in the production source thatis replicated must have a corresponding volume at each copy. A


130

Glossary

replication set is production volume and its associated volume at thelocal copy, the remote replica, or both.

resource An object managed by the Cluster Service that sees all resources asidentical opaque objects. Resources may include physical hardwaredevices such as disk drives and network cards, or logical items suchas disk partitions, TCP/IP addresses, entire applications, anddatabases. A resource is said to be online on a node when it isproviding its service on that specific node.

resource failback The movement of resources back to their preferred location in thecluster. This is usually done under manual user control to avoid asituation where a resource is failed back, and then immediately failsover again because of an unresolved node problem. Microsoftfailover cluster (or MSCS) also allows automatic failback andprovides a timing window to try to avoid repeated failovers.

resource failover The process where control of a resource moves to another node of acluster. Failover can be initiated automatically or manually. Wheninitiated automatically, the cluster management software detects afailure of server node hardware or an application. When manuallyinitiated, the cluster administrator uses the Cluster Administratorsoftware application.

resource group A collection of resources to be managed as a single unit. Usually agroup contains all elements needed to run a specific application, andfor client systems to connect to the service provided by theapplication. Groups allow an administrator to combine resources intolarger logical units and manage them together. Operations performedon a group affect all resources contained within that group.

restore A TimeFinder BCV or SRDF control operation.

In BCV control operations, a restore copies a full BCV mirror back tothe standard device in the pair and reassigns the BCV device as thenext available mirror to the standard device.

In SRDF control operations, a restore copies the full target (R2) deviceback to the source (R1) device in the pair and reassigns the target (R2)device as the next available mirror to the source (R1) device.

See also incremental restore.

RF A remote adapter that provides the link connection and fiber opticprotocol support between the local and remote Symmetrix arrays.


Glossary

The RF cable connection is SCSI fibre (SCSI protocol). An RF differsfrom an RA only in the type of connection; an RA uses an ESCONfibre connection, and an RF uses a SCSI fibre connection.

See also “RA”.

Sscalability The ability to add new components to a storage system as system

load increases.

SCSI Small Computer System Interface. SCSI is a high-speed parallelinterface used to connect microcomputers to SCSI peripheral devices,such as disks, printers, and other computers and local area networks.

secondary image For MirrorView, a LUN that contains a mirror of the primary imageLUN.

secondary imagestate

The secondary image states are synchronized, consistent,synchronizing, and out-of-sync. They describe the data on thesecondary storage system in relation to the data on the primarystorage system.

semisynchronousmode

An SRDF mode of operation that provides an asynchronous mode ofoperation. Applications are notified an I/O (or I/O chain) is completeonce the data is in the cache of the local RA1 Symmetrix array. Anynew data is then written to cache in the remote RA2 Symmetrix array.The remote Symmetrix array acknowledges receipt of the data once itis secure in its cache. If source tracks are pending transfer to a target(R2) device, and a second write is attempted to the source (R1) device,the Symmetrix array disconnects (non-immediate retry request), andwaits for the pending track to transfer to the remote Symmetrix array.

snap-in See “Microsoft Management Console (MMC)”.

snapshot A RecoverPoint term. A snapshot is the difference between oneconsistent image of stored data and the next. Snapshots are takenseconds apart. The application writes to storage; at the same time, thesplitter provides a second copy of the writes to the RecoverPointappliance.

Snapshot replicationmode

A RecoverPoint replication mode that only transfers data that haschanged between one consistent image of the storage subsystem and


132

Glossary

the next. By definition, snapshot replication produces a replica that isnot up-to-date.

Solutions Enabler Also known as SYMCLI, an application written using the SymmetrixApplication Programming Interface (SYMAPI) that retrieves datafrom a Symmetrix array using special low-level SCSI commands.

Solutions Enabler allows you to run commands on the host to obtainconfiguration, status, and performance data from the Symmetrixarrays attached to hosts that are running in an open systemsenvironment.

SYMCLI SRDF and TimeFinder components allow you to performcontrol operations on RDF and BCV devices.

source (R1) device A Symmetrix source (R1) device that is participating in SRDFoperations with a target (R2) device. All writes to this device aremirrored to a target (R2) device in a remote Symmetrix array. An R1device must be assigned to an RDF1 group type.

See also RDF1/RDF2.

source unit In an SRDF configuration, it is the primary data storage subsystem. Itinitiates many of the SRDF synchronization activities. An SRDFconfiguration must have at least one source unit and one target unit.See also target unit.

split A business continuance process that removes the BCV mirror fromthe existing BCV pair and assigns the BCV mirror back to its originaldevice address. The BCV device then holds an instant copy of thedata from the standard device.

split-brain condition A total communication failure while both nodes remain operational.A split-brain condition is a potential cause of logical data corruption.For example, if both sides assume the other is dead and beginprocessing new transactions against their copy of the data, twoseparate and unreconcilable copies of the data can be created.

SRDF Symmetrix Remote Data Facility. SRDF consists of the microcode andhardware required to support Symmetrix remote mirroring.

SRDF Asynchronous(SRDF/A)

A high-performance, extended-distance asynchronous replicationusing a delta set architecture for reduced bandwidth requirementsand no host performance impact.


Glossary

Asynchronous mode provides a point-in-time image on the target(R2) device only slightly behind the source (R1) device. SRDF/Asession data is transferred to the remote Symmetrix system in deltasets, eliminating the redundancy of same-track changes beingtransferred over the link, reducing the required bandwidth. SRDF/Aonly needs enough bandwidth to support the average productionworkload versus peak workloads.

SRDF/A is intended for users who require no host application impactwhile maintaining a consistent, restartable image of their data on theR2 side at all times.

SRDF link Fiber optic connections and channels between two Symmetrix arrays.A minimum of two to a maximum of eight links can exist between thetwo units.

stretch cluster A Microsoft cluster that is geographically distributed across multiplephysical locations.

SYMAPI Symmetrix Application Program Interface. See “Application ProgramInterface (API)”.

SYMCLI See “establish”.

synchronize For MirrorView, the process of updating each secondary image withchanges from a primary image. There are several levels ofsynchronization: synchronization based on a fracture log,synchronization based on the optional write intent log, and fullsynchronization (a complete copy). Synchronization based on thefracture or write intent log requires copying only part of the primaryimage to the secondary images.

synchronized state For SRDF, the state in which the data in the R1 device is identical tothat of the R2 device.

For MirrorView, the state in which the data in the secondary image isidentical to that in the primary. On the next write to the primary, theimage state will change to consistent. See also, “secondary imagestate”.

synchronizing state For MirrorView, when a secondary image is in the process ofsynchronizing. The data in the secondary image is not usable forrecovery until the synchronization operation completes. Thus, animage in the synchronizing state cannot be promoted to the primaryimage. See also, “secondary image state”


134

Glossary

synchronous mode An SRDF mode of operation that ensures 100 percent synchronizedmirroring between the two Symmetrix arrays. This is a synchronousmode of operation. Applications are notified that an I/O (or I/Ochain) is complete when the RA2 Symmetrix array acknowledges thatthe data has been secured in cache.

Ttarget (R2) device A Symmetrix target (R2) device participating in SRDF operations

with a source (R1) device. It resides in the remote, or target,Symmetrix array. It is paired with a source (R1) device in the localSymmetrix array and receives all write data from its mirrored pair.This device is not accessed by user applications during normal I/Ooperations. An R2 device must be assigned to an RDF2 group type.See also RDF1/RDF2.

target unit In an SRDF configuration, this subsystem maintains synchronizationwith the devices it is paired with in the source unit. It can serve as asource unit during disaster recovery. An SRDF configuration musthave at least one source unit and one target unit. See also source unit.

Vvirtual servers See “nodes”.

Wwrite intent log (WIL) For MirrorView, the WIL is a record of recent changes to the primary

image. This record is stored in persistent memory on a private LUNreserved for the mirroring software. If the primary storage systemfails (not catastrophically; that is, the WIL LUNs or the persistentmirror storage was lost), the optional write intent log can be used toquickly synchronize the secondary images when the primary storagesystem becomes available. This eliminates the need for fullsynchronization of the secondary images, which can be a lengthyprocess on very large LUNs.

workload migrations Similar to data center migrations; especially useful for minimizingoutages during preventative maintenance of hardware or software.


Date post:	24-Sep-2020
Category:	Documents
Upload:	others
View:	7 times
Download:	0 times

EMC RecoverPoint/Cluster Enabler Plug-in...EMC Corporation Corporate Headquarters: Hopkinton, MA...

Documents