Front cover

IBM XIV Storage System:

Copy Services and Migration

Learn copy and migration functions

and explore practical scenarios

Integrate Snapshots with Tivoli

FlashCopy Manager

Understand SVC-based
migrations

Lisa Martinez
Rosemary McCutchen
Hank Sautter
Bertrand Dufrasne Stephen Solewin
Aubrey Applewhaite Anthony Vandewerdt
David Denny Ron Verbeek
Jawed Iqbal Pete Wendler
Christina Lara Roland Wolf

ibm.com/redbooks
International Technical Support Organization

IBM XIV Storage System: Copy Services and Migration

April 2010

SG24-7759-00
 Note: Before using this information and the product it supports, read the information in “Notices” on
page ix.

First Edition (April 2010)

This edition applies to Version 10, Release 2.1, of the XIV Storage System software.

© Copyright International Business Machines Corporation 2010. All rights reserved.

Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP Schedule
Contract with IBM Corp.
Contents

Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .x

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
The team that wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv

Chapter 1. Snapshots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Snapshots architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Snapshots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.1 Creating a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2.2 Viewing snapshot details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2.3 Deletion priority . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.2.4 Restore a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.2.5 Overwriting snapshots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.2.6 Unlocking a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.2.7 Locking a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.2.8 Deleting a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.2.9 Automatic deletion of a snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.3 Snapshots consistency group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.3.1 Creating a consistency group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
1.3.2 Creating a snapshot using consistency groups. . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.3.3 Managing a consistency group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.3.4 Deleting a consistency group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.4 Snapshot with Remote Mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
1.5 MySQL database backup example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services . 41
2.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.2 Tivoli Storage FlashCopy Manager . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.3 Windows Server 2008 Volume Shadow Copy Service . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.3.1 VSS architecture and components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.3.2 Microsoft Volume Shadow Copy Service function . . . . . . . . . . . . . . . . . . . . . . . . 47
2.4 XIV VSS provider . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.4.1 XIV VSS Provider installation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
2.4.2 XIV VSS Provider configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2.5 Installing and configuring Tivoli Storage FlashCopy Manager for Microsoft Exchange 52
2.6 Backup scenario for Microsoft Exchange Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

Chapter 3. Volume copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

3.1 Volume copy architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.2 Performing a volume copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3.3 Creating an OS image with volume copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Chapter 4. Remote Mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

4.1 XIV Remote Mirroring overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.1.1 XIV Remote Mirror terminology. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72

© Copyright IBM Corp. 2010. All rights reserved. iii

 4.1.2 XIV Remote Mirroring modes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.2 Mirroring schemes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.2.1 Peer designations and roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.2.2 Operational procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
4.2.3 Mirroring status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.3 XIV Remote Mirroring usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.4 XIV Remote Mirroring actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.4.1 Defining the XIV mirroring target. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.4.2 Setting the maximum initialization and synchronization rates. . . . . . . . . . . . . . . . 90
4.4.3 Connecting XIV mirroring ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.4.4 Defining the XIV mirror coupling and peers: volume. . . . . . . . . . . . . . . . . . . . . . . 92
4.4.5 Activating an XIV mirror coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.4.6 Adding volume mirror coupling to consistency group mirror coupling. . . . . . . . . . 97
4.4.7 Normal operation: volume mirror coupling and CG mirror coupling . . . . . . . . . . . 98
4.4.8 Deactivating XIV mirror coupling: change recording . . . . . . . . . . . . . . . . . . . . . . . 99
4.4.9 Changing role of slave volume or CG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
4.4.10 Changing role of master volume or CG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.4.11 Mirror reactivation and resynchronization: normal direction . . . . . . . . . . . . . . . 102
4.4.12 Reactivation, resynchronization, and reverse direction. . . . . . . . . . . . . . . . . . . 103
4.4.13 Switching roles of mirrored volumes or CGs. . . . . . . . . . . . . . . . . . . . . . . . . . . 103
4.4.14 Adding a mirrored volume to a mirrored consistency group . . . . . . . . . . . . . . . 103
4.4.15 Removing a mirrored volume from a mirrored consistency group . . . . . . . . . . 104
4.4.16 Deleting mirror coupling definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
4.5 Best practice usage scenarios . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.5.1 Failure at primary site: switch production to secondary . . . . . . . . . . . . . . . . . . . 107
4.5.2 Complete destruction of XIV 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.5.3 Using an extra copy for DR . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.5.4 Creating application-consistent data at both local and the remote sites . . . . . . . 109
4.5.5 Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.5.6 Adding data corruption protection to disaster recovery protection . . . . . . . . . . . 110
4.5.7 Communication failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.5.8 Temporary deactivation and reactivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.6 Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
4.7 Advantages of XIV mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.8 Mirroring events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.9 Mirroring statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.10 Boundaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
4.11 Using the GUI or XCLI for Remote Mirroring actions . . . . . . . . . . . . . . . . . . . . . . . . 113
4.11.1 Initial setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
4.11.2 Remote mirror target configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
4.11.3 XCLI examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
4.12 Configuring Remote Mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

Chapter 5. Synchronous Remote Mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

5.1 Synchronous mirroring configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.1.1 Volume mirroring setup and activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
5.1.2 Consistency group setup and configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
5.1.3 Coupling activation, deactivation, and deletion . . . . . . . . . . . . . . . . . . . . . . . . . . 132
5.2 Disaster recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
5.3 Role reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3.1 Switching roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.3.2 Change role . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
5.4 Resynchronization after link failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

iv IBM XIV Storage System: Copy Services and Migration

 5.4.1 Last consistent snapshot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
5.4.2 Last consistent snapshot timestamp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.5 Synchronous mirror step-by-step scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
5.5.1 Phase 1: setup and configuration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
5.5.2 Phase 2: disaster at local site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.5.3 Phase 3: recovery of the primary site . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
5.5.4 Phase 4: switching production back to the primary site . . . . . . . . . . . . . . . . . . . 146

Chapter 6. Asynchronous Remote Mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149

6.1 Asynchronous mirroring configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.1.1 Volume mirroring setup and activation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
6.1.2 Consistency group configuration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
6.1.3 Coupling activation, deactivation, and deletion . . . . . . . . . . . . . . . . . . . . . . . . . . 161
6.2 Role reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
6.3 Resynchronization after link failure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.4 Disaster recovery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6.5 Mirroring process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
6.5.1 Initialization process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
6.5.2 Mirroring ongoing operation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.5.3 Mirroring consistency groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.5.4 Mirroring special snapshots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
6.6 Detailed asynchronous mirroring process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.7 Asynchronous mirror step-by-step illustration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
6.7.1 Mirror initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176
6.7.2 Remote backup scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
6.7.3 DR testing scenario. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
6.8 Pool space depletion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182

Chapter 7. Data migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185

7.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
7.2 Handling I/O requests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
7.3 Data migration steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.3.1 Initial connection setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190
7.3.2 Define the host being migrated to the XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
7.3.3 Creating and activating a data migration volume . . . . . . . . . . . . . . . . . . . . . . . . 196
7.3.4 Create and activate a data migration volume on XIV . . . . . . . . . . . . . . . . . . . . . 200
7.3.5 Complete the data migration on XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
7.4 Command-line interface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.4.1 Using XCLI scripts or batch files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
7.4.2 Sample scripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
7.5 Manually creating the migration volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
7.6 Changing and monitoring the progress of a migration . . . . . . . . . . . . . . . . . . . . . . . . 208
7.6.1 Changing the synchronization rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
7.6.2 Monitoring migration speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.6.3 Monitoring migration via the XIV event log . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
7.6.4 Monitoring migration speed via the fabric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
7.6.5 Monitoring migration speed via the non-XIV storage . . . . . . . . . . . . . . . . . . . . . 211
7.7 Thick-to-thin migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
7.8 Resizing the XIV volume after migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
7.9 Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.9.1 Target connectivity fails . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.9.2 Remote volume LUN is unavailable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
7.9.3 Local volume is not formatted . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218

Contents v
 7.9.4 Host server cannot access the XIV migration volume. . . . . . . . . . . . . . . . . . . . . 218
7.9.5 Remote volume cannot be read . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
7.9.6 LUN is out of range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7.10 Backing out of a data migration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7.10.1 Back-out prior to migration being defined on the XIV . . . . . . . . . . . . . . . . . . . . 219
7.10.2 Back-out after a data migration has been defined but not activated . . . . . . . . . 219
7.10.3 Back-out after a data migration has been activated but is not complete. . . . . . 219
7.10.4 Back-out after a data migration has reached the synchronised state . . . . . . . . 220
7.11 Migration checklist. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
7.12 Device-specific considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
7.12.1 EMC CLARiiON. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.12.2 EMC Symmetrix and DMX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
7.12.3 HDS TagmaStore USP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
7.12.4 HP EVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
7.12.5 IBM DS3000/DS4000/DS5000 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
7.12.6 IBM ESS E20/F20/800 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
7.12.7 IBM DS6000 and DS8000. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
7.13 Sample migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231

Chapter 8. SVC migration with XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243

8.1 Steps to take when using SVC migration with XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
8.2 XIV and SVC interoperability. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
8.2.1 Firmware versions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
8.2.2 Copy functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.2.3 TPC with XIV and SVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.3 Zoning setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
8.3.1 Capacity on demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
8.3.2 Determining XIV WWPNs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
8.3.3 Hardware dependencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.3.4 Sharing an XIV with another SVC cluster or non-SVC hosts . . . . . . . . . . . . . . . 248
8.3.5 Zoning rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.4 Volume size considerations for XIV with SVC. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
8.4.1 SCSI queue depth considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249
8.4.2 XIV volume sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
8.4.3 Creating XIV volumes that are exactly the same size as SVC VDisks . . . . . . . . 252
8.4.4 SVC 2TB volume limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8.4.5 MDisk group creation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8.4.6 SVC MDisk group extent sizes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 252
8.5 Using an XIV for SVC quorum disks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
8.6 Configuring an XIV for attachment to SVC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
8.6.1 XIV setup steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
8.6.2 SVC setup steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
8.7 Data movement strategy overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.7.1 Using SVC migration to move data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259
8.7.2 Using VDisk mirroring to move the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
8.7.3 Using SVC migration with image mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
8.8 Using SVC migration to move data to XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
8.8.1 Determine the required extent size and VDisk candidates . . . . . . . . . . . . . . . . . 261
8.8.2 Create the MDisk group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
8.8.3 Migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
8.9 Using VDisk mirroring to move the data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
8.9.1 Determine the required extent size and VDisk candidates . . . . . . . . . . . . . . . . . 263
8.9.2 Create the MDisk group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264

vi IBM XIV Storage System: Copy Services and Migration

 8.9.3 Set up the IO group for mirroring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
8.9.4 Create the mirror . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
8.9.5 Validating a VDisk copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266
8.9.6 Removing the VDisk copy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
8.10 Using SVC migration with image mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
8.10.1 Create image mode destination volumes on the XIV . . . . . . . . . . . . . . . . . . . . 267
8.10.2 Migrate the VDisk to image mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
8.10.3 Outage step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
8.10.4 Bring the VDisk online. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 270
8.10.5 Migration from image mode to managed mode . . . . . . . . . . . . . . . . . . . . . . . . 271
8.10.6 Remove image mode MDisks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
8.10.7 Use transitional space as managed space . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
8.10.8 Remove non-XIV MDisks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 272
8.11 Future configuration tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
8.11.1 Adding additional capacity to the XIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
8.11.2 Using additional XIV host ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
8.12 Understanding the SVC controller path values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
8.13 SVC with XIV implementation checklist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275

Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277

IBM Redbooks publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Other publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
How to get IBM Redbooks publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Help from IBM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279

Contents vii
viii IBM XIV Storage System: Copy Services and Migration
Notices

This information was developed for products and services offered in the U.S.A.

IBM may not offer the products, services, or features discussed in this document in other countries. Consult
your local IBM representative for information on the products and services currently available in your area. Any
reference to an IBM product, program, or service is not intended to state or imply that only that IBM product,
program, or service may be used. Any functionally equivalent product, program, or service that does not
infringe any IBM intellectual property right may be used instead. However, it is the user's responsibility to
evaluate and verify the operation of any non-IBM product, program, or service.

IBM may have patents or pending patent applications covering subject matter described in this document. The
furnishing of this document does not give you any license to these patents. You can send license inquiries, in
writing, to:
IBM Director of Licensing, IBM Corporation, North Castle Drive, Armonk, NY 10504-1785 U.S.A.

The following paragraph does not apply to the United Kingdom or any other country where such
provisions are inconsistent with local law: INTERNATIONAL BUSINESS MACHINES CORPORATION
PROVIDES THIS PUBLICATION "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESS OR
IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF NON-INFRINGEMENT,
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Some states do not allow disclaimer of
express or implied warranties in certain transactions, therefore, this statement may not apply to you.

This information could include technical inaccuracies or typographical errors. Changes are periodically made
to the information herein; these changes will be incorporated in new editions of the publication. IBM may make
improvements and/or changes in the product(s) and/or the program(s) described in this publication at any time
without notice.

Any references in this information to non-IBM Web sites are provided for convenience only and do not in any
manner serve as an endorsement of those Web sites. The materials at those Web sites are not part of the
materials for this IBM product and use of those Web sites is at your own risk.

IBM may use or distribute any of the information you supply in any way it believes appropriate without incurring
any obligation to you.

Information concerning non-IBM products was obtained from the suppliers of those products, their published
announcements or other publicly available sources. IBM has not tested those products and cannot confirm the
accuracy of performance, compatibility or any other claims related to non-IBM products. Questions on the
capabilities of non-IBM products should be addressed to the suppliers of those products.

This information contains examples of data and reports used in daily business operations. To illustrate them
as completely as possible, the examples include the names of individuals, companies, brands, and products.
All of these names are fictitious and any similarity to the names and addresses used by an actual business
enterprise is entirely coincidental.

COPYRIGHT LICENSE:

This information contains sample application programs in source language, which illustrate programming
techniques on various operating platforms. You may copy, modify, and distribute these sample programs in
any form without payment to IBM, for the purposes of developing, using, marketing or distributing application
programs conforming to the application programming interface for the operating platform for which the sample
programs are written. These examples have not been thoroughly tested under all conditions. IBM, therefore,
cannot guarantee or imply reliability, serviceability, or function of these programs.

© Copyright IBM Corp. 2010. All rights reserved. ix

Trademarks
IBM, the IBM logo, and ibm.com are trademarks or registered trademarks of International Business Machines
Corporation in the United States, other countries, or both. These and other IBM trademarked terms are
marked on their first occurrence in this information with the appropriate symbol (® or ™), indicating US
registered or common law trademarks owned by IBM at the time this information was published. Such
trademarks may also be registered or common law trademarks in other countries. A current list of IBM
trademarks is available on the Web at http://www.ibm.com/legal/copytrade.shtml

The following terms are trademarks of the International Business Machines Corporation in the United States,
other countries, or both:
AIX® FlashCopy® System Storage™
BladeCenter® IBM® System x®
DB2® Lotus® System z®
Domino® Redbooks® Tivoli®
DS4000® Redpaper™ WebSphere®
DS6000™ Redbooks (logo) ® XIV®
DS8000® S/390®

The following terms are trademarks of other companies:

ITIL is a registered trademark, and a registered community trademark of the Office of Government
Commerce, and is registered in the U.S. Patent and Trademark Office.

Snapshot, and the NetApp logo are trademarks or registered trademarks of NetApp, Inc. in the U.S. and other
countries.

VMware, the VMware "boxes" logo and design are registered trademarks or trademarks of VMware, Inc. in the
United States and/or other jurisdictions.

Microsoft, Windows, and the Windows logo are trademarks of Microsoft Corporation in the United States,
other countries, or both.

UNIX is a registered trademark of The Open Group in the United States and other countries.

Linux is a trademark of Linus Torvalds in the United States, other countries, or both.

Other company, product, or service names may be trademarks or service marks of others.

x IBM XIV Storage System: Copy Services and Migration

Preface

This IBM® Redbooks® publication provides a practical understanding of the XIV® Storage
System copy and migration functions. The XIV Storage System has a rich set of copy
functions suited for various data protection scenarios, which enables clients to enhance their
business continuance, data migration, and online backup solutions. These functions allow
point-in-time copies, known as snapshots and full volume copies, and also include remote
copy capabilities in either synchronous or asynchronous mode. These functions are included
in the XIV software and all their features are available at no additional charge.

The various copy functions are reviewed under separate chapters that include detailed
information about usage, as well as practical illustrations.

This book also discusses how to integrate the snapshot function with the IBM Tivoli®
FlashCopy® manager, explains the XIV built-in migration capability, and presents migration
alternatives based on the San Volume Controller (SVC).

Note: GUI and XCLI illustrations included in this book were created with an early version of
the 10.2.1 code, as available at the time of writing. There could be minor differences with
the XIV 10.2.1 code that is publicly released.

This book is intended for anyone who needs a detailed and practical understanding of the XIV
copy functions.

The team who wrote this book

This book was produced by a team of specialists from around the world working at the
International Technical Support Organization, San Jose Center.

Bertrand Dufrasne is an IBM Certified Consulting IT Specialist and Project Leader for
System Storage™ disk products at the International Technical Support Organization, San
Jose Center. He has worked at IBM in various IT areas. He has authored many IBM
Redbooks publications and has also developed and taught technical workshops. Before
joining the ITSO, he worked for IBM Global Services as an Application Architect. He holds a
Master’s degree in Electrical Engineering from the Polytechnic Faculty of Mons.

Aubrey Applewhaite is an IBM Certified Consulting IT Specialist working for the Storage
Services team in the UK. He has worked for IBM since 1996 and has over 20 years of
experience in the IT industry. He has worked in a number of areas, including System x®
servers, operating system administration, and technical support. He currently works in a
customer-facing role providing advice and practical expertise to help IBM customers
implement new storage technology. He specializes in XIV, SVC, DS8000®, and DS5000
hardware. He holds a Bachelor of Science degree in Sociology and Politics from Aston
University and is also a VMware Certified Professional.

David Denny is a Solutions Architect with XIV in the IBM Systems and Technology Group.
David has over 20 years of experience in the IT field, ranging from systems administration to
enterprise storage architect. David is the lead corporate resource for data migrations with XIV.
Prior to joining IBM, David was a Lead Architect of the Enterprise SAN for the DoD Disaster

© Copyright IBM Corp. 2010. All rights reserved. xi

 Recovery Program at the Pentagon following the events of 9/11. He holds a Bachelor of Arts
degree as well a Bachelor of Science degree in Computer Science from Lynchburg College.

Jawed Iqbal is an Advisory Software Engineer and a Team Lead for Tivoli Storage Manager
Client, Data Protection, and FlashCopy Manager products at the IBM Almaden Research
Center in San Jose, CA. Jawed joined IBM in 2000 and worked as Test Lead on several Data
Protection products, including Oracle RDBMS Server, WebSphere®, MS SQL, MS Exchange,
and Lotus® Domino® Server. He holds a master’s degree in Computer Science, a BBA in
Computer Information Systems, and a bachelor’s degree in Math, Stats, and Economics.
Jawed also holds an ITIL® certification.

Christina Lara is a Senior Test Engineer currently working on the XIV storage test team in
Tucson, AZ. She just completed a 1-year assignment as a Assistant Technical Staff Member
(ATSM) to the Systems Group Chief Test Engineer. Christina has just begun her ninth year
with IBM, having held different test and leadership positions within the Storage Division over
that last several years. Her responsibilities included system level testing and field support test
on both DS8000 and ESS800 storage products and test project management. Christina
graduated from the University of Arizona in 1991 with a BSBA in MIS and Operations
Management. In 2002, she received her MBA in Technology Management from the University
of Phoenix.

Lisa Martinez is a Senior Software Engineer working in the DS8000 and XIV System Test
Architecture in Tucson, Arizona. She has extensive experience in Enterprise Disk Test. She
holds a Bachelor of Science degree in Electrical Engineering from the University of New
Mexico and a Computer Science degree from New Mexico Highlands University. Her areas of
expertise include the XIV Storage System and IBM System Storage DS8000, including Copy
Services, with Open Systems and System z®.

Rosemary McCutchen has over 20 years of IT experience and is currently a Certified

Consulting IT Specialist working in Storage ATS in IBM Gaithersburg. There she is
responsible for XIV customer demonstrations, proof of concepts, and workshops, as well as
XIV beta testing. Rosemary has extensive hands-on experience with XIV and has authored
multiple XIV white papers and XIV training documents.

Hank Sautter is a Consulting IT Specialist with Advanced Technical Support in the U.S. He
has 17 years of experience with S/390® and IBM disk storage hardware and Advanced Copy
Services functions working in Tucson, Arizona. His previous 13 years of experience include
IBM Processor microcode development and S/390 system testing while working in
Poughkeepsie, NY. He has worked at IBM for 30 years. Hank's areas of expertise include
enterprise storage performance and disaster recovery implementation for large systems and
open systems. He writes and presents on these topics. He holds a BS degree in Physics.

Stephen Solewin is an XIV Corporate Solutions Architect based in Tucson, Arizona. He has
13 years of experience working on IBM storage, including Enterprise and Midrange Disk, LTO
drives and libraries, SAN, Storage Virtualization, and software. Steve has been working on
the XIV product line since March of 2008, working with both clients and various IBM teams
worldwide. Steve holds a Bachelor of Science degree in Electrical Engineering from the
University of Arizona, where he graduated with honors.

Anthony Vandewerdt is a Senior IT Specialist who currently works for IBM STG Storage
Systems Sales in Australia. He has 21 years of experience providing pre-sales and post-sales
technical support at IBM. Anthony has extensive hands-on experience with nearly all IBM
storage products, especially DS8000, SVC, XIV, ESS800, and Brocade and Cisco SAN
switches. He has worked in a wide variety of post-sales technical support roles including
country and Asia Pacific storage support. Anthony has also worked as an instructor for STG
Education.

xii IBM XIV Storage System: Copy Services and Migration

 Ron Verbeek is a Senior Consulting IT Specialist with Storage and Data System Services,
IBM Global Technology Services Canada. He has over 22 years of experience in the
computing industry, with the last 10 years spent working on storage and data solutions. He
holds multiple product and industry certifications, including SNIA Storage Architect. Ron
spends most of his client time in technical pre-sales solutioning, defining, and architecting
storage optimization solutions. He has extensive experience in data transformation services
and information life-cycle consulting. He holds a Bachelor of Science degree in Mathematics
from McMaster University in Canada.

Pete Wendler is a Software Engineer for IBM Systems and Technology Group, Storage
Platform, located in Tucson, Arizona. In his 10 years working for IBM, Peter has worked in
client support for enterprise storage products, solutions testing, and development of the IBM
DR550 archive appliance. He currently holds a position in technical marketing at IBM. Peter
received a Bachelor of Science degree from Arizona State University in 1999.

Roland Wolf is a Certified IT Specialist in Germany. He has worked for IBM for 23 years and
has 15 years of experience with high-end disk storage hardware in S/390 and Open Systems
environments. He is working in Field Technical Sales Support for storage systems. His areas
of expertise include performance analysis and disaster recovery solutions in enterprises
utilizing the unique capabilities and features of the IBM disk storage servers, DS8000 and
XIV. He has contributed to various IBM Redbooks publications including ESS, DS80000
Architecture, and DS8000 Copy Services. He holds a Ph.D. in Theoretical Physics.

Special thanks to Rami Elron for his help with and advice on many of the topics covered in
this book.

Thanks to the following people for their contributions to this project:

John Bynum, Aviad Offer, Jim Segdwick, Brian Sherman, Juan Yanes

Now you can become a published author, too!

Here's an opportunity to spotlight your skills, grow your career, and become a published
author - all at the same time! Join an ITSO residency project and help write a book in your
area of expertise, while honing your experience using leading-edge technologies. Your efforts
will help to increase product acceptance and customer satisfaction, as you expand your
network of technical contacts and relationships. Residencies run from two to six weeks in
length, and you can participate either in person or as a remote resident working from your
home base.

Find out more about the residency program, browse the residency index, and apply online at:
ibm.com/redbooks/residencies.html

Preface xiii
Comments welcome
Your comments are important to us!

We want our books to be as helpful as possible. Send us your comments about this book or
other IBM Redbooks publications in one of the following ways:
򐂰 Use the online Contact us review Redbooks form found at:
ibm.com/redbooks
򐂰 Send your comments in an e-mail to:
[email protected]
򐂰 Mail your comments to:
IBM Corporation, International Technical Support Organization
Dept. HYTD Mail Station P099
2455 South Road
Poughkeepsie, NY 12601-5400

Stay connected to IBM Redbooks

򐂰 Find us on Facebook:
http://www.facebook.com/pages/IBM-Redbooks/178023492563?ref=ts
򐂰 Follow us on twitter:
http://twitter.com/ibmredbooks
򐂰 Look for us on LinkedIn:
http://www.linkedin.com/groups?home=&gid=2130806
򐂰 Explore new Redbooks publications, residencies, and workshops with the IBM Redbooks
weekly newsletter:
https://www.redbooks.ibm.com/Redbooks.nsf/subscribe?OpenForm
򐂰 Stay current on recent Redbooks publications with RSS Feeds:
http://www.redbooks.ibm.com/rss.html

xiv IBM XIV Storage System: Copy Services and Migration

 1

Chapter 1. Snapshots
The XIV Storage System has a rich set of copy functions suited for various data protection
scenarios, which enables clients to enhance their business continuance, data migration, and
online backup solutions. This chapter provides an overview of the snapshot function for the
XIV product.

A snapshot is a point-in-time copy of a volume’s data. The XIV snapshot is based on several
innovative technologies to ensure minimal degradation of or impact on system performance.

A volume copy is an exact copy of a system volume and differs in approach to a snapshot in
that a full data copy is performed in the background. Snapshots make use of pointers and do
not necessarily copy all the data to the second instance of a volume.

With these definitions in mind, we explore the architecture and functions of snapshots within
the XIV Storage System.

© Copyright IBM Corp. 2010. All rights reserved. 1

1.1 Snapshots architecture
Before we begin discussing snapshots we provide a short review of XIV’s architecture. For
more information refer to IBM XIV Storage System: Architecture, Implementation, and Usage,
SG24-7659.

The XIV system consists of several servers with 12 disk drives each and memory that acts as
cache. All the servers are connected to each other and certain servers act as interface
servers to the SAN and the host servers (Figure 1-1).

Server

Network (FC/Ethernet)

Module 4 Module 5 Module 6 Module 7 Module 8 Module 9

Ethernet

Switch 1 Switch 2
Module 1 Module 15

Module 2 Module 3 Module 10 Module 11 Module 12 Module 13 Module 14

Figure 1-1 XIV architecture: modules and disk drives

2 IBM XIV Storage System: Copy Services and Migration

When a logical volume or LUN is created on an XIV system, the volume’s data is divided into
pieces 1 MB in size, called partitions. Each partition is duplicated for data protection and
stored on disks of different modules. All partitions of a volume are pseudo-randomly
distributed across the modules and disk drives, as shown in Figure 1-2.

XIV Architecture
• Split volume data in 1MB • Store both copies in
partitions different modules
• Maintain a copy of each • Spread data of a volume
partition across all disk drives
pseudo randomly

Volume

Da ta M odule 1 Da ta M odule 2

Da ta M odule 3

Figure 1-2 XIV architecture: distribution of data

Chapter 1. Snapshots 3
 A logical volume is represented by pointers to partitions that make up the volume. If a
snapshot is taken of a volume, the pointers are just copied to form the snapshot volume, as
shown in Figure 1-3. No space is consumed for the snapshot volume up to now.

Vol

• Logical volume and its

partitions: Partitions
are spread across all
disk drives and
actually each partition
exists two times (not
shown here)
Vol snap

• A snapshot of a
volume is taken.
Pointers point to the
same partitions as the
original volume

• There is an update of Vol snap

a data partition of the
original volume. The
updated partition is
written to a new
location.

Figure 1-3 XIV architecture: snapshots

When an update is performed on the original data, the update is stored in a new position and
a pointer of the original volume now points to the new partition, whereas the snapshot volume
still points to the old partition. Now we use up more space for the original volume and its
snapshot and it has the size of a partition (1 MB). This method is called redirect-on-write.

4 IBM XIV Storage System: Copy Services and Migration

It is important to note that data on a volume comprises two fundamental building blocks.
Metadata is information about how the data is stored on the physical volume and the data
itself in the blocks. Metadata management is the key to rapid snapshot performance. A
snapshot points to the partitions of its master volume for all unchanged partitions. When the
data is modified, a new partition is allocated for the modified data. In other words, the XIV
Storage System manages a set of pointers based on the volume and the snapshot. Those
pointers are modified when changes are made to the user data. Managing pointers to data
enables XIV to instantly create snapshots, as opposed to physically copying the data into a
new partition. Refer to Figure 1-4.

Data layout before modification

Empty
Empty
Snapshot Pointer Volume A Volume Pointer
to Partition to Partition

Host modifies data in Volume A

Empty
Volume A Volume Pointer
to Partition
Snapshot Pointer Snapshot of A
to Partition

Figure 1-4 Example of a redirect-on-write operation

The actual metadata overhead for a snapshot is small. When the snapshot is created, the
system does not require new pointers because the volume and snapshot are exactly the
same, which means that the time to create the snapshot is independent of the size or number
of snapshots present in the system. As data is modified, new metadata is created to track the
changes to the data.

Note: The XIV system minimizes the impact to the host for write operations by performing
a redirect-on-write operation. As the host writes data to a volume with a snapshot
relationship, the incoming information is placed into a newly allocated partition. Then the
pointer to the data for the master volume is modified to point at the new partition. The
snapshot volume continues to point at the original data partition.

Because the XIV Storage System tracks the snapshot changes on a partition basis, data is
only copied when a transfer is less than the size of a partition. For example, a host writes
4 KB of data to a volume with a snapshot relationship. The 4 KB is written to a new partition,
but in order for the partition to be complete, the remaining data must be copied from the
original partition to the newly allocated partition.

The alternative to redirect-on-write is the copy on write function. Most other systems do not
move the location of the volume data. Instead, when the disk subsystem receives a change, it
copies the volume’s data to a new location for the point-in-time copy. When the copy is
complete, the disk system commits the newly modified data. Therefore, each individual
modification takes longer to complete, as the entire block must be copied before the change
can be made.

As the storage assigned to the snapshot is completely utilized, the XIV Storage System
implements a deletion mechanism to protect itself from overutilizing the set pool space.
Deletion of snapshots is further explained in 1.2.8, “Deleting a snapshot” on page 19.

Chapter 1. Snapshots 5
 If you know in advance that an automatic deletion is possible, a pool can be expanded to
accommodate additional snapshots. This function requires that there is available space on
the system for the storage pool. See Figure 1-5.

Snapshot space on a single disk

Snapshot free partition

Utilization before a
Snapshot 2 new allocation
Snapshot 1

Snapshot 3 Snapshot 3 allocates

Snapshot 3 a partition and
Snapshot 2 Snapshot 1 is
Snapshot 1 deleted, because
there must always
be at least one free
partition for any
subsequent snapshot.
Snapshot 3
Snapshot 2
Snapshot free partition

Figure 1-5 Diagram of automatic snapshot deletion

Each snapshot has a deletion priority property that is set by the user. There are four priorities,
with 1 being the highest priority and 4 being the lowest priority. The system uses this priority
to determine which snapshot to delete first. The lowest priority becomes the first candidate for
deletion. If there are multiple snapshots with the same deletion priority, the XIV system
deletes the snapshot that was created first. Refer to 1.2.3, “Deletion priority” on page 12 for
an example of working with deletion priorities.

A snapshot also has a unique ability to be unlocked. By default, a snapshot is locked on

creation and is only readable. Unlocking a snapshot allows the user to modify the data in the
snapshot for post-processing.

When unlocked, the snapshot takes on the properties of a volume and can be resized or
modified. As soon as the snapshot has been unlocked, the modified property is set. The
modified property cannot be reset after a snapshot is unlocked, even if the snapshot is
relocked without modification.

In certain cases, it might be important to duplicate a snapshot. When duplicating a snapshot,

the duplicate snapshot points to the original data and has the same creation date as the
original snapshot, if the first snapshot has not been unlocked. This feature can be beneficial
when the user wants to have one copy for a backup and another copy for testing purposes.

If the first snapshot is unlocked and the duplicate snapshot already exists, the creation time
for the duplicate snapshot does not change. The duplicate snapshot points to the original
snapshot. If a duplicate snapshot is created from the unlocked snapshot, the creation date is
the time of duplication and the duplicate snapshot points at the original snapshot.

6 IBM XIV Storage System: Copy Services and Migration

For the further discussion we must introduce two other terms, pools and consistency groups
(Figure 1-6).

Terminology
• Storage Pool Storage Pool
– Administrative construct
for controlling usage of
data capacity Consistency Group
• Volume
– Data capacity spreads
across all disks in IBM
XIV system Volume Volume
• Snapshot
– Point in time image
– Same storage pool as
source
Snapshot Snapshot
• Consistency group
– Multiple volumes that Snapshot Group
require consistent
snapshot creation
– All in same storage pool
• Snapshot group
– Group of consistent
snapshots

Figure 1-6 XIV terminology

A storage pool is just a logical entity that represents storage capacity. Volumes are created in
a storage pool and snapshots of a volume are within the same storage pool. Because
snapshots require capacity as the source and the snapshot volume differ over time, space for
snapshots must be set aside when defining a storage pool (Figure 1-7). A storage pool can
be resized as needed as long as there is enough free capacity in the XIV Storage System.

Figure 1-7 Creating a storage pool with capacity for snapshots

Chapter 1. Snapshots 7
 An application can utilize many volumes on the XIV Storage System. For example, a
database application can span several volumes for journaling and user data. In this case, the
snapshot for the volumes must occur at the same moment in time so that the journal and data
are synchronized. The consistency group allows the user to perform the snapshot on all the
volumes assigned to the group at the same moment in time, therefore enforcing data
consistency.

The XIV Storage System creates a special snapshot related to the Remote Mirroring
functionality. During the recovery process of lost links, the system creates a snapshot of all
the volumes in the system. This snapshot is used if the synchronization process fails. The
data can be restored to a point of known consistency. A special value of the deletion priority is
used to prevent the snapshot from being automatically deleted. Refer to 1.4, “Snapshot with
Remote Mirror” on page 33, for an example of this snapshot.

1.2 Snapshots
The creation and management of snapshots with the XIV Storage System are simple and
easy to perform. This section guides you through the life cycle of a snapshot, providing
examples of how to interact with the snapshots using the GUI. This section also discusses
duplicate snapshots and the automatic deletion of snapshots.

1.2.1 Creating a snapshot

Snapshot™ creation is a simple and easy task to accomplish. Using the Volumes and
snapshots view, right-click the volume and select Create Snapshot. Figure 1-8 depicts how
to make a snapshot of the ITSO_Volume volume.

Figure 1-8 Creating a snapshot

8 IBM XIV Storage System: Copy Services and Migration

The new snapshot is displayed in Figure 1-9. The XIV Storage System uses a specific naming
convention. The first part is the name of the volume followed by the word snapshot and then a
number or count of snapshots for the volume. The snapshot is the same size as the master
volume. However, it does not display how much space has been used by the snapshot.

Figure 1-9 View of the new snapshot

From this view shown in Figure 1-9, there are three other details:
򐂰 First is the locked property of the snapshot. By default, a snapshot is locked, which means
the it is write inhibited at the time of creation.
򐂰 Secondly, the modified property is displayed to the right of the locked property. In this
example, the snapshot has not been modified.
򐂰 Third, the creation date is displayed. For this example, the snapshot was created on
12 June 2009 at 21:39.

You might want to create a duplicate snapshot, for example, if you want to keep this snapshot
as it is and you want another one that you want to modify,

The duplicate has the same creation date as the first snapshot, and it also has a similar
creation process. From the Volumes and snapshots view, right-click the snapshot to duplicate.
Select Duplicate from the menu to create a new duplicate snapshot. Figure 1-10 provides an
example of duplicating the snapshot, ITSO_Volume.snapshot_00001.

Figure 1-10 Creating a duplicate snapshot

After selecting Duplicate from the menu, the duplicate snapshot is displayed directly under
the original snapshot.

Chapter 1. Snapshots 9
 Note: The creation date of the duplicate snapshot in Figure 1-11 is the same creation date
as the original snapshot. Even though it is not shown, the duplicate snapshot points to the
master volume, not the original snapshot.

Figure 1-11 View of the new duplicate snapshot

Example 1-1 provides an example of creating a snapshot and a duplicate snapshot with the
Extended Command Line Interface (XCLI).

In the following examples we use the XIV Session XCLI. You could also use the XCLI
command. In this case, however, specify the configuration file or the IP address of the XIV
that you are talking to as well as the user ID and password. Use the XCLI command to
automate tasks with batch jobs. For simplicity, we used the XIV Session XCLI in our
examples.

Example 1-1 Creating a snapshot and a duplicate with the XCLI Session
snapshot_create vol=ITSO_Volume
snapshot_duplicate snapshot=ITSO_Volume.snapshot_00001

After the snapshot is created, it must be mapped to a host in order to access the data. This
action is performed in the same way as mapping a normal volume.

Important: A snapshot is an exact replica of the original volume. Certain hosts do not
properly handle having two volumes with the same exact metadata describing them. In
these cases, you must map the snapshot to a different host to prevent failures.

Creation of a snapshot is only done in the volume’s storage pool. A snapshot cannot be
created in a storage pool other than the one that owns the volume. If a volume is moved to
another storage pool, the snapshots are moved with the volume to the new storage pool
(provided that there is enough space).

10 IBM XIV Storage System: Copy Services and Migration

1.2.2 Viewing snapshot details
After creating the snapshots, you might want to view the details of the snapshot for creation
date, deletion priority, and whether the volume has been modified. Using the GUI, select
Snapshot Tree from the Volumes menu, as shown in Figure 1-12.

Figure 1-12 Selecting the Snapshot Tree view

The GUI displays all the volumes in a list.

Scroll down to the snapshot of interest and select the snapshot by clicking its name. Details of
the snapshot are displayed in the upper right panel. Looking at the volume ITSO_Volume, it
contains a snapshot 00001 and a duplicate snapshot 00002. The snapshot and the duplicate
snapshot have the same creation date of 2009-06-12 21:39:08, as shown in Figure 1-13. In
addition, the snapshot is locked, has not been modified, and has a deletion priority of 1 (which
is the highest priority, so it will be deleted last).

Figure 1-13 Viewing the snapshot details

Chapter 1. Snapshots 11
 Along with these properties, the tree view shows a hierarchal structure of the snapshots. This
structure provides details about restoration and overwriting snapshots. Any snapshot can be
overwritten by any parent snapshot, and any child snapshot can restore a parent snapshot or
a volume in the tree structure.

In Figure 1-13 on page 11, the duplicate snapshot is a child of the original snapshot, or in
other words, the original snapshot is the parent of the duplicate snapshot. This structure has
nothing to do with how the XIV Storage System manages the pointers with the snapshots, but
is intended to provide an organizational flow for snapshots.

Example 1-2 is an example of viewing the snapshot data in the XCLI Session. Due to space
limitations, only a small portion of the data is displayed from the output.

Example 1-2 Viewing the snapshots on the XCLI session

snapshot_list vol=ITSO_Volume

Name Size (GB) Master Name

ITSO_Volume.snapshot_00001 17 ITSO_Volume
ITS0_Volume.snapshot_00002 17 ITSO_Volume

1.2.3 Deletion priority

Deletion priority enables the user to rank the importance of the snapshots within a pool. For
the current example, the duplicate snapshot ITSO_Volume.snapshot_00002 is not as
important as the original snapshot ITSO_Volume.snapshot_00001. Therefore, the deletion
priority is reduced.

If the snapshot space is full, the duplicate snapshot is deleted first even though the original
snapshot is older.

To modify the deletion priority, right-click the snapshot in the Volumes and snapshots view
and select Change Deletion Priority, as shown in Figure 1-14.

Figure 1-14 Changing the deletion priority

12 IBM XIV Storage System: Copy Services and Migration

After clicking Change Deletion Priority, select the desired deletion priority from the dialog
window and accept the change by clicking OK. Figure 1-15 shows the four options that are
available for setting the deletion priority. The lowest priority setting is 4, which causes the
snapshot to be deleted first. The highest priority setting is 1, and these snapshots are deleted
last. All snapshots have a default deletion priority of 1, if not specified on creation.

Figure 1-15 Lowering the priority for a snapshot

Figure 1-16 confirms that the duplicate snapshot has had its deletion priority lowered to 4. As
shown in the upper right panel, the delete priority is reporting a 4 for snapshot
ITSO_Volume.snapshot_00002.

Figure 1-16 Confirming the modification to the deletion priority

To change the deletion priority for the XCLI Session, specify the snapshot and new deletion
priority, as illustrated in Example 1-3.

Example 1-3 Changing the deletion priority for a snapshot

snapshot_change_priority snapshot=ITSO_Volume.snapshot_00002 delete_priority=4

Chapter 1. Snapshots 13
 The GUI also lets you specify the deletion priority when you create the snapshot. Instead of
selecting Create Snapshot, you select Create Snapshot (Advanced), as shown in
Figure 1-17).

Figure 1-17 Create Snapshot Advanced

A panel is presented that allows you to specify the deletion priority, but it also allows you to
use your own volume name for the snapshot.

Figure 1-18 Advanced snapshot options

1.2.4 Restore a snapshot

The XIV Storage System provides the ability to restore the data from a snapshot back to the
master volume, which can be helpful for operations where data was modified incorrectly and
you want to restore the data. From the Volumes and snapshots view, right-click the volume
and click Restore. This action opens a dialog box where you can select which snapshot is to
be used to restore the volume. Click OK to perform the restoration.

14 IBM XIV Storage System: Copy Services and Migration

 Figure 1-19 illustrates selecting the Restore action on the ITSO_Volume volume.

Figure 1-19 Snapshot volume restore

After you perform the restore action, you return to the Volumes and snapshots panel. The
process is instantaneous, and none of the properties (creation date, deletion priority, modified
properties, or locked properties) of the snapshot or the volume have changed.

Specifically, the process modifies the pointers to the master volume so that they are
equivalent to the snapshot pointer. This change only occurs for partitions that have been
modified. On modification, the XIV Storage System stores the data in a new partition and
modifies the master volume’s pointer to the new partition. The snapshot pointer does not
change and remains pointing at the original data. The restoration process restores the pointer
back to the original data and frees the modified partition space.

If a snapshot is taken and later on the original volume increases in size, you can still do a
restore operation. The snapshot still has the original volume size and when you restore it to
the original volume this volume also will have the original size again.

The XCLI Session (or XCLI command) provides more options for restoration than the GUI.
With the XCLI, you can restore a snapshot to a parent snapshot (Example 1-4).

Example 1-4 Restoring a snapshot to another snapshot

snapshot_restore snapshot=ITSO_Volume.snapshot_00002
target_snapshot=ITSO_Volume.snapshot_00001

1.2.5 Overwriting snapshots

For your regular backup jobs you can decide whether you always want to create new
snapshots (and let the system delete the old ones) or whether you prefer to overwrite the
existing snapshots with the latest changes to the data. For instance, a backup application
requires the latest copy of the data to perform its backup operation. This overwrite operation
modifies the pointers to the snapshot data to be reset to the master volume. Therefore, all

Chapter 1. Snapshots 15
 pointers to the original data are lost, and the snapshot appears as new. Storage that was
allocated for the data changes between the volume and its snapshot is released.

From either the Volumes and Snapshots view or the Snapshots Tree view, right-click the
snapshot to overwrite. Select Overwrite from the menu and a dialog box opens. Click OK to
validate the overwriting of the snapshot. Figure 1-20 illustrates overwriting the snapshot
named ITSO_Volume.snapshot_00001.

Figure 1-20 Overwriting a snapshot

It is important to note that the overwrite process modifies the snapshot properties and
pointers when involving duplicates. Figure 1-21 shows two changes to the properties. The
snapshot named ITSO_Volume.snapshot_00001 has a new creation date. The duplicate
snapshot still has the original creation date. However, it no longer points to the original
snapshot. Instead, it points to the master volume according to the snapshot tree, which
prevents a restoration of the duplicate to the original snapshot. If the overwrite occurs on the
duplicate snapshot, the duplicate creation date is changed, and the duplicate is now pointing
to the master volume.

Figure 1-21 Snapshot tree after the overwrite process has occurred

The XCLI performs the overwrite operation through the snapshot_create command. There is
an optional parameter in the command to specify which snapshot to overwrite. If the optional
parameter is not used, a new snapshot volume is created.

Example 1-5 Overwriting a snapshot

snapshot_create vol=ITSO_Volume overwrite=ITSO_Volume.snapshot_00001

1.2.6 Unlocking a snapshot

At certain times, it might be beneficial to modify the data in a snapshot. This feature is useful
for performing tests on a set of data or performing other types of data-mining activities.

16 IBM XIV Storage System: Copy Services and Migration

There are two scenarios that you must investigate when unlocking snapshots. The first
scenario is to unlock a duplicate. By unlocking the duplicate, none of the snapshot properties
are modified, and the structure remains the same. This method is straightforward and
provides a backup of the master volume along with a working copy for modification. To unlock
the snapshot, simply right-click the snapshot and select Unlock, as shown in Figure 1-22.

Figure 1-22 Unlocking a snapshot

The results in the Snapshots Tree window show that the locked property is off and the
modified property is on for ITSO_Volume.snapshot_00002. Even if the volume is relocked or
overwritten with the original master volume, the modified property remains on. Also note that
in Figure 1-23 the structure is unchanged. If an error occurs in the modified duplicate
snapshot, the duplicate snapshot can be deleted, and the original snapshot duplicated a
second time to restore the information.

Figure 1-23 Unlocked duplicate snapshot

Chapter 1. Snapshots 17
 For the second scenario, the original snapshot is unlocked and not the duplicate. Figure 1-24
shows the new property settings for ITSO_Volume.snapshot.00001. At this point, the duplicate
snapshot mirrors the unlocked snapshot, because both snapshots still point to the original
data. While the unlocked snapshot is modified, the duplicate snapshot references the original
data. If the unlocked snapshot is deleted, the duplicate snapshot remains, and its parent
becomes the master volume.

Figure 1-24 Unlocked original snapshot

Because the hierarchal snapshot structure was unmodified, the duplicate snapshot can be
overwritten by the original snapshot. The duplicate snapshot can be restored to the master
volume. Based on the results, this process is no different from the first scenario. There is still
a backup and a working copy of the data.

Unlocking a snapshot is the same as unlocking a volume (Example 1-6).

Example 1-6 Unlocking a snapshot with the XCLI Session commands

vol_unlock vol=ITSO_Volume.snapshot_00001

1.2.7 Locking a snapshot

If the changes made to a snapshot must be preserved, you can lock an unlocked snapshot.
Figure 1-25 shows locking the snapshot named ITSO_Redbooks.snapshot.00001. From the
Volumes and snapshots panel, right-click the snapshot to lock and select Lock. The snapshot
is locked immediately.

Figure 1-25 Locking a snapshot

18 IBM XIV Storage System: Copy Services and Migration

 The locking process completes immediately, preventing further modification to the snapshot.
In Figure 1-26, the ITSO_Volume.00001 snapshot shows that both the lock property is on and
the modified property is on.

Even though there has not been a change to the snapshot, the system does not remove the
modified property.

Figure 1-26 Validating that the snapshot is locked

The XCLI lock command (vol_lock), which is shown in Example 1-7, is almost a mirror
operation of the unlock command. Only the actual command changes, but the same
operating parameters are used when issuing the command.

Example 1-7 Locking a snapshot

vol_lock vol=ITSO_Redbooks.snapshot_00001

1.2.8 Deleting a snapshot

When a snapshot is no longer needed, delete it. Figure 1-27 illustrates how to delete a
snapshot. In this case, the modified snapshot redbook_markus_01.snapshot.00001 is no
longer needed. To delete the snapshot, right-click it and select Delete from the menu. A
dialog box appears requesting that you validate the operation.

Figure 1-27 Deleting a snapshot

Chapter 1. Snapshots 19
 Figure 1-28 no longer displays the snapshot ITSO_volume.snapshot.00001. Note that the
volume and the duplicate snapshot are unaffected by the removal of this snapshot. In fact, the
duplicate becomes the child of the master volume. The XIV Storage System provides the
ability to restore the duplicate snapshot to the master volume or to overwrite the duplicate
snapshot from the master volume even after deleting the original snapshot.

Figure 1-28 Validating the snapshot is removed

The delete snapshot command (snapshot_delete) operates the same as the creation
snapshot. Refer to Example 1-8.

Example 1-8 Deleting a snapshot

snapshot_delete snapshot=ITSO_Volume.snapshot_00001

Important: If you delete a volume all snapshots associated with the volume are also
deleted.

20 IBM XIV Storage System: Copy Services and Migration

1.2.9 Automatic deletion of a snapshot
The XIV Storage System has a feature in place to protect a storage pool from becoming full. If
the space allocated for snapshots becomes full, the XIV Storage System automatically
deletes a snapshot. Figure 1-29 shows a storage pool with a single 17 GB volume labeled
XIV_ORIG_VOL. The host connected to this volume is sequentially writing to a file that is stored
on this volume. While the data is written, a snapshot called XIV_ORIG_VOL.snapshot.00006 is
created, and one minute later, a second snapshot is taken (not a duplicate), which is called
XIV_ORIG_VOL.snapshot.00007.

Figure 1-29 Snapshot before the automatic deletion

With this scenario, a duplicate does not cause the automatic deletion to occur. Because a
duplicate is a mirror copy of the original snapshot, the duplicate does not create the additional
allocations in the storage pool.

Approximately one minute later, the oldest snapshot (XIV_ORIG_VOL.snapshot_00006) is

removed from the display. The storage pool is 51 GB in size, with a snapshot size of 34 GB,
which is enough for one snapshot. If the master volume is unmodified, many snapshots can
exist within the pool, and the automatic deletion does not occur. If there were two snapshots
and two volumes, it might take longer to cause the deletion, because the volumes utilize
different portions of the disks, and the snapshots might not have immediately overlapped.

Chapter 1. Snapshots 21
 To examine the details of the scenario at the point where the second snapshot is taken, a
partition is in the process of being modified. The first snapshot caused a redirect on write, and
a partition was allocated from the snapshot area in the storage pool. Because the second
snapshot occurs at a different time, this action generates a second partition allocation in the
storage pool space. This second allocation does not have available space, and the oldest
snapshot is deleted. Figure 1-30 shows that the master volume XIV_ORIG_VOL and the newest
snapshot XIV_ORIG_VOL.snapshot.00007 are present. The oldest snapshot
XIV_ORIG_VOL.snapshot.00006 was removed.

Figure 1-30 Snapshot after automatic deletion

22 IBM XIV Storage System: Copy Services and Migration

 To determine the cause of removal, you must go to the Events panel under the System menu.
As shown on Figure 1-31, the event “SNAPSHOT_DELETED_DUE_TO_POOL_EXHAUSTION” is logged.
The snapshot name XIV_ORIG_VOL.snapshot.00006 and time 2008-07-31 15:17:31 are also
logged for future reference.

Figure 1-31 Record of automatic deletion

1.3 Snapshots consistency group

The purpose of a consistency group is to pool multiple volumes together so that a snapshot
can be taken of all the volumes at the same moment in time. This action creates a
synchronized snapshot of all the volumes and is ideal for applications that span multiple
volumes, for example, a database application that has the logs on one volume and the
database on another volume. When creating a backup of the database, it is important to
synchronize the data so that it is consistent. If the data is inconsistent, a database restore is
not possible, because the log and the data are different and therefore part of the data can be
lost.

1.3.1 Creating a consistency group

There are two methods of creating a consistency group. The first method is to create the
consistency group and add the volumes in one step. The second method creates the
consistency group and then adds the volumes in a subsequent step. If you also use
consistency groups to manage Remote Mirroring, you must first create an empty consistency
group, mirror it, and later add mirrored volumes to the consistency group.

Chapter 1. Snapshots 23
 Restriction: Volumes in a consistency group must be in the same storage pool. A
consistency group cannot have volumes from different pools.

Starting at the Volumes menu, select the volume that is to be added to the consistency group.
To select multiple volumes, hold down the Ctrl key and click each volume. After the volumes
are selected, right-click a selected volume to bring up the Operations menu. From there, click
Create consistency group with these Volumes. Refer to Figure 1-32 for an example of this
operation.

Figure 1-32 Creating a consistency group with these Volumes

After selecting the Create option from the menu, a dialog window appears. Enter the name of
the consistency group. Because the volumes are added during creation, it is not possible to
change the pool name. Figure 1-33 shows the process of creating a consistency group. After
the name is entered, click Create.

Figure 1-33 Naming the consistency group

Viewing the volumes displays the owning consistency group. As in Figure 1-34, the two
volumes contained in the xiv_volume_copy pool are now owned by the xiv_db_cg consistency
group. The volumes are displayed in alphabetical order and do not reflect a preference or
internal ordering.

Figure 1-34 Viewing the volumes after creating a consistency group

24 IBM XIV Storage System: Copy Services and Migration

In order to obtain details about the consistency group, the GUI provides a panel to view the
information. Under the Volumes menu, select Consistency Groups. Figure 1-35 illustrates
how to access this panel.

Figure 1-35 Accessing the consistency group view

This selection sorts the information by consistency group. The panel allows you to expand the
consistency group and see all the volumes owned by that consistency group. In Figure 1-36,
there are two volumes owned or contained by the xiv_db_cg consistency group. In this
example, a snapshot of the volumes has not been created.

Figure 1-36 Consistency Group view

From the consistency group view, you can create a consistency group without adding
volumes. On the menu bar at the top of the window, there is an icon to add a new consistency
group. By clicking the Add consistency group icon shown in Figure 1-37, a creation dialog box
appears, as shown in Figure 1-33 on page 24. Then provide a name and the storage pool for
the consistency group.

Figure 1-37 Adding a new consistency group

Chapter 1. Snapshots 25
 When created, the consistency group appears in the Consistency Groups view of the GUI
(Figure 1-38). The new group does not have any volumes associated with it. A new
consistency group named xiv_db_cg is created. The consistency group cannot be expanded
yet, because there are no volumes contained in the consistency group xiv_db_cg.

Figure 1-38 Validating new consistency group

Using the Volumes view in the GUI, select the volumes to add to the consistency group. You
can select multiple volumes by holding Ctrl down and clicking the desired volumes. After
selecting the desired volumes, right-click the volumes and select Add to Consistency
Group. Figure 1-39 shows two volumes being added to a consistency group:
򐂰 xiv_vmware_1
򐂰 xiv_vmware_2

Figure 1-39 Adding volumes to a consistency group

26 IBM XIV Storage System: Copy Services and Migration

 After selecting the volumes to add, a dialog box opens asking for the consistency group to
which to add the volumes. Figure 1-40 adds the volumes to the xiv_db_cg group. Clicking OK
completes the operation.

Figure 1-40 Selecting a consistency group for adding volumes

Using the XCLI Session (or XCLI command), the process must be done in two steps. First,
create the consistency group, then the volumes are added. Example 1-9 provides an example
of setting up a consistency group and adding volumes using the XCLI.

Example 1-9 Creating consistency groups and adding volumes with the XCLI
cg_create cg=xiv_new_cg pool=ITSO_Volume_CG
cg_add_vol cg=xiv_new_cg vol=ITSO_Volume_01
cg_add_vol cg=xiv_new_cg vol=ITSO_Volume_02

1.3.2 Creating a snapshot using consistency groups

When the consistency group is created and the volumes added, the snapshot can be created.
From the consistency group view on the GUI, select the consistency group to copy. As in
Figure 1-41, right-click the group and select Create Snapshots Group from the menu. The
system immediately creates the snapshot.

Figure 1-41 Creating a snapshot using consistency groups

Chapter 1. Snapshots 27
 The new snapshots are created and displayed beneath the volumes in the consistency group
view (Figure 1-42). These snapshots have the same creation date and time. Each snapshot is
locked on creation and has the same defaults as a regular snapshot. The snapshots are
contained in a group structure (called a snapshot group) that allows all the snapshots to be
managed by a single operation.

Figure 1-42 Validating the new snapshots in the consistency group

Adding volumes to a consistency group does not prevent you from creating a single volume
snapshot. If a single volume snapshot is created, it is not displayed in the consistency group
view. The single volume snapshot is also not consistent across multiple volumes. However,
the single volume snapshot does work according to all the rules defined previously in 1.2,
“Snapshots” on page 8.

With the XCLI, when the consistency group is set up, it is simple to create the snapshot. One
command creates all the snapshots within the group at the same moment in time.

Example 1-10 Creating a snapshot group

cg_Snapshots_create cg=xiv_new_cg

1.3.3 Managing a consistency group

After the snapshots are created within a consistency group, you have several options
available. The same management options for a snapshot are available to a consistency
group. Specifically, the deletion priority is modifiable, and the snapshot or group can be
unlocked and locked, and the group can be restored or overwritten. Refer to 1.2, “Snapshots”
on page 8, for specific details about performing these operations.

28 IBM XIV Storage System: Copy Services and Migration

In addition to the snapshot functions, you can remove a volume from the consistency group.
By right-clicking the volume, a menu opens. Click Remove from Consistency Group and
validate the removal on the dialog window that opens. Figure 1-43 provides an example of
removing the xiv_windows_1 volume from the consistency group.

Figure 1-43 Removing a volume from a consistency group

Removing a volume from a consistency group after a snapshot is performed prevents

restoration of any snapshots in the group. If the volume is added back into the group, the
group can be restored.

Chapter 1. Snapshots 29
 To obtain details about a consistency group, you can select Snapshots Group Tree from the
Volumes menu. Figure 1-44 shows where to find the group view.

Figure 1-44 Selecting the Snapshot Group Tree

30 IBM XIV Storage System: Copy Services and Migration

From the Snapshots Group Tree view, you can see many details. Select the group to view on
the left panel by clicking the group snapshot. The right panes provide more in-depth
information about the creation time, the associated pool, and the size of the snapshots. In
addition, the consistency group view points out the individual snapshots present in the group.
Refer to Figure 1-45 for an example of the data that is contained in a consistency group.

Figure 1-45 Snapshots Group Tree view

To display all the consistency groups in the system, issue the XCLI cg_list command.

Example 1-11 Listing the consistency groups

cg_list

Name Pool Name

Group1 GCHI_THIN_01
EXCH_CLU_CONSGROUP GCHI_THIN_01
snapshot_test snapshot_test
Tie xiv_pool
mirror_cg redbooks_mirror
xiv_db_cg xiv_volume_copy
MySQL Group redbooks_markus
xiv_new_cg redbooks_markus

More details are available by viewing all the consistency groups within the system that have
snapshots. The groups can be unlocked or locked, restored, or overwritten. All the operations
discussed in the snapshot section are available with the snap_group operations.

Chapter 1. Snapshots 31
 Example 1-12 illustrates the snap_group_list command.

Example 1-12 Listing all the consistency groups with snapshots

snap_group_list

Name CG Snapshot Time Deletion Priority

xiv_db_cg.snap_group_00001 xiv_db_cg 2008-08-07 18:59:06 1
MySQL Group.snap_group_00001 MySQL Group 2008-08-08 18:16:53 1
xiv_new_cg.snap_group_00001 xiv_new_cg 2008-08-08 20:39:57 1

1.3.4 Deleting a consistency group

Before a consistency group can be deleted, the associated volumes must be removed from
the consistency group. On deletion of a consistency group, the snapshots become
independent snapshots and remain tied to their volume. To delete the consistency group,
right-click the group and select Delete. Validate the operation by clicking OK. Figure 1-46
provides an example of deleting the consistency group called xiv_db_cg.

Figure 1-46 Deleting a consistency group

In order to delete a consistency group with the XCLI, you must first remove all the volumes
one at a time. As in Example 1-13, each volume in the consistency group is removed first.
Then the consistency group is available for deletion. Deletion of the consistency group does
not delete the individual snapshots. They are tied to the volumes and were removed from the
consistency group when you removed the volumes.

Example 1-13 Deleting a consistency group

cg_remove_vol vol=ITSO_Redbooks_03
cg_remove_vol vol=ITSO_Redbooks_04
cg_delete cg=xiv_new_cg

32 IBM XIV Storage System: Copy Services and Migration

1.4 Snapshot with Remote Mirror
XIV has a special snapshot (shown in Figure 1-47) that is automatically created by the
system. During the recovery phase of a Remote Mirror, the system creates a snapshot on the
target to ensure a consistent copy.

Important: This snapshot has a special deletion priority and is not deleted automatically if
the snapshot space becomes fully utilized.

When the synchronization is complete, the snapshot is removed by the system because it is
no longer needed. The following list describes the sequence of events to trigger the creation
of the special snapshot. Note that if a write does not occur while the links are broken, the
system does not create the special snapshot. The events are:
1. Remote Mirror is synchronized.
2. Loss of connectivity to remote system occurs.
3. Writes continue to the primary XIV Storage System.
4. Mirror paths are reestablished (here the snapshot is created) and synchronization starts.

Figure 1-47 Special snapshot during Remote Mirror synchronization operation

For more details about Remote Mirror refer to Chapter 5, “Synchronous Remote Mirroring” on
page 125.

Important: The special snapshot is created regardless of the amount of pool space on the
target pool. If the snapshot causes the pool to be overutilized, the mirror remains inactive.
The pool must be expanded to accommodate the snapshot, then the mirror can be
reestablished.

Chapter 1. Snapshots 33
1.5 MySQL database backup example
MySQL is an open source database application that is used by many Web programs. For
more information go to:
http://www.mysql.com

The database has several important files:

򐂰 The database data
򐂰 The log data
򐂰 The backup data

The MySQL database stores the data in a set directory and cannot be separated. The backup
data, when captured, can be moved to a separate system. The following scenario shows an
incremental backup of a database and then uses snapshots to restore the database to verify
that the database is valid.

The first step is to back up the database. For simplicity, a script is created to perform the
backup and take the snapshot. Two volumes are assigned to a Linux® host (Figure 1-48). The
first volume contains the database and the second volume holds the incremental backups in
case of a failure.

Figure 1-48 XIV view of the volumes

On the Linux host, the two volumes are mapped onto separate file systems. The first file
system xiv_pfe_1 maps to volume redbook_markus_09, and the second file system xiv_pfe_2
maps to volume redbook_markus_10. These volumes belong to the consistency group MySQL
Group so that when the snapshot is taken, snapshots of both volumes are taken at the same
moment.

To perform the backup you must configure the following items:

򐂰 The XIV XCLI must be installed on the server. This way, the backup script can invoke the
snapshot instead of relying on human intervention.
򐂰 Secondly, the database must have the incremental backups enabled. To enable the
incremental backup feature, MySQL must be started with the --log-bin feature
(Example 1-14). This feature enables the binary logging and allows database restorations.

Example 1-14 Starting MySQL

./bin/mysqld_safe --no-defaults --log-bin=backup

The database is installed on /xiv_pfe_1. However, a pointer in /usr/local is made, which

allows all the default settings to coexist, and yet the database is stored on the XIV volume. To
create the pointer, use the command in Example 1-15. Note that the source directory must be
changed for your particular installation. You can also install the MySQL application on a local
disk and change the default data directory to be on the XIV volume.

Example 1-15 MySQL setup

cd /usr/local
ln -s /xiv_pfe_1/mysql-5.0.51a-linux-i686-glibc23 mysql

34 IBM XIV Storage System: Copy Services and Migration

The backup script is simple, and depending on the implementation of your database, the
following script might be too simple. However, the following script (Example 1-16) does force
an incremental backup and copies the data to the second XIV volume. Then the script locks
the tables so that no more data can be modified. When the tables are locked, the script
initiates a snapshot, which saves everything for later use. Finally, the tables are unlocked.

Example 1-16 Script to perform backup

# Report the time of backing up
date

# First flush the tables this can be done while running and
# creates an incremental backup of the DB at a set point in time.
/usr/local/mysql/bin/mysql -h localhost -u root -p password < ~/SQL_BACKUP

# Since the mysql daemon was run specifying the binary log name
# of backup the files can be copied to the backup directory on another disk
cp /usr/local/mysql/data/backup* /xiv_pfe_2

# Secondly lock the tables so a Snapshot can be performed.

/usr/local/mysql/bin/mysql -h localhost -u root -p password < ~/SQL_LOCK

# XCLI command to perform the backup

# ****** NOTE User ID and Password are set in the user profile *****
/root/XIVGUI/xcli -c xiv_pfe cg_Snapshots_create cg="MySQL Group"

# Unlock the tables so that the database can continue in operation.

/usr/local/mysql/bin/mysql -h localhost -u root -p password < ~/SQL_UNLOCK

When issuing commands to the MySQL database, the password for the root user is stored in
an environment variable (not in the script, as was done in Example 1-16 for simplicity).
Storing the password in an environment variable allows the script to perform the action
without requiring user intervention. For the script to invoke the MySQL database, the SQL
statements are stored in separate files and piped into the MySQL application. Example 1-17
provides the three SQL statements that are issued to perform the backup operation.

Example 1-17 SQL commands to perform backup operation

SQL_BACKUP
FLUSH TABLES

SQL_LOCK
FLUSH TABLES WITH READ LOCK

SQL_UNLOCK
UNLOCK TABLES

Chapter 1. Snapshots 35
 Before running the backup script, a test database, which is called redbook, is created. The
database has one table, which is called chapter, which contains the chapter name, author,
and pages. The table has two rows of data that define information about the chapters in the
redbook. Figure 1-49 shows the information in the table before the backup is performed.

Figure 1-49 Data in database before backup

Now that the database is ready, the backup script is run. Example 1-18 is the output from the
script. Then the snapshots are displayed to show that the system now contains a backup of
the data.

Example 1-18 Output from the backup process

[root@x345-tic-30 ~]# ./mysql_backup

Mon Aug 11 09:12:21 CEST 2008

Command executed successfully.

[root@x345-tic-30 ~]# /root/XIVGUI/xcli -c xiv_pfe snap_group_list cg="MySQLGroup"

Name CG Snapshot Time Deletion Priority
MySQL Group.snap_group_00006 MySQL Group 2008-08-11 15:14:24 1

[root@x345-tic-30 ~]# /root/XIVGUI/xcli -c xiv_pfe time_list

Time Date Time Zone Daylight Saving Time
15:17:04 2008-08-11 Europe/Berlin yes
[root@x345-tic-30 ~]#

36 IBM XIV Storage System: Copy Services and Migration

To show that the restore operation is working, the database is dropped (Figure 1-50) and all
the data is lost. After the drop operation is complete, the database is permanently removed
from MySQL. It is possible to perform a restore action from the incremental backup. For this
example, the snapshot function is used to restore the entire database.

Figure 1-50 Dropping the database

The restore script, shown in Example 1-19, stops the MySQL daemon and unmounts the
Linux file systems. Then the script restores the snapshot and finally remounts and starts
MySQL.

Example 1-19 Restore script

[root@x345-tic-30 ~]# cat mysql_restore
# This resotration just overwrites all in the database and puts the
# data back to when the snapshot was taken. It is also possible to do
# a restore based on the incremental data; this script does not handle
# that condition.

# Report the time of backing up

date

# First shutdown mysql

mysqladmin -u root -p password shutdown

# Unmount the filesystems

umount /xiv_pfe_1
umount /xiv_pfe_2

#List all the snap groups

/root/XIVGUI/xcli -c xiv_pfe snap_group_list cg="MySQL Group"

#Prompt for the group to restore

echo "Enter Snapshot group to restore: "
read -e snap_group

Chapter 1. Snapshots 37
 # XCLI command to perform the backup
# ****** NOTE User ID and Password are set in the user profile *****
/root/XIVGUI/xcli -c xiv_pfe snap_group_restore snap_group="$snap_group"

# Mount the FS
mount /dev/dm-2 /xiv_pfe_1
mount /dev/dm-3 /xiv_pfe_2

# Start the MySQL server

cd /usr/local/mysql
./configure

Example 1-20 shows the output from the restore action.

Example 1-20 Output from the restore script

[root@x345-tic-30 ~]# ./mysql_restore
Mon Aug 11 09:27:31 CEST 2008
STOPPING server from pid file
/usr/local/mysql/data/x345-tic-30.mainz.de.ibm.com.pid
080811 09:27:33 mysqld ended

Name CG Snapshot Time Deletion Priority

MySQL Group.snap_group_00006 MySQL Group 2008-08-11 15:14:24 1
Enter Snapshot group to restore:
MySQL Group.snap_group_00006
Command executed successfully.
NOTE: This is a MySQL binary distribution. It's ready to run, you don't
need to configure it!

To help you a bit, I am now going to create the needed MySQL databases
and start the MySQL server for you. If you run into any trouble, please
consult the MySQL manual, that you can find in the Docs directory.

Installing MySQL system tables...

OK
Filling help tables...
OK

To start mysqld at boot time you have to copy

support-files/mysql.server to the right place for your system

PLEASE REMEMBER TO SET A PASSWORD FOR THE MySQL root USER !

To do so, start the server, then issue the following commands:
./bin/mysqladmin -u root password 'new-password'
./bin/mysqladmin -u root -h x345-tic-30.mainz.de.ibm.com password 'new-password'

Alternatively you can run:

./bin/mysql_secure_installation

which also gives the option of removing the test

databases and anonymous user created by default. This is
strongly recommended for production servers.

See the manual for more instructions.

38 IBM XIV Storage System: Copy Services and Migration

You can start the MySQL daemon with:
cd . ; ./bin/mysqld_safe &

You can test the MySQL daemon with mysql-test-run.pl

cd mysql-test ; perl mysql-test-run.pl

Please report any problems with the ./bin/mysqlbug script!

The latest information about MySQL is available on the Web at

http://www.mysql.com
Support MySQL by buying support/licenses at http://shop.mysql.com
Starting the mysqld server. You can test that it is up and running
with the command:
./bin/mysqladmin version
[root@x345-tic-30 ~]# Starting mysqld daemon with databases from
/usr/local/mysql/data

When complete, the data is restored and the redbook database is available, as shown in
Figure 1-51.

Figure 1-51 Database after restore operation

Chapter 1. Snapshots 39
40 IBM XIV Storage System: Copy Services and Migration
 2

Chapter 2. Tivoli Storage FlashCopy

Manager and Volume Shadow
Copy Services
This chapter explains how the XIV Snapshot function can be combined with the Microsoft®
Volume Shadow Copy Services (VVS) and IBM Tivoli Storage FlashCopy Manager to provide
efficient and reliable application or database backup and recovery solutions.

After a brief overview of the Microsoft VSS architecture and an introduction to IBM Tivoli
Storage FlashCopy Manager, we cover the requirements, configuration, and implementation
of the XIV VSS Provider with the Tivoli Storage FlashCopy Manager for backing up Microsoft
Exchange Server data.

© Copyright IBM Corp. 2010. All rights reserved. 41

2.1 Overview
This section includes an introduction to the IBM Tivoli Storage FlashCopy Manager, which
provides application-integrated snapshot backup and restore support. The product provides
the tools and information needed to create and manage volume-level snapshots of Microsoft
SQL Server and Microsoft Exchange server data.

Tivoli Storage FlashCopy Manager uses Microsoft Volume Shadow Copy Services in a
Windows® environment. VSS relies on a VSS hardware provider.

We explain in subsequent sections the installation of the XIV VSS Provider and provide
detailed installation and configuration information for the IBM Tivoli Storage FlashCopy
Manager. We have also included usage scenarios.

2.2 Tivoli Storage FlashCopy Manager

Tivoli Storage FlashCopy Manager V2.1 is a package that is easy to install, configure, and
deploy, and integrates in a seamless manner with the IBM System Storage DS3000,
DS4000®, DS5000, DS8000, IBM SAN Volume Controller, and IBM XIV Storage System
products.

IBM Tivoli Storage FlashCopy Manager V2.1 is a standalone product designed to perform
near-instant application-aware snapshot backups, with minimal performance impact, for IBM
DB2®, Oracle, SAP, Microsoft SQL Server, and Microsoft Exchange. It improves application
availability and service levels through high-performance, near-instant restore capabilities that
help reduce downtime.

In addition, it can satisfy advanced data protection and data reduction needs through optional
integration with IBM Tivoli Storage Manager V6.

42 IBM XIV Storage System: Copy Services and Migration

Tivoli Storage FlashCopy Manager (Figure 2-1) is the successor to the former Tivoli Storage
Manager (TSM) for Advanced Copy Services and TSM for Copy Services products.

Application FlashCopy Manager

System Local
Online,
Online,near
nearinstant
instant
Application Snapshot snapshot
snapshotbackups
backups
Data Versions
with
withminimal
minimal
Snapshot
Backup performance
performanceimpact
impact
Sn
a
R e ps h
st o o t
re High
Highperformance,
performance,
• Oracle
• DB2
near
nearinstant
instantrestore
restore
• SAP capability
capability
• SQL Server
• Microsoft For IBM Storage Storage Manager 6

Exchange Integrated
Integratedwith
withIBM
IBM
With
Server Storage
Storage Hardware
Hardware
SVC Optional
XIV TSM
DS8000 Simplified
Simplifieddeployment
deployment
DS 3/4/5* Backup
Integration
*VSS Integration
1
Figure 2-1 Tivoli Storage FlashCopy Manager overview

Tivoli Storage FlashCopy Manager V2.1 has the following key features:
򐂰 Step-by-step product provision using Microsoft Management Console (MMC) for Microsoft
Exchange Server and Microsoft SQL server
򐂰 Performs application snapshots without using the TSM Server
򐂰 Easy to integrate with TSM environment
򐂰 Diagnostic snapshot tool
򐂰 Provider complete reporting tool
򐂰 Manages and schedules application snapshots
򐂰 Maintains application snapshot history
򐂰 Configuration verification tool
򐂰 Incremental and differential support for Exchange VSS Backups
򐂰 Oracle ASM support
򐂰 SAP Control File Backup

For more detailed information refer to the IBM Tivoli Storage Web site:
http://www.ibm.com/software/tivoli

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 43
 Figure 2-2 shows the Tivoli Storage FlashCopy Manager Management Console.

Figure 2-2 Tivoli Storage FlashCopy Manager: Management Console

2.3 Windows Server 2008 Volume Shadow Copy Service

Microsoft first introduced Volume Shadow Copy Services in Windows 2003 Server and all of
its server line and subsequent releases after that. VSS provides a framework and the
mechanisms to create consistent point-in-time copies (known as shadow copies) of
databases and applications data. It consists of a set of Microsoft COM APIs that enable
volume-level snapshots to be performed while the applications that contain data on those
volumes remain online and continue to write.

Without VSS, if you do not have an online backup solution implemented, you either must stop
or quiesce applications during the backup process, or live with the side effects of an online
backup with inconsistent data and open files that could not be backed up.

With VSS, you can produce consistent shadow copies by coordinating tasks with business
applications, file system services, backup applications, fast recovery solutions, and storage
hardware such as the XIV Storage System.

44 IBM XIV Storage System: Copy Services and Migration

2.3.1 VSS architecture and components
Figure 2-3 shows the VSS architecture and how the VSS service interacts with the other
components to create a shadow copy of a volume, or, when it pertains to XIV, a volume
snapshot.

Figure 2-3 VSS components

The components of the VSS architecture are:

򐂰 VSS Service
The VSS Service is at the core of the VSS architecture. It is the Microsoft Windows
service that directs all of the other VSS components that are required to create the
volumes shadow copies (snapshots). This Windows service is the overall coordinator for
all VSS operations.
򐂰 Requestor
This is the software application that commands that a shadow copy be created for
specified volumes. The VSS requestor is provided by Tivoli Storage FlashCopy Manager
and is installed with the Tivoli Storage FlashCopy Manager software.
򐂰 Writer
This is a component of a software application that places the persistent information for the
shadow copy on the specified volumes. A database application (such as SQL Server or
Exchange Server) or a system service (such as Active Directory) can be a writer.
Writers serve two main purposes by:
– Responding to signals provided by VSS to interface with applications to prepare for
shadow copy
– Providing information about the application name, icons, files, and a strategy to restore
the files.
Writers prevent data inconsistencies.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 45
 For exchange data, the Microsoft Exchange Server contains the writer components and
requires no configuration.
For SQL data, Microsoft SQL Server contains the writer components (SqlServerWriter). It
is installed with the SQL Server software and requires the following minor configuration
tasks:
– Set the SqlServerWriter service to automatic. This enables the service to start
automatically when the machine is rebooted.
– Start the SqlServerWriter service.
򐂰 Provider
This is the application that produces the shadow copy and also manages its availability. It
can be a system provider (such as the one included with the Microsoft Windows operating
system), a software provider, or a hardware provider (such as the one available with the
XIV storage system).
For XIV, you must install and configure the IBM XIV VSS Provider.

VSS uses the following terminology to characterize the nature of volumes participating in a
shadow copy operation:
򐂰 Persistent
This is a shadow copy that remains after the backup application completes its operations.
This type of shadow copy also survives system reboots.
򐂰 Non-persistent
This is a temporary shadow copy that remains only as long as the backup application
needs it in order to copy the data to its backup repository.
򐂰 Transportable
This is a shadow copy volume that is accessible from a secondary host so that the backup
can be off-loaded. Transportable is a feature of hardware snapshot providers. On an XIV
you can mount a snapshot volume to another host.
򐂰 Source volume
This is the volume that contains the data to be shadow copied. These volumes contain the
application data.
򐂰 Target or snapshot volume
This is the volume that retains the shadow-copied storage files. It is an exact copy of the
source volume at the time of backup.

VSS supports the following shadow copy methods:

򐂰 Clone (full copy/split mirror)
A clone is a shadow copy volume that is a full copy of the original data as it resides on a
volume. The source volume continues to take application changes while the shadow copy
volume remains an exact read-only copy of the original data at the point-in-time that it was
created.
򐂰 Copy-on-write (differential copy)
A copy-on-write shadow copy volume is a differential copy (rather than a full copy) of the
original data as it resides on a volume. This method makes a copy of the original data
before it is overwritten with new changes. Using the modified blocks and the unchanged
blocks in the original volume, a shadow copy can be logically constructed that represents
the shadow copy at the point-in-time at which it was created.

46 IBM XIV Storage System: Copy Services and Migration

 򐂰 Redirect-on-write (differential copy)
A redirect-on-write shadow copy volume is a differential copy (rather than a full copy) of
the original data as it resides on a volume. This method is similar to copy-on-write, without
the double-write penalty, and it offers storage space and performance efficient snapshots.
New writes to the original volume are redirected to another location set aside for snapshot.
The advantage of redirecting the write is that only one write takes place, whereas with
copy-on-write, two writes occur (one to copy original data onto the storage space, the
other to copy changed data). The XIV storage system supports redirect-on-write.

2.3.2 Microsoft Volume Shadow Copy Service function

Microsoft VSS accomplishes the fast backup process when a backup application (the
requestor, which is Tivoli Storage FlashCopy Manager in our case) initiates a shadow copy
backup. The VSS service coordinates with the VSS-aware writers to briefly hold writes on
databases, applications, or both. VSS flushes the file system buffers and asks a provider
(such as the XIV provider) to initiate a snapshot of the data. When the snapshot is logically
completed, VSS allows writes to resume and notifies the requestor that the backup has
completed successfully. The (backup) volumes are mounted, but hidden and read-only, ready
to be used when a rapid restore is requested. Alternatively, the volumes can be mounted to a
different host and used for application testing or backup to tape.

The Microsoft VSS FlashCopy process is:

1. The requestor notifies VSS to prepare for a shadow copy creation.
2. VSS notifies the application-specific writer to prepare its data for making a shadow copy.
3. The writer prepares the data for that application by completing all open transactions,
flushing cache and buffers, and writing in-memory data to disk.
4. When the application data is ready for shadow copy, the writer notifies VSS, which in turn
relays the message to the requestor to initiate the commit copy phase.
5. VSS temporarily quiesces application I/O write requests for a few seconds and the VSS
hardware provider performs the snapshot on the storage system.
6. Once the storage snapshot has completed, VSS releases the quiesce, and database or
application writes resume.
7. VSS queries the writers to confirm that write I/Os were successfully held during the
Volume Shadow Copy.

2.4 XIV VSS provider

A VSS hardware provider, such as the XIV VSS Provider, is used by third-party software to
act as an interface between the hardware (storage system) and the operating system. The
third-party application (which can be IBM Tivoli Storage FlashCopy Manager) uses XIV VSS
Provider to instruct the XIV storage system to perform a snapshot of a volume attached to the
host system.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 47
2.4.1 XIV VSS Provider installation
This section illustrates the installation of the XIV VSS Provider. First, make sure that your
Windows system meets the minimum requirements listed below:
򐂰 Supported operating systems: Windows 2003 and later
򐂰 Disk space: 3 MB
򐂰 Host attachment: 1.0.4 or later
򐂰 .NET Framework: 2.0 with SP1 or later

At the time of writing, the XIV VSS Provider 2.0.9 version was available. We used a Windows
2008 64bit host system for our tests.

The XIV VSS Hardware Provider 2.0.9 version can be downloaded at:
http://www.ibm.com/systems/storage/disk/xiv/index.html

The installation of the XIV VSS Provider is a straightforward normal Windows application
program installation.

To start, locate the XIV VSS Provider installation file, also known as the xProv installation file.
If the XIV VSS Provider 2.0.9 is downloaded from the Internet, the file name is
xProvSetup-x64-2.0.9.exe. Execute the file to start the installation.

A Welcome window opens (Figure 2-4). Click Next.

Tip: Uninstall any previous versions of the XIV VSS xProv driver if installed. An upgrade is
not allowed with the 2.0.9 release of XIV VSS provider.

Figure 2-4 XIV VSS provider installation: Welcome window

The License Agreement window is displayed and to continue the installation you must accept
the license agreement.

48 IBM XIV Storage System: Copy Services and Migration

 In the next step you can specify the XIV VSS Provider configuration file directory and the
installation directory. Keep the default directory folder and installation folder or change it to
meet your needs.

The next dialog window is for post-installation operations, as shown in Figure 2-5. Perform a
post-installation configuration during the installation process. The configuration can, however,
be performed at later time. When done, click Next.

Figure 2-5 Installation: post-installation operation

A Confirm Installation window is displayed. If required you can go back to make changes or
confirm the installation by clicking Next.

Once the installation is complete click Close to exit.

2.4.2 XIV VSS Provider configuration

The XIV VSS Provider must now be configured.

If the post installation check box was selected during the installation (Figure 2-5), the XIV
VSS Provider configuration window shown in Figure 2-7 on page 50 is now displayed. If the
post-installation check box had not been selected during the installation, it must be manually
invoked by selecting Start  All Programs  XIV and starting the MachinePool Editor, as
shown in Figure 2-6.

Figure 2-6 Configuration: XIV VSS Provider setup

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 49
 Provide specific information regarding the XIV Storage System IP addresses and user ID and
password with admin privileges. Have that information available.
1. In the dialog shown in Figure 2-7, click Add Machine.

Figure 2-7 XIV Configuration: Machine Pool Editor

2. The New Machine dialog shown in Figure 2-8 is displayed. Enter the user name and
password of an XIV user with administrator privileges (storageadmin role) and the primary
IP address of the XIV Storage System. Then click Validate.

Figure 2-8 XIV configuration: add machine

50 IBM XIV Storage System: Copy Services and Migration

3. When the validation is successful, the Validation Success dialog shown in Figure 2-9
opens. Click OK.

Figure 2-9 XIV configuration: validation success

4. You are now returned to the VSS MachinePool Editor window. The VSS Provider collected
additional information about the XIV storage system, as illustrated in Figure 2-10.

Figure 2-10 XIV Configuration: Machine Pool Editor

5. Click SaveAll.

At this point XIV VSS Provider configuration is complete and you can close the Machine Pool
Editor window. If you must add other XIV Storage Systems, repeat steps 1 to 5.

Once the XIV VSS provider has been configured as just explained, ensure that the operating
system can recognize it. For that purpose, launch the vssadmin command from the operating
system command line:
C:\>vssadmin list providers

Make sure that IBM XIV VSS HW Provider appears among the list of installed VSS providers
returned by the vssadmin command.

Tip: The XIV VSS Provider log file is located in C:\Windows\Temp\xProvDotNet.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 51
 The Windows server is now ready to perform snapshot operations on the XIV Storage
System. Refer to you application documentation for completing the VSS setup.

The next section demonstrates how the Tivoli Storage FlashCopy Manager application uses
the XIV VSS Provider to perform a consistent point-in-time snapshot of the Exchange 2007
and SQL 2008 data on Windows 2008 64bit.

2.5 Installing and configuring Tivoli Storage FlashCopy

Manager for Microsoft Exchange
To install Tivoli Storage FlashCopy Manager, insert the product media into the DVD drive and
the installation starts automatically. If this does not occur, or if you are using a copy or
downloaded version of the media, locate and execute the SetpFCM.exe file. During the
installation, accept all default values.

The Tivoli Storage FlashCopy Manager installation and configuration wizards will guide you
through the installation and configuration steps. After you run the setup and configuration
wizards, your computer is ready to take snapshots.

Tivoli Storage FlashCopy Manager provides the following wizards for installation and
configuration tasks:
򐂰 Setup wizard
Use this wizard to install Tivoli Storage FlashCopy Manager on your computer.
򐂰 Local configuration wizard
Use this wizard to configure Tivoli Storage FlashCopy Manager on your computer to
provide locally managed snapshot support. To manually start the configuration wizard,
double-click Local Configuration in the results pane.
򐂰 Tivoli Storage Manager configuration wizard
Use this wizard to configure Tivoli Storage FlashCopy Manager to manage snapshot
backups using a Tivoli Storage Manager server. This wizard is only available when a Tivoli
Storage Manager license is installed.

Once installed, Tivoli Storage FlashCopy Manager must be configured for VSS snapshot
backups. Use the local configuration wizard for that purpose. These tasks include selecting
the applications to protect, verifying requirements, provisioning, and configuring the
components required to support the selected applications.

52 IBM XIV Storage System: Copy Services and Migration

The configuration process for Microsoft Exchange Server is:
1. Start the Local Configuration Wizard from the Tivoli Storage FlashCopy Manager
Management Console, as shown in Figure 2-11.

Figure 2-11 Tivoli FlashCopy Manager: local configuration wizard for Exchange Server

2. A dialog window is displayed, as shown in Figure 2-12. Select the Exchange Server to
configure and click Next.

Figure 2-12 Local configuration wizard: local data protection selection

Note: The Show System Information button shows the basic information about your
host system.

Tip: Select the check box at the bottom if you do not want the local configuration wizard
to start automatically the next time that the Tivoli Storage FlashCopy Manager
Management Console windows starts.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 53
 3. The Requirements Check dialog window opens, as shown in Figure 2-13. At this stage,
the systems checks that all prerequisites are met.

Figure 2-13 Local Configuration for exchange: requirements check

If any requirement is not met, the configuration wizard does not proceed to the next step.
You may have to upgrade components to fulfill the requirements. The requirements check
can be run again by clicking Re-run once fulfilled. When the check completes
successfully, click Next.

54 IBM XIV Storage System: Copy Services and Migration

4. In this configuration step, the Local Configuration wizard performs all necessary
configuration steps, as shown in Figure 2-14. The steps include provisioning and
configuring the VSS Requestor, provisioning and configuring data protection for the
Exchange Server, and configuring services. When done, click Next.

Figure 2-14 Local configuration for exchange: configuration

Note: By default, details are hidden. Details can be seen or hidden by clicking Show
Details or Hide Details.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 55
 5. The completion window shown in Figure 2-15 is displayed. To run a VSS diagnostic check,
ensure that the corresponding check box is selected and click Finish.

Figure 2-15 Local configuration for exchange: completion

6. The VSS Diagnostic dialog window is displayed. The goal of this step is to verify that any
volume that you select is indeed capable of performing an XIV snapshot using VSS. Select
the XIV mapped volumes to test, as shown in Figure 2-16, and click Next.

Figure 2-16 VSS Diagnostic Wizard: Snapshot Volume Selection

56 IBM XIV Storage System: Copy Services and Migration

 Tip: Any previously taken snapshots can be seen by clicking Snapshots. Clicking the
button refreshes the list and shows all of the existing snapshots.

7. The VSS Snapshot Tests window is displayed, showing a status for each of the snapshots.
This dialog also displays the event messages when clicking Show Details, as shown in
Figure 2-17. When done, click Next.

Figure 2-17 VSS Diagnostic Wizard: Snapshot tests

8. A completion window is displayed with the results, as shown in Figure 3-25. When done,
click Finish.

Note: Microsoft SQL Server can be configured the same way as Microsoft Exchange
Server to perform XIV VSS snapshots for Microsoft SQL Server using Tivoli Storage
FlashCopy Manager.

2.6 Backup scenario for Microsoft Exchange Server

Microsoft Exchange Server is a Microsoft server-line product that provides the capability of
messaging and collaboration software. The main features of Exchange Server are e-mail
exchange, contacts, and calendar functions.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 57
 To perform a VSS snapshot backup of Exchange data, we used the following setup:
򐂰 Windows 2008 64bit
򐂰 Exchange 2007 Server
򐂰 XIV Host Attachment Kit 1.0.4
򐂰 XIV VSS Provider 2.0.9
򐂰 Tivoli Storage FlashCopy Manager 2.0

Microsoft Exchange Server XIV VSS Snapshot backup

On the XIV Storage System a single volume has been created and mapped to the host
system, as illustrated in Figure 2-18. On the Windows host system, the volume has been
initialized as a basic disk and assigned the drive letter G. The G drive has been formatted as
NTFS, and we created a single Exchange Server storage group with a couple of mailboxes on
that drive.

Figure 2-18 Mapped volume to the host system

Tivoli Storage FlashCopy Manager was already configured and tested for XIV VSS snapshot,
as shown in 2.5, “Installing and configuring Tivoli Storage FlashCopy Manager for Microsoft
Exchange” on page 52. To review the Tivoli Storage FlashCopy Manager configuration
settings, use the command shown in Example 2-1.

Example 2-1 Tivoli Storage FlashCopy Manager for Mail: query DP configuration
C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc query tdp

IBM FlashCopy Manager for Mail:

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

FlashCopy Manager for Exchange Preferences

----------------------------------------

BACKUPDESTination................... LOCAL
BACKUPMETHod........................ VSS
BUFFers ............................ 3
BUFFERSIze ......................... 1024

58 IBM XIV Storage System: Copy Services and Migration

DATEformat ......................... 1
LANGuage ........................... ENU
LOCALDSMAgentnode................... sunday
LOGFile ............................ tdpexc.log
LOGPrune ........................... 60
MOUNTWait .......................... Yes
NUMberformat ....................... 1
REMOTEDSMAgentnode..................
RETRies............................. 4
TEMPDBRestorepath...................
TEMPLOGRestorepath..................
TIMEformat ......................... 1

As explained earlier, Tivoli Storage FlashCopy Manger does not use (or need) a TSM server
to perform a snapshot backup. You can see this when you execute the query tsm command,
as shown in Example 2-2. The output does not show a TSM service but
FLASHCOPYMANAGER instead for the NetWork Host Name of Server field. Tivoli Storage
FlashCopy Manager creates a virtual server instead of using a TSM Server to perform a VSS
snapshot backup.

Example 2-2 Tivoli FlashCopy Manager for Mail: query TSM

C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc query tsm

IBM FlashCopy Manager for Mail:

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

FlashCopy Manager Server Connection Information

----------------------------------------------------

Nodename ............................... SUNDAY_EXCH

NetWork Host Name of Server ............ FLASHCOPYMANAGER
FCM API Version ........................ Version 6, Release 1, Level 1.0

Server Name ............................ Virtual Server

Server Type ............................ Virtual Platform
Server Version ......................... Version 6, Release 1, Level 1.0
Compression Mode ....................... Client Determined
Domain Name ............................ STANDARD
Active Policy Set ...................... STANDARD
Default Management Class ............... STANDARD

Example 2-3 shows what options have been configured and used for TSM Client Agent to
perform VSS snapshot backups.

Example 2-3 TSM Client Agent: option file

*======================================================================*
* *
* IBM Tivoli Storage Manager for Databases *
* *
* dsm.opt for the Microsoft Windows Backup-Archive Client Agent *
*======================================================================*
Nodename sunday

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 59
 CLUSTERnode NO
PASSWORDAccess Generate

*======================================================================*
* TCP/IP Communication Options *
*======================================================================*
COMMMethod TCPip
TCPSERVERADDRESS FlashCopymanager
TCPPort 1500
TCPWindowsize 63
TCPBuffSize 32

Before we can perform any backup, we must ensure that VSS is properly configured for
Microsoft Exchange Server and that the DSMagent service is running (Example 2-4).

Example 2-4 Tivoli Storage FlashCopy Manger: Query Exchange Server

C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc query exchange

IBM FlashCopy Manager for Mail:

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

Querying Exchange Server to gather storage group information, please wait...

Microsoft Exchange Server Information

-------------------------------------

Server Name: SUNDAY

Domain Name: sunday.local
Exchange Server Version: 8.1.375.1 (Exchange Server 2007)

Storage Groups with Databases and Status

----------------------------------------

First Storage Group

Circular Logging - Disabled
Replica - None
Recovery - False
Mailbox Database Online
User Define Public Folder Online

STG3G_XIVG2_BAS
Circular Logging - Disabled
Replica - None
Recovery - False
2nd MailBox Online
Mail Box1 Online

Volume Shadow Copy Service (VSS) Information

--------------------------------------------

Writer Name : Microsoft Exchange Writer

Local DSMAgent Node : sunday
Remote DSMAgent Node :

60 IBM XIV Storage System: Copy Services and Migration

Writer Status : Online
Selectable Components : 8

Our test Microsoft Exchange Storage Group is on drive G:\ and it is called STG3G_XIVG2_BAS. It
contains two mailboxes:
򐂰 Mail Box1
򐂰 2nd MailBox

Now we can take a full backup of the storage group by executing the backup command, as
shown in Example 2-5.

Example 2-5 Tivoli Storage FlashCopy Manger: full XIV VSS snapshot backup
C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc backup STG3G_XIVG2_BAS full

IBM FlashCopy Manager for Mail:

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

Updating mailbox history on FCM Server...

Mailbox history has been updated successfully.

Querying Exchange Server to gather storage group information, please wait...

Connecting to FCM Server as node 'SUNDAY_EXCH'...

Connecting to Local DSM Agent 'sunday'...
Starting storage group backup...

Beginning VSS backup of 'STG3G_XIVG2_BAS'...

Executing system command: Exchange integrity check for storage group

'STG3G_XIVG2_BAS'

Files Examined/Completed/Failed: [ 4 / 4 / 0 ] Total Bytes: 44276

VSS Backup operation completed with rc = 0

Files Examined : 4
Files Completed : 4
Files Failed : 0
Total Bytes : 44276

Note that we did not specify a disk drive here. Tivoli Storage FlashCopy Manager finds out
which disk drives to copy with snapshot when doing a backup of a Microsoft Exchange
Storage Group. This is the advantage of an application-aware snapshot backup process.

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 61
 To see a list of the available VSS snapshot backups issue a query command, as shown in
Example 2-6.

Example 2-6 Tivoli Storage FlashCopy Manger: query full VSS snapshot backup
C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc query TSM STG3G_XIVG2_BAS full

IBM FlashCopy Manager for Mail:

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

Querying FlashCopy Manager server for a list of database backups, please wait...

Connecting to FCM Server as node 'SUNDAY_EXCH'...

Backup List
-----------

Exchange Server : SUNDAY

Storage Group : STG3G_XIVG2_BAS

Backup Date Size S Fmt Type Loc Object Name/Database Name

------------------- ----------- - ---- ---- --- -------------------------
06/30/2009 22:25:57 101.04MB A VSS full Loc 20090630222557
91.01MB Logs
6,160.00KB Mail Box1
4,112.00KB 2nd MailBox

To show that a restore operation is working, we deleted the 2nd Mailbox mail box, as shown
in Example 2-7.

Example 2-7 Deleting the mailbox and adding a file

G:\MSExchangeSvr2007\Mailbox\STG3G_XIVG2_BAS>dir
Volume in drive G is XIVG2_SJCVTPOOL_BAS
Volume Serial Number is 344C-09F1

06/30/2009 11:05 PM <DIR> .

06/30/2009 11:05 PM <DIR> ..
06/30/2009 11:05 PM 4,210,688 2nd MailBox.edb
:

G:\MSExchangeSvr2007\Mailbox\STG3G_XIVG2_BAS> del “2nd MailBox.edb”

To perform a restore, all the mailboxes must be unmounted first. A restore will be done at the
volume level, called instant restore (IR), then the recovery operation will run, applying all the
logs, and then mount the mail boxes, as shown in Example 2-8.

Example 2-8 Tivoli Storage FlashCopy Manager: VSS Full Instant Restore and recovery.
C:\Program Files\Tivoli\TSM\TDPExchange>tdpexcc Restore STG3G_XIVG2_BAS Full
/RECOVer=APPL
YALLlogs /MOUNTDAtabases=Yes

IBM FlashCopy Manager for Mail:

62 IBM XIV Storage System: Copy Services and Migration

FlashCopy Manager for Microsoft Exchange Server
Version 6, Release 1, Level 1.0
(C) Copyright IBM Corporation 1998, 2009. All rights reserved.

Starting Microsoft Exchange restore...

Beginning VSS restore of 'STG3G_XIVG2_BAS'...

Starting snapshot restore process. This process may take several minutes.

VSS Restore operation completed with rc = 0

Files Examined : 0
Files Completed : 0
Files Failed : 0
Total Bytes : 0

Recovery being run. Please wait. This may take a while...

C:\Program Files\Tivoli\TSM\TDPExchange>

Note: Instant restore is at the volume level. It does not show the total number of files
examined and completed like a normal backup process does.

To verify that the restore operation worked, open the Exchange Management Console and
check that the storage group and all the mailboxes have been mounted. Furthermore, verify
that the 2nd Mailbox.edb file exists.

See the Tivoli Storage FlashCopy Manager: Installation and User’s Guide for Windows,
SC27-2504, or Tivoli Storage FlashCopy Manager for AIX: Installation and User’s Guide,
SC27-2503, for more and detailed information about Tivoli Storage FlashCopy Manager and
its functions.

The latest information about the Tivoli Storage FlashCopy Manager is available on the Web
at:
http://www.ibm.com/software/tivoli

Chapter 2. Tivoli Storage FlashCopy Manager and Volume Shadow Copy Services 63
64 IBM XIV Storage System: Copy Services and Migration
 3

Chapter 3. Volume copy

The XIV Storage System provides the ability to copy a volume into another volume. This
valuable feature, known as volume copy, is best used for duplicating an image of the volume
when the data residency is extremely long and the information diverges after the copy is
complete.

© Copyright IBM Corp. 2010. All rights reserved. 65

3.1 Volume copy architecture
The volume copy feature provides an instantaneous copy of data from one volume to another
volume. By utilizing the same functionality of the snapshot, the system modifies the target
volume to point at the source volume’s data. After the pointers are modified, the host has full
access to the data on the volume.

After the XIV Storage System completes the setup of the pointers to the source data, a
background copy of the data is performed. The data is copied from the source volume to a
new area on the disk, and the pointers of the target volume are then updated to use this new
space. The copy operation is done in such a way as to minimize the impact to the system. If
the host performs an update before the background copy is complete, a redirect on write
occurs, which allows the volume to be readable and writable before the volume copy
completes.

3.2 Performing a volume copy

Performing a volume copy is a simple task. The only requirement is that the target volume
must be created before the copy can occur.

If the sizes of the volumes differ, the size of the target volume is modified to match the source
volume when the copy is initiated. The resize operation does not require user intervention.

66 IBM XIV Storage System: Copy Services and Migration

Figure 3-1 illustrates making a copy of volume redbook_markus_01. The target volume for this
example is redbook_chris_01. By right-clicking the source volume, a menu appears and you
can then select Copy This Volume. This action causes a dialog box to open.

Figure 3-1 Initiating a copy volume process

From the dialog box, select redbook_chris_01 and click OK. The system then asks you to
validate the copy action.

The XIV Storage System instantly performs the update process and displays a completion
message. When the copy process is complete, the volume is available for use.

Chapter 3. Volume copy 67

 Figure 3-2 provides an example of the volume selection.

Figure 3-2 Target volume selection

To create a volume copy with the XCLI, the source and target volumes must be specified in
the command. In addition, the -y parameter must be specified to provide an affirmative
response to the validation questions. See Example 3-1.

Example 3-1 Performing a volume copy

xcli -c MZ_PFE_1 -y vol_copy vol_src=xiv_vmware_1 vol_trg=xiv_vmware_2

3.3 Creating an OS image with volume copy

This section describes another usage of the volume copy feature. In certain cases, you might
want to install another operating system (OS) image. By using volume copy, the installation
can be done immediately. Usage of VMware simplified the need for SAN boot. However, this
example can be applied to any OS installation in which the hardware configuration is similar.

VMware allows the resources of a server to be separated into logical virtual systems, each
containing its own OS and resources. When creating the configuration, it is extremely
important to have the hard disk assigned to the virtual machine to be a mapped raw LUN. If
the hard disk is a VMware File System (VMFS), the volume copy fails because there are
duplicate file systems in VMware. In Figure 3-3, the mapped raw LUN is the XIV volume that
was mapped to the VMware server.

Figure 3-3 Configuration of the virtual machine in VMware

68 IBM XIV Storage System: Copy Services and Migration

To perform the volume copy:
1. Validate the configuration for your host. With VMware, ensure that the hard disk assigned
to the virtual machine is a mapped raw LUN. For a disk directly attached to a server, the
SAN boot must be enabled and the target server must have the XIV volume discovered.
2. Shut down the source server or OS. If the source remains active, there might be data in
memory that is not synchronized to the disk. If this step is skipped, unexpected results can
occur.
3. Perform volume copy from the source volume to the target volume.
4. Power on the new system.

A demonstration of the process is simple using VMware. Starting with the VMware resource
window, power off the virtual machines for both the source and the target. The summary
described in Figure 3-4 shows that both XIV Source VM (1), the source, and XIV Source VM
(2), the target, are powered off.

Figure 3-4 VMware virtual machine summary

Looking at the XIV Storage System before the copy (Figure 3-5), xiv_vmware_1 is mapped to
the XIV Source VM (1) in VMware and has utilized 1 GB of space. This information shows that
the OS is installed and operational. The second volume, xiv_vmware_2, is the target volume
for the copy and is mapped to XIV Source VM (2) and is 0 in size. At this point, the OS has not
been installed on the virtual machine and thus the OS is not usable.

Figure 3-5 The XIV volumes before the copy

Because the virtual machines are powered off, simply initiate the copy process as just
described.

Selecting xiv_vmware_1 as the source, copy the volume to the target xiv_vmware_2. The
copy completes immediately and is available for usage.

To verify that the copy is complete, the used area of the volumes must match, as shown in
Figure 3-6.

Figure 3-6 The XIV volumes after the copy

Chapter 3. Volume copy 69

 After the copy is complete, power up the new virtual machine to use the new operating
system. Both servers usually boot up normally with only minor modifications to the host. In
this example, the server name we had to changed because there were two servers on the
network with the same name. Refer to Figure 3-7.

Figure 3-7 VMware summary showing both virtual machines powered on

Figure 3-8 shows the second virtual machine console with the Windows operating system
powered on.

Figure 3-8 Booted Windows server

70 IBM XIV Storage System: Copy Services and Migration

 4

Chapter 4. Remote Mirroring

The Remote Mirroring function of the XIV Storage System provides a real-time copy between
two or more storage systems supported over Fibre Channel (FC) or iSCSI links. This feature
provides a method to protect data from site failures.

Remote Mirroring can be a synchronous copy solution where write operations are completed
on both copies (local and remote sites) before they are considered to be complete (see
Chapter 5, “Synchronous Remote Mirroring” on page 125). This type of remote mirroring is
normally used for short distances to minimize the effect of I/O delays inherent to the distance
to the remote site.

Remote Mirroring can also be an asynchronous solution were consistent sets of data are
copied to the remote location at specified intervals and host I/O operations are complete after
writing to the primary (see Chapter 6, “Asynchronous Remote Mirroring” on page 149). This
is typically used for long distances between sites.

Note: For asynchronous mirroring over iSCSI links, a reliable, dedicated network must be
available. It requires consistent network bandwidth and a non-shared link.

Unless otherwise noted, this chapter describes the basic concepts, functions, and terms that
are common to both XIV synchronous and asynchronous mirroring.

© Copyright IBM Corp. 2010. All rights reserved. 71

4.1 XIV Remote Mirroring overview
The purpose of mirroring is to create a set of consistent data that can be used by production
applications in the event of problems with production volumes or for other purposes.

XIV remote mirroring is application and operating system independent, and does not require
server processor cycle usage.

4.1.1 XIV Remote Mirror terminology

It is worth going through and becoming familiar with several terms used throughout the next
chapters involving remote mirroring. A number of terms, meanings, and usage with regards to
XIV and synchronous remote mirroring are noted below:
򐂰 Local site: This site is made up of the primary storage and the servers running
applications with the XIV Storage System.
򐂰 Remote site: This site holds the mirror copy of the data on another XIV Storage System
and usually standby servers as well. In this case, the remote site is capable of becoming
the active production site with consistent data available in the event of a failure at the local
site.
򐂰 Primary: This denotes the XIV designated under normal conditions to serve hosts and
have its data replicated to a secondary XIV for disaster recovery purposes.
򐂰 Secondary. This denotes the XIV designated under normal conditions to act as the mirror
(backup) for the primary, and that could be set to replace the primary if the primary fails.
򐂰 Consistency groups (CG): A consistency group is a set of related volumes on the same
XIV Storage System that are treated as a single consistent unit. Consistency groups are
supported within Remote Mirroring.
򐂰 Coupling: This is the pairing of volumes or consistency groups (CGs) to form a mirror
relationship between the source of the replication (master) and the target (slave).
򐂰 Peer : This is one side of a coupling. It can either be a volume or a consistency group.
However, peers must be of the same type (that is, both volumes or CGs). Whenever a
coupling is defined, a role is specified for each peer. One peer is designated as the master
and the other peer is designated as the slave.
򐂰 Role: This denotes the actual role that the peer is fulfilling:
– Master : A role that indicates that the peer serves host requests and acts as the source
for replication. Changing a peer’s role to master from slave may be warranted after a
disruption of the current master’s service either due to a disaster or to planned service
maintenance.
– Slave: A role that indicates that the peer does not serve host requests and acts as the
target for replication. Changing a peer’s role to slave from master may be warranted
after the peer is recovered from a site/system/link failure or disruption that led to the
promotion of the other peer from slave to master. Changing roles can also be done in
preparation for supporting a planned service maintenance.
򐂰 Sync job: This applies to async mirroring only. It denotes a synchronization procedure run
by the master at specified user-configured intervals corresponding to the asynchronous
mirroring definition or upon manual execution of a dedicated XCLI command (the related
command is mirror_create_snapshot). The resulting job is dubbed snapshot mirror sync
job or ad-hoc sync job, or manual sync job in contrast with a scheduled sync job. The sync
job entails synchronization of data updates recorded on the master since the creation time
of the most recent snapshot that was successfully synchronized.

72 IBM XIV Storage System: Copy Services and Migration

 򐂰 Asynchronous schedule interval: This applies to asynchronous mirroring only. It
represents, per given coupling, how often the master automatically runs a new sync job.
For example, if the pertinent mirroring configuration parameter specifies a 60-minute
interval, then during a period of 1 day, 24 sync jobs will be created.
򐂰 Recovery Point Objective (RPO): The RPO is a setting that is only applicable to
asynchronous mirroring. It represents an objective set by the user implying the maximal
currency difference considered acceptable between the mirror peers (the actual difference
between mirror peers can be shorter or longer than the RPO set).
An RPO of zero indicates that no currency difference between the mirror peers is
acceptable. An RPO that is greater than zero indicates that the replicated volume is less
current or lags somewhat behind the master volume, and that there is a potential for
certain transactions that have been run against the production volume to be rerun when
applications come up on the replicated volume.
For XIV asynchronous mirroring, the required RPO is user-specified. The XIV system then
reports effective RPO and compares it to the required RPO.
Connectivity, bandwidth, and distance between the XIV system that contains the
production volume and the XIV system that contains the replicated copy directly impact
RPO. More connectivity, greater bandwidth, and less distance typically enable a lower
RPO.

4.1.2 XIV Remote Mirroring modes

As mentioned in our introduction, XIV supports both synchronous mirroring and
asynchronous mirroring:
򐂰 XIV synchronous mirroring
XIV synchronous mirroring is designed to accommodate a requirement for zero RPO.
To ensure that data is also written to the Secondary XIV (slave role), an acknowledgement
of the write operation to the host is only issued after the data has been written to both XIV
systems. This ensures the consistency of mirroring peers. A write acknowledgement is
sent to the host once the write data has been cached into two separate XIV modules at
each site. This is depicted in Figure 4-1.

Host Server

2
4
1. Host Write to Master XIV 3
(data placed in cache of 2
Modules)
Local XIV Remote XIV
2. Master replicates to Slave (Master)
XIV (data placed in cache of (Slave)
2 Modules)
3. Slave acknowledges write
complete to Master
4. Master acknowledges write
complete to application

Figure 4-1 XIV synchronous mirroring

Chapter 4. Remote Mirroring 73

 Host read operations are performed from the Primary XIV (master role), whereas writing is
performed at the primary (master role) and replicated to the Secondary XIV systems.
Refer to 5.5, “Synchronous mirror step-by-step scenario” on page 137, for more details.
򐂰 XIV asynchronous mirroring
XIV asynchronous mirroring is designed to provide a consistent replica of data on a target
peer through timely replication of data changes recorded on a source peer.
XIV Asynchronous mirroring exploits the XIV snapshot function, which creates a
point-in-time (PiT) image. In XIV asynchronous mirroring, successive snapshots
(point-in-time images) are made and used to create consistent data on the slave peers.
The system sync job copies the data corresponding to the differences between two
designated snapshots on the master (most_recent and last_replicated).
For XIV asynchronous mirroring, acknowledgement of write complete is returned to the
application as soon as the write data has been received at the local XIV system, as shown
in Figure 4-2. Refer to 6.6, “Detailed asynchronous mirroring process” on page 173, for
details.

Application Server

3
2
4
1. Host Write to Master XIV
(data placed in cache of 2
Modules)) Local XIV
(Master) Remote XIV
2. Master acknowledges write (Slave)
complete to application
3. Master replicates to Slave
4. Slave acknowledges write
complete

Figure 4-2 XIV asynchronous mirroring

74 IBM XIV Storage System: Copy Services and Migration

4.2 Mirroring schemes
Mirroring, whether synchronous or asynchronous, requires two or more XIV systems. The
source and target of the asynchronous mirroring can reside on the same site and form a local
mirroring or they can reside on different sites and enable a disaster recovery plan. Figure 4-3
shows how peers can be spread across multiple storage systems and sites.

Replication Scheme XIV System E

XIV System B
XIV System A

Mirrored CG M irro red

Master

Mirrored Vol Mirrored CG

Master Master

Storage
Pool
Mirrored CG
Slave
XIV System D
XIV System C

Mirrored Vol
Slave

Mirrored Vol Mirrored CG

Slave Mirrored Vol
Master Slave

Storage S tora ge
Pool Po o l

Figure 4-3 Mirroring replication schemes

Up to 16 targets can be referenced by a single system. A system can host replication sources
and separate replication targets simultaneously.

In a bi-directional configuration, an XIV system concurrently functions as the replication

source (master) for one or more couplings, and as the replication target (slave) for other
couplings. If production applications are eventually running at both sides, the applications at
each site are independent from each other to ensure data consistency in case of a site failure.

Figure 4-3 illustrates possible schemes for how mirroring can be configured.

Chapter 4. Remote Mirroring 75

 Figure 4-4 shows remote mirror connections as shown in the XIV GUI.

Figure 4-4 XIV GUI showing the remote mirror connections

4.2.1 Peer designations and roles

A peer (volume or consistency group) is assigned either a master or a slave role when the
mirror is defined. By default, in a new mirror definition, the location of the master designates
the primary system, and the slave designates the secondary system. A mirror must have
exactly one primary and exactly one secondary. The actual function of the peer is determined
based on the peer role (see below).

Important: A single XIV can contain both master volumes and CGs (mirroring to another
XIV) and slave volumes and CGs (mirroring from another XIV). Peers in a master role and
peers in a slave role on the same XIV system must belong to different mirror couplings.

76 IBM XIV Storage System: Copy Services and Migration

The various mirroring role status options are:
򐂰 Designations:
– Primary: the designation of the source peer, which is initially assigned the master role
– Secondary: the designation of the target peer, which initially plays the slave role
򐂰 Role status:
– Master: denotes the peer with the source data in a mirror coupling. Such peers serve
host requests and are the source for synchronization updates to the slave peer. In
synchronous mirroring, slave and master roles can be switched (switch_role
command) if the status is synchronized). For both synchronous and asynchronous
mirroring, the master can be changed (change_role command) to a slave if the status is
inactive.
– Slave: denotes the active target peer in a mirror. Such peers do not serve host
requests and accept synchronization updates from a corresponding master. A slave
LUN could be accessed in read-only mode by a host. In synchronous mirroring, slave
and master roles can be switched (switch_role command) if the status is
synchronized. For both synchronous and asynchronous mirroring, a slave can be
changed (change_role command) to a master regardless of the synchronization state.
As a master the LUN accepts write I/Os. The change_role and switch_role commands
are relevant to disaster recovery situations and failover scenarios.

Consistency group
With mirroring (synchronous or asynchronous), the major reason for consistency groups is to
handle a large number of mirror pairs as a group (mirrored volumes are consistent). Instead
of dealing with many volume remote mirror pairs individually, consistency groups simplify the
handling of many pairs considerably.

Important: If your mirrored volumes are in a mirrored consistency group you cannot do
mirroring operations like deactivate or change_role on a single volume basis. If you want to
do this, you must remove the volume from the consistency group (refer to “Removing a
volume from a mirrored consistency group” on page 132 or “Removing a volume from a
mirrored consistency group” on page 159).

Consistency groups also play an important role in the recovery process. If mirroring was
suspended (for example, due to complete link failure), data on different slave volumes at the
remote XIV are consistent. However, when the links are up again and resynchronization is
started, data spread across several slave volumes is not consistent until the master state is
synchronized. To preserve the consistent state of the slave volumes, the XIV system
automatically creates a snapshot of each slave volume and keeps it until the remote mirror
volume pair is synchronized (the snapshot is kept until all pairs are synchronized in order to
enable restoration to the same consistent point in time). If the remote mirror pairs are in a
consistency group, then the snapshot is taken for the whole group of slave volumes and the
snapshots are preserved until all pairs are synchronized. Then the snapshot is deleted
automatically.

Chapter 4. Remote Mirroring 77

4.2.2 Operational procedures
Mirroring operations involve configuration, initialization, ongoing operation, handling of
communication failures, and role switching activities.

The following list defines the mirroring operation activities:

򐂰 Configuration
Local and remote replication peers are defined by an administrator who specifies the
master and slave peers roles. These peers can be volumes or consistency groups. The
secondary peer provides a backup of the primary.
򐂰 Initialization
Mirroring operations begin with a master volume that contains data and a formatted slave
volume. The first step is to copy the data from the master volume (or CG) to the slave
volume (or CG). This process is called initialization. Initialization is performed once in the
lifetime of a mirror. After it is performed, both volumes or CGs are considered to be
synchronized to a specific point in time. The completion of initialization marks the first
point-in-time that a consistent master replica on the slave is available. Details of the
process differ depending on the mirroring mode (synchronous or asynchronous). Refer to
5.5, “Synchronous mirror step-by-step scenario” on page 137, for synchronous mirroring
and 6.6, “Detailed asynchronous mirroring process” on page 173, for asynchronous
mirroring.
򐂰 Ongoing operation
After the initialization process is complete, mirroring ensues.
In synchronous mirroring, normal ongoing operation means that all data written to the
primary volume or CG is first mirrored to the slave volume or CG. At any point in time, the
master and slave volumes or CGs will be identical except for any unacknowledged
(pending) writes.
In asynchronous mirroring, ongoing operation means that data is written to the master
volume or CG and then replicated on the slave volume or CG at specified intervals.
򐂰 Monitoring
The XIV System effectively monitors the mirror activity and places events in the event log
for error conditions. Alerts can be set up to notify the administrator of such conditions. You
must have set up SNMP trap monitoring tools or e-mail notification to be informed about
abnormal mirroring situations.
򐂰 Handling of communication failures
From time to time the communication between the sites might break down. The master
continues to serve host requests, yet mirroring will only resume once the link is restored.
Events will be generated for link failures.
򐂰 Role switching (synchronous mirroring only)
If required, mirror peer roles of slave and master can be switched. A role switching is
always initiated at the master site. Usually, this is done for certain maintenance operations
or because of a drill that tests the disaster recovery procedures.
򐂰 Role change
In case of a disaster at the primary site, the master peer might fail. To allow read/write
access to the volumes at the remote site, the volume’s role must be changed from slave to
master. A role change only changes the role of the XIV volumes or CGs to which the
command was addressed. Remote mirror peer volumes or CGs are not changed
automatically. That is why changing roles on both mirror sides if mirroring is to be restored
is imperative (if possible).

78 IBM XIV Storage System: Copy Services and Migration

4.2.3 Mirroring status
The status of a mirror is affected by a number of factors such as the links between the XIVs or
the initialization state.

Link status
The link status reflects the connection from the master to the slave volume or CG. A link has
a direction (from local site to remote or vice versa). A failed link or a failed secondary system
both result in a link error status. The link state is one of the factors determining the mirror
operational status. Link states are as follows:
򐂰 OK: link is up and functioning
򐂰 Error: link is down

Figure 4-5 and Figure 4-6 depict how the link status is reflected in the XIV GUI, respectively.

Figure 4-5 Link up

Figure 4-6 Link down

If there are several links (at least two) in one direction and one link fails, this usually does not
affect mirroring as long as the bandwidth of the remaining link is high enough to keep up with
the data traffic.

Monitoring the link utilization

The mirroring bandwidth of the links must be high enough to cope with the data traffic caused
by the changes on the master volumes. During the planning phase, before setting up
mirroring, monitor the write activity to the local volumes. The bandwidth of the links for
mirroring must be as large as the peak write workload.

Chapter 4. Remote Mirroring 79

 After mirroring has been implemented, from time to time monitor the utilization of the links.
The XIV statistics panels allow you to select targets to show the data traffic to remote XIV
Systems, as shown in Figure 4-7.

Figure 4-7 Monitoring link utilization

Mirror operational status

Mirror operational status is defined as either operational or non_operational.
򐂰 Mirroring is operational if:
– The activation state is active.
– The link is UP.
– Both peers have different roles (master or slave).
– The mirror is active.
򐂰 Mirroring is non_operational if:
– The mirror is inactive.
– The link is in an error state or deactivated (link down).

Synchronous mirroring states

Note: This section only applies to synchronous mirroring.

The synchronization status reflects the consistency of the data between the master and slave
volumes. Because the purpose of the remote mirroring feature is to ensure that the slave
volumes are an identical copy of the master volumes, this status indicates whether this
objective is currently being achieved.

80 IBM XIV Storage System: Copy Services and Migration

The following states or statuses are possible.
򐂰 Initializing
The first step in remote mirroring is to create a copy of all the data from the master volume
or CG to the slave volume or CG. During this initial copy phase, the status remains
initializing.
򐂰 Synchronized (master volume or CG only)/consistent (slave volume or CG only)
This status indicates that all data that has been written to the master volume or CG has
also been written to the slave volume or CG. Ideally, the master and slave volumes or CGs
must always be synchronized. However, this does not always indicate that the two
volumes are absolutely identical in case of a disaster because there are situations when
there might be a limited amount of data that was written to one volume, but that was not
yet written to its peer volume. This means that the write operations have not yet been
acknowledged. These are also known as pending writes or data in flight.
򐂰 Unsynchronized (master volume only)/inconsistent (slave volume only)
After a volume or CG has completed the initializing stage and achieved the synchronized
status it can become unsynchronized (master) or inconsistent (slave). This occurs when it
is not known whether all the data that has been written to the master volume has also
been written to the slave volume. This status can occur in the following cases:
– The communications link is down and as a result certain data might have been written
to the master volume, but was not yet written to the slave volume.
– Secondary XIV is down. This is similar to communication link errors because in this
state, the Primary XIV is updated, whereas the secondary is not.
– Remote mirroring is deactivated. As a result, certain data might have been written to
the master volume and not to the secondary volume.

The XIV keeps track of the partitions that have been modified on the master volumes and
when the link is operational again or the remote mirroring is reactivated. These changed
partitions can be sent to the remote XIV and applied to the slave volumes there.

Asynchronous mirroring states

Note: This section only applies to asynchronous mirroring.

The mirror states can be one of the following:

򐂰 Inactive: The synchronization process is disabled. It is possible to delete a mirror.
򐂰 Initializing: The initial copy is not done yet. Synchronization does not start until the
initialization completes.
򐂰 When initialization is complete, the synchronization process is enabled. It is possible to
run sync jobs and copy data between master and slave. The possible synchronization
states are:
– RPO_OK: Synchronization has completed within the specified sync job interval time
(RPO).
– RPO_Lagging: Synchronization has completed but took longer that the specified
interval time (RPO).

Chapter 4. Remote Mirroring 81

4.3 XIV Remote Mirroring usage
Remote Mirroring solutions can be used to address multiple types of failures and planned
outages, from events affecting a single XIV system or its components, to events affecting an
entire data center or campus, or events affecting an entire geographical region. When the
production XIV system and the disaster recovery (DR) XIV system are separated by
increasing distance, disaster recovery protection for more levels of failures is possible, as
illustrated in Figure 4-8. A global distance disaster recovery solution protects from
single-system failures, local disasters, and regional disasters.

Remote Mirroring

Local Disaster Regional Disasters

• Terrorist Attacks • Electric grid failures
• Human Error • Natural disasters
• HVAC failures - Floods
Single System Failure • Power failures - Hurricanes
• Component failures • Building Fire - Earthquakes
• Single system failures • Architectural failures
• Planned Maintenance

High Availability Metro Distance Recovery Global Distance Recovery

IBM System StorageTM © 200 9 IBM C orp ora tio n 3

Figure 4-8 Disaster recovery protection levels

Several configurations are possible:

򐂰 Single-site high-availability XIV Remote Mirroring configuration
Protection for the event of a failure or planned outage of an XIV system (single-system
failure) can be provided by a zero-distance high-availability (HA) solution including another
XIV system in the same location (zero distance). Typical usage of this configuration is an
XIV synchronous mirroring solution that is part of a high-availability clustering solution
including both servers and XIV storage systems. Figure 4-9 shows a single-site
high-availability configuration (where both XIV systems are in the same data center).

Figure 4-9 Single site HA configuration

82 IBM XIV Storage System: Copy Services and Migration

򐂰 Metro region XIV Remote Mirroring configuration
Protection for the event of a failure or planned outage of an entire location (local disaster)
can be provided by a metro distance disaster recovery solution, including another XIV
system in a different location within a metro region. The two XIV systems may be in
different buildings on a corporate campus or in different buildings within the same city
(typically up to approximately 100 km apart). Typical usage of this configuration is an XIV
synchronous mirroring solution. Figure 4-10 shows a metro region disaster recovery
configuration.

Figure 4-10 Metro region disaster recovery configuration

򐂰 Out-of-region XIV Remote Mirroring configuration

Protection for the event of a failure or planned outage of an entire geographic region
(regional disaster) can be provided by a global distance disaster recovery solution
including another XIV system in a different location outside the metro region. (The two
locations may be separated by up to a global distance.) Typical usage of this configuration
is an XIV asynchronous mirroring solution. Figure 4-11 shows an out-of-region disaster
recovery configuration.

Figure 4-11 Out-of-region disaster recovery configuration

Chapter 4. Remote Mirroring 83

 򐂰 Metro region plus out-of-region XIV mirroring configuration
Certain volumes may be protected by a metro distance disaster recovery configuration,
and other volumes may be protected by a global distance disaster recovery configuration,
as shown in the configuration in Figure 4-12. Typical usage of this configuration is an XIV
synchronous Mirroring solution for a set of volumes with a requirement for zero RPO, and
an XIV asynchronous mirroring solution for a set of volumes with a requirement for a low,
but non-zero RPO. Figure 4-12 shows a metro region plus out-of-region configuration.

Figure 4-12 Metro region plus out-of-region configuration

Using snapshots
Snapshots can be used with Remote Mirroring to provide copies of production data for
business or IT purposes. Moreover, when used with Remote Mirroring, snapshots provide
protection against data corruption.

84 IBM XIV Storage System: Copy Services and Migration

Like any continuous or near-continuous remote mirroring solution, XIV Remote Mirroring
cannot protect against software data corruption because the corrupted data will be copied as
part of the remote mirroring solution. However, the XIV snapshot function provides a
point-in-time image that may be used for rapid restore in the event of software data corruption
(that occurred after the snapshot was taken), and XIV snapshot may be used in combination
with XIV Remote Mirroring, as illustrated in Figure 4-13.

Remote Mirroring
Point in Time
Copy

Local Disaster Regional Disasters

• Terrorist Attacks • Electric grid failures
• Human Error • Natural d isasters
• HVAC failures - Floods
Data Corruption Single System Failure
• Power failures - Hurricanes
• Component failures • Building Fire - Earthquakes
• Single system failures • Architectural failures
Point In Time • Planned Maintenance
Disk Backup,
High Availability Metro Distance Recovery Global Distance Recovery
Extra Copies

IBMS t St TM 8
Figure 4-13 Combining snapshots with Remote Mirroring

Note that recovery using a snapshot warrants deletion and recreation of the mirror.
򐂰 XIV snapshot (within a single XIV system)
Protection for the event of software data corruption can be provided by a point-in-time
backup solution using the XIV snapshot function within the XIV system that contains the
production volumes. Figure 4-14 shows a single-system point-in-time online backup
configuration.

Figure 4-14 Point-in-time online backup configuration

򐂰 XIV local snapshot plus Remote Mirroring configuration

An XIV snapshot of the production (local) volume may be used in addition to XIV Remote
Mirroring of the production volume when protection from logical data corruption is required
in addition to protection against failures and disasters. The additional XIV snapshot of the
production volume provides a quick restore to recover from data corruption. An additional
Snapshot of the production (local) volume may also be used for other business or IT
purposes (for example, reporting, data mining, development and test, and so on).

Chapter 4. Remote Mirroring 85

 Figure 4-15 shows an XIV local snapshot plus Remote Mirroring configuration.

Figure 4-15 Local snapshot plus Remote Mirroring configuration

򐂰 XIV remote snapshot plus Remote Mirroring configuration

An XIV snapshot of the consistent replicated data at the remote site may be used in
addition to XIV Remote Mirroring to provide an additional consistent copy of data that can
be used for business purposes such as data mining, reporting, and for IT purposes, such
as remote backup to tape or development, test, and quality assurance. Figure 4-16 shows
an XIV remote snapshot plus Remote Mirroring configuration.

Figure 4-16 XIV remote snapshot plus Remote Mirroring configuration

4.4 XIV Remote Mirroring actions

These XIV Remote Mirroring actions are the fundamental building blocks of XIV Remote
Mirroring solutions and usage scenarios.

4.4.1 Defining the XIV mirroring target

In order to connect two XIV systems for remote mirroring, each system must be defined to be
a mirroring target of the other. An XIV mirroring target is an XIV system with volumes that
receive data copied through XIV remote mirroring. Defining an XIV mirroring target for an XIV
system simply involves giving the target a name and specifying whether Fibre Channel or
iSCSI protocol will be used to copy the data. For a practical illustration refer to 4.11.2,
“Remote mirror target configuration” on page 118.

86 IBM XIV Storage System: Copy Services and Migration

XIV Remote Mirroring copies data from a peer on one XIV system to a peer on another XIV
system (the mirroring target system). Whereas the basic underlying mirroring relationship is a
one-to-one relationship between two peers, XIV systems may be connected in several
different ways:
򐂰 XIV target configuration: one-to-one
The most typical XIV Remote Mirroring configuration is a one-to-one relationship between
a local XIV system (production system) and a remote XIV system (DR system), as shown
in Figure 4-17. This configuration is typical where there is a single production site and a
single disaster recovery (DR) site.

Target

M
Figure 4-17 One-to-one target configuration

During normal remote mirroring operation, one XIV system (at the DR site) will be active
as a mirroring target. The other XIV system (at the local production site) will be active as a
mirroring target only when it becomes available again after an outage and switch of
production to the DR site. Changes made while production was running at the DR site are
copied back to the original production site, as shown in Figure 4-18.

Target

M
Figure 4-18 Copying changes back to production

In a configuration with two identically provisioned sites, production may be periodically

switched from one site to another as part of normal operation, and the XIV system that is
the active mirroring target will be switched at the same time. (The mirror_switch_roles
command allows for switching roles in both synchronous and asynchronous mirroring.
Note that there are special requirements for doing so with asynchronous mirroring.)

Chapter 4. Remote Mirroring 87

 򐂰 XIV target configuration: synchronous and asynchronous one-to-one
XIV supports both synchronous and asynchronous mirroring (for different peers) on the
same XIV system, so a single local XIV system could have certain volumes synchronously
mirrored to a remote XIV system, whereas other peers are asynchronously mirrored to the
same remote XIV system as shown in Figure 4-19. Highly response-time-sensitive
volumes could be asynchronously mirrored and less response-time-sensitive volumes
could be synchronously mirrored to a single remote XIV.

Figure 4-19 Synchronous and asynchronous peers

򐂰 XIV target configuration: fan-out

A single local (production) XIV system may be connected to two remote (DR) XIV systems
in a fan-out configuration, as shown in Figure 4-20. Both remote XIV systems could be at
the same location, or each of the two target systems could be at a different location.
Certain volumes on the local XIV system are copied to one remote XIV system, and other
volumes on the same local XIV system are copied to a different remote XIV system. This
configuration may be used when each XIV system at the DR site has less available
capacity than the XIV system at the local site.

Target

Target
Figure 4-20 Fan-out target configuration

88 IBM XIV Storage System: Copy Services and Migration

򐂰 XIV target configuration: synchronous and asynchronous fan-out
XIV supports both synchronous and asynchronous mirroring (for different peers) on the
same XIV system, so a single local XIV system could have certain peers synchronously
mirrored to a remote XIV system at a metro distance, whereas other peers are
asynchronously mirrored to a remote XIV system at a global distance, as shown in
Figure 4-21. This configuration may be used when higher priority data is synchronously
mirrored to another XIV system within the metro area, and lower priority data is
asynchronously mirrored to an XIV system within or outside the metro area.

Target

Target

Figure 4-21 Synchronous and asynchronous fan-out

򐂰 XIV target configuration: fan-in

Two (or more) local XIV systems may have peers mirrored to a single remote XIV system
in a fan-in configuration, as shown in Figure 4-22. This configuration must be evaluated
carefully and used with caution because it includes the risk of overloading the single
remote XIV system. The performance capability of the single remote XIV system must be
carefully reviewed before implementing a fan-in configuration.
This configuration may be used in situations where there is a single disaster recovery data
center supporting multiple production data centers, or when multiple XIV systems are
mirrored to a single XIV system at a service provider.

Target

Figure 4-22 Fan-in configuration

Chapter 4. Remote Mirroring 89

 򐂰 XIV target configuration: bi-directional
Two different XIV systems may have different volumes mirrored in a bi-directional
configuration, as shown in Figure 4-23. This configuration may be used for situations
where there are two active production sites and each site provides a DR solution for the
other. Each XIV system is active as a production system for certain peers and as a
mirroring target for other peers.

S Target

Target M

Figure 4-23 Bi-directional configuration

4.4.2 Setting the maximum initialization and synchronization rates

The XIV system allows a user-specifiable maximum rate (in MBps) for remote mirroring
coupling initialization, and a different user-specifiable maximum rate for re-synchronization.
The initialization rate and resynchronization rate are specified for each mirroring target using
the XCLI command target_config_sync_rates. As such, if different rates are required for
different volumes for a single remote target XIV system, multiple logical targets may be
defined for the single physical remote XIV system. The actual effective initialization or
synchronization rate will also be dependent on the number and speed of connections
between the XIV systems. The maximum initialization rate must be less than or equal to the
maximum sync job rate (asynchronous mirroring only), which must be less than or equal to
the maximum resynchronization rate. The defaults are:
򐂰 Maximum initialization rate: 100 MBps
򐂰 Maximum sync job: 300 MBps
򐂰 Maximum resync rate: 300 MBps

90 IBM XIV Storage System: Copy Services and Migration

4.4.3 Connecting XIV mirroring ports
After defining remote mirroring targets, one-to-one connections must be made between ports
on each XIV system. For an illustration of these actions using the GUI or the XCLI, refer to
4.11, “Using the GUI or XCLI for Remote Mirroring actions” on page 113.
򐂰 FC ports
For XIV Fibre Channel (FC) ports, connections are unidirectional—from an initiator port
(port 4 is configured as a Fibre Channel initiator by default) on the source XIV system to a
target port (typically port 2) on the target XIV system. Use a minimum of four connections
(two connections in each direction, from ports in two different modules, using a total of
eight ports) to provide availability protection. Connections must be made between ports on
modules with the same number on each XIV system (for example, from module 9 to
module 9 and from module 6 to module 6, as shown in Figure 4-24).

9 FC SAN 9
8 8
7 7
6 6
Data,
5 , FC SAN Data,
5 ,
Mgt
4
Data, , Mgt
4
Data, ,
Mgt Mgt

Figure 4-24 Connecting XIV mirroring ports (FC connections)

In Figure 4-24, the solid lines represent mirroring connections used during normal
operation (the mirroring target system is on the right), and the dotted lines represent
mirroring connections used when production is running at the disaster recovery site and
changes are being copied back to the original production site (mirroring target is on the
left.)
XIV Fibre Channel ports may be easily and dynamically configured as initiator or target
ports.
򐂰 iSCSI ports
For iSCSI ports, connections are bi-directional.
Use a minimum of two connections (with each of these ports in a different module) using a
total of four ports to provide availability protection. In Figure 4-25 on page 92, the solid
lines represent data flow during normal operation and the dotted lines represent data flow
when production is running at the disaster recovery site and changes are being copied
back to the original production site.

Chapter 4. Remote Mirroring 91

 Connections must be made between ports on modules with the same number on each
XIV system (for example, from module 9 to module 9 and from module 7 to module 7, as
shown in Figure 4-25).

9
8
IP Network 9
8
7 7

Data, , Data, ,
IP Network
DMatgat , , DMatgat , ,
Mgt Mgt

Figure 4-25 Connecting XIV mirroring ports (iSCSI connections)

Note: For asynchronous mirroring over iSCSI links, a reliable, dedicated network must
be available. It requires consistent network bandwidth and a non-shared link.

4.4.4 Defining the XIV mirror coupling and peers: volume

After the mirroring targets have been defined, a coupling or mirror may be defined, creating a
mirroring relationship between two peers.

Before discussing actions involved in creating mirroring pairs, we must introduce the basic
XIV concepts used in the discussion.

Storage pools, volumes, and consistency groups

An XIV storage pool is a purely administrative construct used to manage XIV logical and
physical capacity allocation.

An XIV volume is a logical volume that is presented to an external server as a logical unit
number (LUN). An XIV volume is allocated from logical and physical capacity within a single
XIV storage pool. The physical capacity on which data for an XIV volume is stored is always
spread across all available disk drives in the XIV system

The XIV system is data aware. It monitors and reports the amount of physical data written to
a logical volume and does not copy any part of the volume that has not been used yet to store
any actual data.

92 IBM XIV Storage System: Copy Services and Migration

In Figure 4-26, seven logical volumes have been allocated from a storage pool with 40 TB of
capacity. Remember that the capacity assigned to a storage pool and its volumes is spread
across all available physical disk drives in the XIV system.

40TB
Storage
Pool

Figure 4-26 Storage pool with seven volumes

With Remote Mirroring, the concept of consistency group represents a logical container for a
group of volumes, allowing them to be managed as a single unit. Instead of dealing with many
volume remote mirror pairs individually, consistency groups simplify the handling of many
pairs considerably.

An XIV consistency group exists within the boundary of an XIV storage pool in a single XIV
system (in other words, you can have different CGs in different storage pools within an XIV
storage system, but a CG cannot span multiple storage pools). All volumes in a particular
consistency group are in the same XIV storage pool.

In Figure 4-27, an XIV storage pool with 40 TB capacity contains seven logical volumes. One
consistency group has been defined for the XIV storage pool, but no volumes have been
added to or created in the consistency group.

40TB CG
Storage
Pool

Figure 4-27 Consistency group defined

Volumes may be easily and dynamically (that is, without stopping mirroring or application
I/Os) added to a consistency group.

Chapter 4. Remote Mirroring 93

 In Figure 4-28, five of the seven existing volumes in the storage pool have been added to the
consistency group in the storage pool. One or more additional volumes may be dynamically
added to the consistency group at any time. Also, volumes may be dynamically moved from
another storage pool to the storage pool containing the consistency group, and then added to
the consistency group.

40TB CG
Storage
Pool

Figure 4-28 Volumes added to the consistency group

Volumes may also be easily and dynamically removed from an XIV consistency group. In
Figure 4-29, one of the five volumes has been removed from the consistency group, leaving
four volumes remaining in the consistency group. It is also possible to remove all volumes
from a consistency group.

40TB CG
Storage
Pool

Figure 4-29 Volume removed from the consistency group

Dependent write consistency

XIV Remote Mirroring provides dependent write consistency, preserving the order of
dependent writes in the mirrored data. Dependent write consistency is also referred to as
crash consistency or power-loss consistency, and applications and databases are developed
to be able to perform a fast restart from volumes that are consistent in terms of dependent
writes.

94 IBM XIV Storage System: Copy Services and Migration

Dependent writes: normal operation
Applications or databases often manage dependent write consistency using a 3-step process
such as the sequence of three writes shown in Figure 4-30. Even when the writes are
directed at different logical volumes, the application ensures that the writes are committed in
order during normal operation.

2) Update Record
DB

1) Intend to update DB

3) DB updated Log

Figure 4-30 Dependent writes: normal operation

Dependent writes: failure scenario

In the event of a failure, applications or databases manage dependent writes, as shown in
Figure 4-31. If the database record is not updated (step 2), the application does not allow DB
updated (step 3) to be written to the log.

2) Update Record
x DB

1) Intend to update DB

3) DB updated Log

Figure 4-31 Dependent writes: failure scenario

Just as the application or database manages dependent write consistency for the production
volumes, the XIV system must manage dependent write consistency for the mirror target
volumes.

Chapter 4. Remote Mirroring 95

 If multiple volumes will have dependent write activity, they may be put into a single storage
pool in the XIV system and then added to an XIV consistency group to be managed as a
single unit for remote mirroring. Any mirroring actions are taken simultaneously against the
mirrored consistency group as a whole, preserving dependent write consistency. Mirroring
actions cannot be taken against an individual volume pair while it is part of a mirrored CG.
However, an individual volume pair may be dynamically removed from the mirrored
consistency group.

XIV also supports creation of application-consistent data in the remote mirroring target
volumes, as discussed 4.5.4, “Creating application-consistent data at both local and the
remote sites” on page 109.

Defining mirror coupling and peers

After the remote mirroring targets have been defined, a coupling or mirror may be defined,
creating a mirroring relationship between two peers.

The two peers in the mirror coupling may be either two volumes (volume peers) or two
consistency groups (CG peers), as shown in Figure 4-32.

SITE 1 SITE 2
Production DR Test/Recovery Servers

Volume
Coupling/Mirror
M Defined S
Volume
Volume Peer Coupling/Mirror Volume Peer
Designated M Defined S Designated
Primary Volume Secondary
Coupling/Mirror
M Defined S

CG
Consistency Group Peer Coupling/Mirror Consistency Group Peer
Primary Designation (P) P/M Defined S/S Secondary Designation (S)
Master Role (M) Slave Role (S)

Figure 4-32 Defining mirror coupling

Each of the two peers in the mirroring relationship is given a designation and a role. The
designation indicates the original or normal function of each of the two peers—either primary
or secondary. The peer designation does not change with operational actions or commands.
(If necessary, the peer designation may be changed by explicit user command or action.)

The role of a peer indicates its current (perhaps temporary) operational function (either
master or slave). The operational role of a peer may change as the result of user commands
or actions. Peer roles typically change during DR testing or a true disaster recovery and
production site switch.

When a mirror coupling is created, the first peer specified (for example, the volumes or CG at
site 1, as shown in Figure 4-32) is the source for data to be replicated to the target system, so
it is given the primary designation and the master role.

The second peer specified (or automatically created by the XIV system) when the mirroring
coupling is created is the target of data replication, so it is given the secondary designation
and the slave role.

96 IBM XIV Storage System: Copy Services and Migration

 When a mirror coupling relationship is first created, no data movement occurs.

4.4.5 Activating an XIV mirror coupling

When an XIV mirror coupling is first activated, all actual data existing on the master is copied
to the slave. This process is referred to as initialization. XIV Remote Mirroring copies volume
identification information (that is, physical volume ID/PVID) and any actual data on the
volumes. Space that has not been used is not copied.

Initialization may take a significant amount of time if a large amount of data exists on the
master when a mirror coupling is activated. As discussed earlier, the rate for this initial copy of
data can be specified by the user. The speed of this initial copy of data will also be affected by
the connectivity and bandwidth (number of links and link speed) between the XIV primary and
secondary systems.

As an option to remove the impact of distance on initialization, XIV mirroring may be initialized
with the target system installed locally, and the target system may be disconnected after
initialization, shipped to the remote site and reconnected, and mirroring reactivated.

If a remote mirroring configuration is set up when a volume is first created (that is, before any
application data has been written to the volume), initialization will be very quick.

When an XIV consistency group mirror coupling is created, the CG must be empty so there is
no data movement and the initialization process is extremely fast.

The mirror coupling status at the end of initialization differs for XIV synchronous mirroring and
XIV asynchronous mirroring (see “Synchronous mirroring states” on page 80 and “Storage
pools, volumes, and consistency groups” on page 92), but in either case, when initialization is
complete, a consistent set of data exists at the remote site. See Figure 4-33.

SITE 1 SITE 2
Production DR Test/Recovery Servers

Volume
Coupling/Mirror
M Active S
Volume
Volume Peer Coupling/Mirror Volume Peer
Designated M Active S Designated
Primary Volume Secondary
Coupling/Mirror
M Active S

CG
Consistency Group Peer Coupling/Mirror Consistency Group Peer
Primary Designation (P) P/M Active S/S Secondary Designation (S)
Master Role (M) Slave Role (S)

Figure 4-33 Active mirror coupling

4.4.6 Adding volume mirror coupling to consistency group mirror coupling

Once a volume mirror coupling has completed initialization, the master volume may be added
to a mirrored consistency group in the same storage pool (note that with each mirroring type

Chapter 4. Remote Mirroring 97

 there are certain additional constraints, such as same role, target, schedule, and so on). The
slave volume is automatically added to the consistency group on the remote XIV system.

In Figure 4-34, three active volume couplings that have completed initialization have been
moved into the active mirrored consistency group.

SITE 1 SITE 2
Production DR Test/Recovery Servers

P/M S/S
CG
Consistency Group Peer Coupling/Mirror Consistency Group Peer
Primary Designation (P) Active Secondary Designation (S)
Master Role (M) Slave Role (S)

Figure 4-34 Consistency group mirror coupling

One or more additional mirrored volumes may be added to a mirrored consistency group at a
later time in the same way.

It is also important to realize that in a CG all volumes have the same role. Also, consistency
groups are handled as a single entity and, for example, in asynchronous mirroring, a delay in
replicating a single volume affects the status of the entire CG.

4.4.7 Normal operation: volume mirror coupling and CG mirror coupling

XIV mirroring normal operation begins after initialization has completed successfully and all
actual data on the master volume at the time of activation has been copied to the slave
volume. During normal operation, a consistent set of data is available on the slave volumes.

Normal operation, statuses, and reporting differ for XIV synchronous mirroring and XIV
asynchronous mirroring. Refer to Chapter 5, “Synchronous Remote Mirroring” on page 125,
and Chapter 6, “Asynchronous Remote Mirroring” on page 149, for details.

98 IBM XIV Storage System: Copy Services and Migration

 During normal operation, a single XIV system may contain one or more mirrors of volume
peers as well as one or more mirrors of CG peers, as shown in Figure 4-35.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Remote Target

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary M Active S Designated Secondary
Master Role Slave Role

CG
Coupling/Mirror
CG Peer Active CG Peer
Designated Primary Designated Secondary
Master Role M S Slave Role

Figure 4-35 Normal operations: volume mirror coupling and CG mirror coupling

4.4.8 Deactivating XIV mirror coupling: change recording

An XIV mirror coupling may be deactivated by a user command. In this case, the mirror
transitions to standby mode, as shown in Figure 4-36.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary M Standby S Designated Secondary
Master Role Master Role

CG
Coupling/Mirror
CG Peer Standby CG Peer
Designated Primary Designated Secondary
Master Role M S Master Role

Figure 4-36 Deactivating XIV mirror coupling: change recording

During standby mode, a consistent set of data is available at the remote site (site 2, in our
example). The currency of the consistent data ages in comparison to the master volumes,
and the gap increases while mirroring is in standby mode.

Chapter 4. Remote Mirroring 99

 In synchronous mirroring, during standby mode, XIV metadata is used to note which parts of
a master volume have changed but have not yet been replicated to the slave volume
(because mirroring is not currently active). The actual changed data is not retained in cache,
so there is no danger of exhausting cache while mirroring is in standby mode.

When synchronous mirroring is reactivated by a user command or communication is restored,

the metadata is used to resynchronize changes from the master volumes to the slave
volumes. XIV mirroring records changes for master volumes only. If it is desirable to record
changes to both peer volumes while mirroring is in standby mode, the slave volume must be
changed to a master volume.

Note that in asynchronous mirroring, metadata is not used and the comparison between the
most_recent and last_replicated snapshots indicates the data that must be replicated.

Planned deactivation of XIV remote mirroring may be done to suspend remote mirroring
during a planned network outage or DR test, or to reduce bandwidth during a period of peak
load.

4.4.9 Changing role of slave volume or CG

When XIV mirroring is active, the slave volume or CG is locked and write access is prohibited.
To allow write access to a slave peer, in case of failure or unavailability of the master, the
slave volume role must be changed to the master role. Refer to Figure 4-37.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary M Standby M Designated Secondary
Master Role Master Role

CG
Coupling/Mirror
CG Peer Standby CG Peer
Designated Primary Designated Secondary
Master Role M M Master Role

Figure 4-37 Changing role of slave volume or CG

Changing the role of a volume from slave to master allows the volume to be accessed. In
synchronous mirroring, changing the role also starts metadata recording for any changes
made to the volume. This metadata may be used for resynchronization (if the new master
volume remains the master when remote mirroring is reactivated). In asynchronous mirroring,
changing a peer's role automatically reverts the peer to its last_replicated snapshot.

When mirroring is in standby mode, both volumes may have the master role, as shown in the
following section. When changing roles, both peer roles must be changed if possible (the
exception being a site disaster or complete system failure). Changing the role of a slave
volume or CG is typical during a true disaster recovery and production site switch.

100 IBM XIV Storage System: Copy Services and Migration

4.4.10 Changing role of master volume or CG
During a true disaster recovery, to resume production at the remote site a slave must have its
role changed to the master role.

In synchronous mirroring, changing a peer role from master to slave allows the slave to
accept mirrored data from the master and cause deletion of metadata that was used to record
any changes while the peer had the master role.

In asynchronous mirroring, changing a peer's role automatically reverts the peer to its
last_replicated snapshot. If at any point in time the command is run on the slave (changing
the slave to a master), the former master must first be changed to the slave role (upon
recovery of the primary site) before changing the secondary role back from master to slave.

Both peers may temporarily have the master role when a failure at site 1 has resulted in a true
disaster recovery production site switch from site 1 to site 2. When site 1 becomes available
again and there is a requirement to switch production back to site 1, the production changes
made to the volumes at site 2 must be resynchronized to the volumes at site 1. In order to do
this, the peers at site 1 must change their role from master to slave, as shown in Figure 4-38.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary S Standby M Designated Secondary
Slave Role Master Role

CG
Coupling/Mirror
CG Peer Standby CG Peer
Designated Primary Designated Secondary
Slave Role S M Master Role

Figure 4-38 Changing role to slave volume and CG

Chapter 4. Remote Mirroring 101

4.4.11 Mirror reactivation and resynchronization: normal direction
In synchronous mirroring, when mirroring has been in standby mode, any changes to
volumes with the master role were recorded in metadata. Then when mirroring is reactivated,
changes recorded in metadata for the current master volumes are resynchronized to the
current slave volumes. Refer to Figure 4-39.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Remote Target

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary M Active S Designated Secondary
Master Role Slave Role

CG
Coupling/Mirror
CG Peer Active CG Peer
Designated Primary Designated Secondary
Master Role M S Slave Role

Figure 4-39 Mirror reactivation and resynchronization: normal direction

The rate for this resynchronization of changes can be specified by the user in MBps using the
XCLI target_config_sync_rates command.

When XIV mirroring is reactivated in the normal direction, changes recorded at the primary
peers are copied to the secondary peers.

Examples of mirror deactivation and reactivation in the same direction are:

򐂰 Remote mirroring is temporarily inactivated due to communication failure and then
automatically reactivated by the XIV system when communication is restored.
򐂰 Remote mirroring is temporarily inactivated to create an extra copy of consistent data at
the secondary.
򐂰 Remote mirroring is temporarily inactivated via user action during peak load in an
environment with constrained network bandwidth.

102 IBM XIV Storage System: Copy Services and Migration

4.4.12 Reactivation, resynchronization, and reverse direction
When XIV mirroring is reactivated in the reverse direction, as shown in the previous section,
changes recorded at the secondary peers are copied to the primary peers. The primary peers
must change the role from master to slave before mirroring can be reactivated in the reverse
direction. See Figure 4-40.

SITE 1 SITE 2
Production Servers DR Test/Recovery Servers

Remote Target

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary S Active
M Designated Secondary
Slave Role Master Role

CG
Coupling/Mirror
CG Peer Active CG Peer
Designated Primary Designated Secondary
Slave Role S M Master Role

Figure 4-40 Reactivation and resynchronization

A typical usage example of this scenario is when returning to the primary site after a true
disaster recovery with production switched to the secondary peers at the remote site.

4.4.13 Switching roles of mirrored volumes or CGs

When mirroring is active and synchronized (consistent), the master and slave roles of
mirrored volumes or consistency groups may be switched simultaneously. Role switching is
typical for returning mirroring to the normal direction after changes have been mirrored in the
reverse direction after a production site switch. Role switching is also typical for any planned
production site switch. For asynchronous mirroring, host server write activity and replication
activity must be paused very briefly before and during the role switch.

4.4.14 Adding a mirrored volume to a mirrored consistency group

First make sure that the following constraints are respected:
򐂰 Volume and CG must be associated with the same pool
򐂰 Volume is not already part of a CG
򐂰 Command must be issued only on the master CG
򐂰 Command must not be run during initialization of volume or CG

Chapter 4. Remote Mirroring 103

 򐂰 The volume mirroring settings must be identical to those of the CG:
– Mirroring type
– Mirroring role
– Mirroring status
– Mirroring target
– Target pool
򐂰 Both volume synchronization status and mirrored CG synchronization status is RPO OK.

To add a volume mirror to a mirrored consistency group (for instance, when an application
needs additional capacity):
1. Define XIV volume mirror coupling from the additional master volume at XIV 1 to the slave
volume at XIV 2.
2. Activate XIV remote mirroring from the additional master volume at XIV 1 to the slave
volume at XIV 2.
3. Monitor initialization until it is complete. Volume coupling initialization must be complete
before the coupling can be moved to a mirrored CG.
4. Add the additional master volume at XIV 1 to the master consistency group at XIV 1. (The
additional slave volume at XIV 2 will be automatically added to the slave consistency
group at XIV 2.)

In Figure 4-41, one volume has been added to the mirrored XIV consistency group. The
volumes must be in a volume peer relationship and must have completed initialization

SITE 1 SITE 2
Production DR Test/Recovery Servers

M/P S/S

CG
Coupling/Mirror
Active
Consistency Group Peer Consistency Group Peer
Primary Designation (P) Secondary Designation (S)
Master Role (M) Slave Role (S)

Figure 4-41 Adding a mirrored volume to a mirrored consistency group

Refer also to 4.4.4, “Defining the XIV mirror coupling and peers: volume” on page 92, and
4.4.6, “Adding volume mirror coupling to consistency group mirror coupling” on page 97, for
additional details.

4.4.15 Removing a mirrored volume from a mirrored consistency group

If a volume in a mirrored consistency group is no longer being used by an application or if
actions must be taken against the individual volume, it can be dynamically removed from the
consistency group.

104 IBM XIV Storage System: Copy Services and Migration

To remove a volume mirror from a mirrored consistency group:
1. Remove the master volume from the master consistency group at site 1. (The slave
volume at site 2 will be automatically removed from the slave CG.)
2. When a mirrored volume is removed from a mirrored CG, it retains its mirroring status and
settings and continues remote mirroring until deactivated.

In Figure 4-42, one volume has been removed from the example mirrored XIV consistency
group with three volumes. After being removed from the mirrored CG, a volume will continue
to be mirrored as part of a volume peer relationship.

Site 1 Site 2
Production DR Test/Recovery Servers

Volume
Coupling/Mirror
P/M Active S/S
Volume
Coupling/Mirror
P/M Active S/S

P/M S/S
CG
Consistency Group Peer Coupling/Mirror Consistency Group Peer
Primary Designation (P) Active Secondary Designation (S)
Master Role (M) Slave Role (S)

Figure 4-42 Removing a mirrored volume from a mirrored CG

Chapter 4. Remote Mirroring 105

4.4.16 Deleting mirror coupling definitions
When an XIV mirror coupling is deleted, all metadata and mirroring definitions are deleted,
and the peers do not have any relationship at all (Figure 4-43). However, any volumes and
consistency groups mirroring snapshots remain on the local and remote XIV systems. In
order to restart XIV mirroring, a full copy of data is required.

Site 1 Site 2
Production Servers DR Test/Recovery Servers

Figure 4-43 Deleting mirror coupling definitions

Typical usage of mirror deletion is a one-time data migration using remote mirroring. This
includes deleting the XIV mirror couplings after the migration is complete.

106 IBM XIV Storage System: Copy Services and Migration

4.5 Best practice usage scenarios
The following best practice usage scenarios begin with the normal operation remote
mirroring environment shown in Figure 4-44.

Site 1 Site 2
Production Servers DR Test/Recovery Servers

Target
XIV 1 XIV 2

Volume
Volume Peer Coupling/Mirror Volume Peer
Designated Primary M Active S Designated Secondary
Master Role Slave Role

CG
Coupling/Mirror
CG Peer Active CG Peer
Designated Primary Designated Secondary
Master Role M S Slave Role

Figure 4-44 Remote Mirroring environment for scenarios

4.5.1 Failure at primary site: switch production to secondary

This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2,
followed by a failure at XIV 1 with the assumption that the data already existing on the XIV
system at XIV 1 will be available for resynchronization when XIV 1 is repaired and returned to
operation.
1. XIV remote mirroring may have been deactivated by the failure.
2. Change the role of the peer at XIV 2 from slave to master. This allows the peer to be
accessed for writes from a host server, and also causes recording of any changes in
metadata for synchronous mirroring. For asynchronous mirroring, changing the role from
slave to master causes the last replicated snapshot to be restored to the volume. Now
both XIV 1 and XIV 2 peers have the master role.
3. Map the master (secondary) peers at XIV 2 to the DR servers.
4. Bring the XIV 2 peers (now with the master role) online to the DR servers to begin
production workload at XIV 2.
5. When the failure at XIV 1 has been corrected and XIV 1 is available, deactivate mirrors at
XIV 1 if they are not already inactive.
6. Unmap XIV 1 peers from servers if necessary.
7. Activate remote mirroring from the master peers at XIV 2 to the slave peers at XIV 1. This
starts resynchronization of production changes from XIV 2 to XIV 1.
8. Monitor the progress to ensure that resynchronization is complete.
9. Quiesce production applications at XIV 2 to ensure that application-consistent data is
copied to XIV 1.

Chapter 4. Remote Mirroring 107

 10.Unmap master peers at XIV 2 from DR servers.
11.For asynchronous mirroring, monitor completion of sync job and change the replication
interval to never.
12.Monitor to ensure that no more data is flowing from XIV 2 to XIV 1.
13.Switch roles of master and slave. XIV 1 peers now have the master role and XIV 2 peers
now have the slave role.
14.For asynchronous mirroring, change the replication schedule to the desired interval.
15.Map master peers at XIV 1 to the production servers.
16.Bring master peers online to XIV 1 production servers.

4.5.2 Complete destruction of XIV 1

This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2,
followed by complete destruction of XIV 1.
1. Change the role of the peer at XIV 2 from slave to master. This allows the peer to be
accessed for writes from a host server.
2. Map the new master peer at XIV 2 to the DR servers at XIV 2.
3. Bring the XIV 2 peer (now with a master role) online to XIV 2 DR servers to begin
production workload at XIV 2.
4. Deactivate XIV remote mirroring from the master peer at XIV 2 if necessary. (It may have
already been deactivated by the XIV 1 failure.)
5. Delete XIV remote mirroring from the master peer at XIV 2.
6. Rebuild XIV 1, including configuration of the new XIV system at XIV 1, the definition of
remote targets for both XIV 1 and XIV 2, and the definition of connectivity between XIV 1
and XIV 2.
7. Define XIV remote mirroring from the master peer at XIV 2 to the slave peer at XIV 1.
8. Activate XIV remote mirroring from the master peer at XIV 2 to the slave peer at XIV 1.
This causes a full copy of all actual data on the master peer at XIV 2 to the slave volume at
XIV 1.
9. Monitor initialization until it is complete.
10.Quiesce the production applications at XIV 2 to ensure that all application-consistent data
is copied to XIV 1.
11.Unmap master peers at XIV 2 from DR servers.
12.For asynchronous mirroring, monitor completion of the sync job and change the replication
interval to never.
13.Monitor to ensure that no more data is flowing from XIV 2 to XIV 1.
14.You can do a switch roles, which simultaneously changes the role of the peers at XIV 1
from slave to master and changes the role of the peers at XIV 2 from master to slave.
15.For asynchronous mirroring, change the replication schedule to the desired interval.
16.Map master peers at XIV 1 to the production servers.
17.Bring master peers online to XIV 1 production servers.
18.Change the designation of the master peer at XIV 1 to primary.
19.Change the designation of the slave peer at XIV 2 to secondary.

108 IBM XIV Storage System: Copy Services and Migration

4.5.3 Using an extra copy for DR
This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2.
1. Create a Snapshot or volume copy of the consistent data at XIV 2. (The procedure is
slightly different for XIV synchronous mirroring and XIV asynchronous mirroring. For
asynchronous mirroring, consistent data is on the last replicated snapshot.)
2. Unlock the snapshot or volume copy.
3. Map the snapshot/volume copy to DR servers at XIV 2.
4. Bring the snapshot/volume copy at XIV 2 online to DR servers to begin disaster recovery
testing at XIV 2.
5. When DR testing is complete, unmap the snapshot/volume copy from XIV 2 DR servers.
6. Delete the snapshot/volume copy if desired.

4.5.4 Creating application-consistent data at both local and the remote sites
This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2. This
scenario may be used when the fastest possible application restart is required.
1. No actions are taken to change XIV remote mirroring.
2. Briefly quiesce the application at XIV 1 or place the database into hot backup mode.
3. Ensure that all data has been copied from the master peer at XIV 1 to the slave peer at
XIV 2.
4. Issue Create Mirrored Snapshot at the master peer. This creates an additional snapshot
at the master and slave.
5. Resume normal operation of the application or database at XIV 1.
6. Unlock the snapshot or volume copy.
7. Map the snapshot/volume copy to DR servers at XIV 2.
8. Bring the snapshot or volume copy at XIV 2 online to XIV 2 servers to begin disaster
recovery testing or other functions at XIV 2.
9. When DR testing or other use is complete, unmap the snapshot/volume copy from XIV 2
DR servers.
10.Delete the snapshot/volume copy if desired.

4.5.5 Migration
A migration scenario involves a one-time movement of data from one XIV system to another
(for example, migration to new XIV hardware.) This scenario begins with existing connectivity
between XIV 1 and XIV 2.
1. Define XIV remote mirroring from the master volume at XIV 1 to the slave volume at XIV 2.
2. Activate XIV remote mirroring from the master volume at XIV 1 to the slave volume at XIV
2.
3. Monitor initialization until it is complete.
4. Deactivate XIV remote mirroring from the master volume at XIV 1 to the slave volume at
XIV 2.
5. Delete XIV remote mirroring from the master volume at XIV 1 to the slave volume at XIV 2.

Chapter 4. Remote Mirroring 109

 6. Remove connectivity between the XIV systems at XIV 1 and XIV 2.
7. Redeploy the XIV system at XIV 1 if desired.

4.5.6 Adding data corruption protection to disaster recovery protection

This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2
followed by creation of an additional snapshot of the master volume at XIV 1 to be used in the
event of application data corruption. To create a dependent-write consistent snapshot, no
changes are required to XIV remote mirroring.
1. Periodically issue Create Mirrored Snapshot at the master peer. This creates an additional
snapshot at the master and slave.
2. When production data corruption is discovered, quiesce the application and take any
steps necessary to prepare the application to be restored.
3. Deactivate and delete mirroring.
4. Restore production volumes from the appropriate snapshots.
5. Bring production volumes online and begin production access.
6. Remove remote volumes from the consistency group.
7. Delete or format remote volumes.
8. Delete any mirroring snapshots existing at the production site.
9. Remove production volumes from the consistency group.
10.Define and activate mirroring. Initialization results in a full copy of data.

If an application-consistent snapshot is desired, the following alternative procedure is used:

1. Periodically quiesce the application (or place into hot backup mode).
2. Create a snapshot of the production data at XIV 1. (The procedure may be slightly
different for XIV synchronous mirroring and XIV asynchronous mirroring. For
asynchronous mirroring, a duplicate snapshot or a volume copy of the last replicated
snapshot may be used.)
3. As soon as the snapshot or volume copy relationship has been created, resume normal
operation of the application.
4. When production data corruption is discovered, deactivate mirroring.
5. Remove master peers from the consistency group at XIV 1 if necessary. (Slave peers will
be automatically removed from the consistency group at XIV 2.)
6. Delete mirroring.
7. Restore the production volume from the snapshot or volume copy at XIV 1.
8. Delete any remaining mirroring-related snapshots or snapshot groups at XIV 1.
9. Delete secondary volumes at XIV 2.
10.Remove XIV 1 volumes (primary) from the consistency group.
11.Define remote mirroring peers from XIV 1 to XIV 2.
12.Activate remote mirroring peers from XIV 1 to XIV 2 (full copy is required).

110 IBM XIV Storage System: Copy Services and Migration

4.5.7 Communication failure
This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2
followed by a failure in the communication network used for XIV remote mirroring from XIV 1
to XIV 2.
1. No action is required to change XIV remote mirroring.
2. When communication between the two XIV systems is not available, XIV remote mirroring
is automatically deactivated and changes to the master volume are recorded in metadata.
3. When communication between the XIV systems at XIV 1 and XIV 2 is restored, XIV
mirroring is automatically reactivated, resynchronizing changes from the master at XIV 1
to the slave at XIV 2.

4.5.8 Temporary deactivation and reactivation

This scenario begins with normal operation of XIV remote mirroring from XIV 1 to XIV 2,
followed by user deactivation of XIV remote mirroring for a period of time. This scenario may
be used to temporarily suspend XIV remote mirroring during a period of peak activity if there
is not enough bandwidth to handle the peak load or if the response time impact during peak
activity is unacceptable.
1. Deactivate XIV remote mirroring from the master volume at XIV 1 to the slave volume at
XIV 2. Changes to the master volume at XIV 1 will be recorded in metadata for
synchronous mirroring.
2. Wait until it is acceptable to reactivate mirroring.
3. Reactivate XIV remote mirroring from the master volume at XIV 1 to the slave volume at
XIV 2.

4.6 Planning
The most important planning considerations for XIV Remote Mirroring are those related to
ensuring availability and performance of the mirroring connections between XIV systems, as
well as the performance of the XIV systems. Planning for snapshot capacity usage is also
extremely important.

To optimize availability, XIV remote mirroring connections must be spread across multiple
ports on different adapter cards in different modules, and must be connected to different
networks.

To optimize capacity usage, the number and frequency of snapshots (both those required for
asynchronous replication and any additional user-initiated snapshots) and the workload
change rates must be carefully reviewed. If not enough information is available, a snapshot
area that is 30% of the pool size may be used as a starting point. Storage pool snapshot
usage thresholds must be set to trigger notification (for example, SNMP, e-mail, SMS) when
the snapshot area capacity reaches 50%, and snapshot usage must be monitored continually
to understand long-term snapshot capacity requirements.

Chapter 4. Remote Mirroring 111

4.7 Advantages of XIV mirroring
XIV remote mirroring provides all the functions typical of remote mirroring solutions in addition
to the following advantages:
򐂰 Both synchronous and asynchronous mirroring are supported on a single XIV system.
򐂰 XIV mirroring is supported for consistency groups and individual volumes and mirrored
volumes may be dynamically moved into and out of mirrored consistency groups.
򐂰 XIV mirroring is data aware. Only actual data is replicated.
򐂰 Synchronous mirroring automatically resynchronizes couplings when a connection
recovers after a network failure.
򐂰 Both FC and iSCSI protocols are supported, and both may be used to connect between
the same XIV systems.
򐂰 XIV mirroring provides an option to automatically create slave volumes.
򐂰 XIV allows user specification of initialization and resynchronization speed.

4.8 Mirroring events

The XIV system generates events for user actions, failures, and changes in mirroring status.
These events can be used to trigger SNMP traps and send e-mails or text messages.

Thresholds for RPO and for link disruption may be specified by the user and trigger an event
when the threshold is reached.

4.9 Mirroring statistics

The XIV system provides Remote Mirroring performance statistics via both the graphical user
interface (GUI) and the command-line interface (XCLI) using the mirror_statistics_get
command.

Performance statistics from the FC or IP network components are also extremely useful.

4.10 Boundaries
With Version 10.2, the XIV Storage System has the following boundaries or limits:
򐂰 Maximum remote systems: The maximum number of remote systems that can be
attached to a single primary is 16.
򐂰 Number of remote mirrors: The combined number of master and slave volumes (including
in mirrored CG) cannot exceed 512.
򐂰 Distance: Distance is only limited by the response time of the medium used. Use
asynchronous mirroring when the distance causes unacceptable delays to the host I/O in
synchronous mode.
򐂰 Consistency groups are supported within Remote Mirroring. The maximum number of
consistency groups is 256.

112 IBM XIV Storage System: Copy Services and Migration

 򐂰 Snapshots: Snapshots are allowed with either the primary or secondary volumes without
stopping the mirror. There are also special-purpose snapshots used in the mirroring
process. Space must be available in the storage pool for snapshots.
򐂰 Master and slave peers cannot be the target of a copy operation and cannot be restored
from a snapshot. Peers cannot be deleted or formatted without deleting the coupling first.
򐂰 Master volumes cannot be resized or renamed if the link is operational.

4.11 Using the GUI or XCLI for Remote Mirroring actions

This section illustrates Remote Mirroring definition actions through the GUI or XCLI.

4.11.1 Initial setup

When preparing to set up Remote Mirroring, take the following questions into consideration:
򐂰 Will the paths be configured via SAN or direct attach, FC or iSCSI?
򐂰 Is the desired port configured as an initiator or a target?
– The port 4 default configuration an initiator.
– Port 2 is suggested as the target port for remote mirror links.
– Ports can be changed if needed.
򐂰 How many pairs will be copied?
This is related to the bandwidth needed between sites.
򐂰 How many secondary machines will be used for a single primary?

Remote Mirroring can be set up on paths that are either direct or SAN attached via FC or
iSCSI protocols. For most disaster recovery solutions, the secondary system will be located at
a geographically remote site. The sites will be connected using either SAN connectivity with
Fibre Channel Protocol (FCP) or Ethernet with iSCSI. In certain cases, using direct connect
might be the option of choice if the machines are located near each other and could be used
for initialization before the target XIV Storage System is moved to the remote site.

Bandwidth considerations must be taken into account when planning the infrastructure to
support the Remote Mirroring implementation. Knowing when the peak write rate occurs for
systems attached to the storage will help with the planning for the number of paths needed to
support the Remote Mirroring function and any future growth plans.

When the protocol has been selected, it is time to determine which ports on the XIV Storage
System will be used. The port settings are easily displayed using the XCLI Session
environment and the command fc_port_list for Fibre Channel or ipinterface_list for
iSCSI.

There must always be a minimum of two paths configured within Remote Mirroring for FCP
connections, and these paths must be dedicated to Remote Mirroring. These two paths must
be considered a set. Use port 4 and port 2 in the selected interface module for this purpose.
For redundancy, additional sets of paths must be configured in different interface modules.

Fibre Channel paths for Remote Mirroring have slightly more requirements for setup, and we
look at this interface first.

Chapter 4. Remote Mirroring 113

 As seen in Example 4-1, in the Role column each Fibre Channel port is identified as either a
target or an initiator. Simply put, a target in a Remote Mirror configuration is the port that will
be receiving data from the other system, whereas an initiator is the port that will be doing the
sending of data. In this example, there are three initiators configured. Initiators, by default, are
configured on FC:X:4 (X is the module number). In this highlighted example, port 4 in module
6 is configured as the initiator.

Example 4-1 The fc_port_list output command

>> fc_port_list
Component ID Status Currently Functioning WWPN Port ID Role
1:FC_Port:4:1 OK yes 5001738000130140 00030A00 Target
1:FC_Port:4:2 OK yes 5001738000130141 0075002E Target
1:FC_Port:4:3 OK yes 5001738000130142 00750029 Target
1:FC_Port:4:4 OK yes 5001738000130143 00750027 Initiator
1:FC_Port:5:1 OK yes 5001738000130150 00611000 Target
1:FC_Port:5:2 OK yes 5001738000130151 0075001F Target
1:FC_Port:5:3 OK yes 5001738000130152 00021D00 Target
1:FC_Port:5:4 OK yes 5001738000130153 00000000 Initiator
1:FC_Port:6:1 OK yes 5001738000130160 00070A00 Target
1:FC_Port:6:2 OK yes 5001738000130161 006D0713 Target
1:FC_Port:6:3 OK yes 5001738000130162 00000000 Target
1:FC_Port:6:4 OK yes 5001738000130163 0075002F Initiator
1:FC_Port:9:1 OK yes 5001738000130190 00DDEE02 Target
1:FC_Port:9:2 OK yes 5001738000130191 00FFFFFF Target
1:FC_Port:9:3 OK yes 5001738000130192 00021700 Target
1:FC_Port:9:4 OK yes 5001738000130193 00021600 Initiator
1:FC_Port:8:1 OK yes 5001738000130180 00060219 Target
1:FC_Port:8:2 OK yes 5001738000130181 00021C00 Target
1:FC_Port:8:3 OK yes 5001738000130182 002D0027 Target
1:FC_Port:8:4 OK yes 5001738000130183 002D0026 Initiator
1:FC_Port:7:1 OK yes 5001738000130170 006B0F00 Target
1:FC_Port:7:2 OK yes 5001738000130171 00681813 Target
1:FC_Port:7:3 OK yes 5001738000130172 00021F00 Target
1:FC_Port:7:4 OK yes 5001738000130173 00021E00 Initiator
>>

The iSCSI connections are shown in Example 4-2 using the command ipinterface_list.
The output has been truncated to show just the iSCSI connections in which we are interested
here. The command also displays all Ethernet connections and settings. In this example we
have two connections displayed for iSCSI—one connection in module 7 and one connection
in module 8.

Example 4-2 The ipinterface_list command

>> ipinterface_list
Name Type IP Address Network Mask Default Gateway MTU Module Ports
itso_m8_p1 iSCSI 9.11.237.156 255.255.254.0 9.11.236.1 4500 1:Module:8 1
itso_m7_p1 iSCSI 9.11.237.155 255.255.254.0 9.11.236.1 4500 1:Module:7 1

114 IBM XIV Storage System: Copy Services and Migration

Alternatively, a single port can be queried by selecting a system in the GUI, followed by
selecting Mirror Connectivity (Figure 4-45).

Figure 4-45 Selecting Mirror Connectivity

Click the connecting links between the systems of interest to view the ports.

Right-click a specific port and select Properties, the output of which is shown in Figure 4-46.
This particular port is configured as a target.

Figure 4-46 Port properties displayed with GUI

Another way to query the port configuration is to select the desired system, click the curved
arrow (at the bottom right of the window) to display the ports on the back of the system, and

Chapter 4. Remote Mirroring 115

 hover the mouse over a port, as shown in Figure 4-47. This view displays all the information
that is shown in Figure 4-46 on page 115.

Figure 4-47 Port information from the patch panel view

Similar information can be displayed for the iSCSI connections using the GUI, as shown in
Figure 4-48. This view can be seen either by right-clicking the Ethernet port (similar to the
Fibre Channel port shown in Figure 4-47) or by selecting the system, then selecting Hosts
and LUNs  iSCSI Connectivity. This sequence displays the same two iSCSI definitions
that are shown with the XCLI command.

Figure 4-48 iSCSI connectivity

By default, Fibre Channel ports 2 and 4 (target and initiator, respectively) from every module
are designed to be used for Remote Mirroring. For example, port 4 module 8 (initiator) on the
local machine is connected to port 2 module 8 (target) on the remote machine. When setting
up a new system, it is best to plan for any Remote Mirroring and reserve these ports for that
purpose. However different ports could be used as needed.

116 IBM XIV Storage System: Copy Services and Migration

In the event that a port role does need to be changed, you can change the port role with both
the XCLI and the GUI. Use the XCLI fc_port_config command to change a port, as shown in
Example 4-3. Using the output from fc_port_list, we can get the fc_port name to be used in
the command, changing the port role to be either initiator or target, as needed.

Example 4-3 XCLI command to configure a port

fc_port_config fc_port=1:FC_Port:4:3 role=initiator
Command completed successfully

fc_port_list
Component ID Status Currently Functioning WWPN Port ID Role
1:FC_Port:4:3 OK yes 5001738000130142 00750029 Initiator

To perform the same function with the GUI, select the primary system, open the patch panel
view, and right-click the port, as shown in Figure 4-49.

Figure 4-49 Configure ports

Chapter 4. Remote Mirroring 117

 Selecting Configure opens a configuration window, as shown in Figure 4-50, which allows
the port to be enabled (or disabled), its role defined as target or initiator, and, finally, the
speed for the port configured (Auto, 1 Gbps, 2 Gbps, or 10 Gbps).

Figure 4-50 Configure port with GUI

Planning for Remote Mirroring is important when determining how many copy pairs will exist.
All volumes defined in the system can be mirrored. A single primary system is limited to a
maximum of 16 secondary systems. Volumes cannot be part of an XIV data migration and a
remote mirror volume at the same time. Data migration information can be found in Chapter 7,
“Data migration” on page 185.

4.11.2 Remote mirror target configuration

The connections to the target (secondary) XIV system must be defined. We assume that the
physical connections and zoning have been set up. Target configuration is done from the
mirror connectivity menu. The first step is to add the target system. To do this right-click the
system image and select Create Target, as shown in Figure 4-51.

Figure 4-51 Create target

118 IBM XIV Storage System: Copy Services and Migration

Then define the type of mirroring to be used (mirroring or migration) and the type of
connection (iSCSI or FC), as shown in Figure 4-52.

Figure 4-52 Target type and protocol

Repeat the same process to define the local (primary) XIV system as a target for the
secondary XIV system.

Next, as shown in Figure 4-53, connections are defined by clicking the line between the two
XIV systems to display the link status detail screen.

Figure 4-53 Define connections

Chapter 4. Remote Mirroring 119

 Connections are easily defined by clicking Show Auto Detected Connections. This shows
the possible connections and provides an Approve button to define the detected connections.
Remember that for FCP ports an initiator must be connected to a target and the proper zoning
must be established for the connections to be successful. The possible connections are
shown in light grey, as depicted in Figure 4-54.

Figure 4-54 Show possible connections

Connections can also be defined by clicking a port on the primary system and dragging the
the corresponding port on the target system. This is shown as a blue line in Figure 4-55.

Figure 4-55 Graphically define a connection

Releasing the mouse button initiates the connection and then the status can be displayed, as
shown in Figure 4-56.

Figure 4-56 Define connection and view status

120 IBM XIV Storage System: Copy Services and Migration

Right-click a path and you have options to Activate, Deactivate, and Delete the selected path,
as shown in Figure 4-57.

Figure 4-57 Paths actions menu

Deleting the connections between two XIV systems is done from the Mirror Connectivity
display. Right-click the connecting links and select Delete, as illustrated in Figure 4-58.

Figure 4-58 Delete mirror connections

Chapter 4. Remote Mirroring 121

4.11.3 XCLI examples
XCLI commands can be used to configure connectivity between the primary XIV system and
the target or secondary XIV system (Figure 4-59).

target_define target="WSC_1300331" protocol=FC xiv_features=yes

target_mirroring_allow target="WSC_1300331"
target_define target="WSC_6000639" system_id=639 protocol=FC xiv_features=yes
target_mirroring_allow target="WSC_6000639"

target_port_add fcaddress=50017380014B0183 target="WSC_1300331"

target_port_add fcaddress=50017380027F0180 target="WSC_6000639"
target_port_add fcaddress=50017380014B0193 target="WSC_1300331"
target_port_add fcaddress=50017380027F0190 target="WSC_6000639"
target_port_add fcaddress=50017380027F0183 target="WSC_6000639"
target_port_add fcaddress=50017380014B0181 target="WSC_1300331"
target_connectivity_define local_port="1:FC_Port:8:4"
fcaddress=50017380014B0181 target="WSC_1300331"
target_port_add fcaddress=50017380027F0193 target="WSC_6000639"
target_port_add fcaddress=50017380014B0191 target="WSC_1300331"
target_connectivity_define local_port="1:FC_Port:9:4"
fcaddress=50017380014B0191 target="WSC_1300331"
target_connectivity_define target="WSC_6000639" local_port="1:FC_Port:8:4"
fcaddress="50017380027F0180"
target_connectivity_define target="WSC_6000639" local_port="1:FC_Port:9:4"
fcaddress="50017380027F0190"
Figure 4-59 Define target XCLI commands

XCLI commands can also be used to delete the connectivity between the primary XIV System
and the secondary XIV system (Figure 4-60).

target_connectivity_delete local_port="1:FC_Port:8:4"
fcaddress=50017380014B0181 target="WSC_1300331"
target_port_delete fcaddress=50017380014B0181 target="WSC_1300331"
target_connectivity_delete local_port="1:FC_Port:8:4"
fcaddress=50017380027F0180 target="WSC_6000639"
target_port_delete fcaddress=50017380027F0180 target="WSC_6000639"
target_connectivity_delete local_port="1:FC_Port:9:4"
fcaddress=50017380014B0191 target="WSC_1300331"
target_port_delete fcaddress=50017380014B0191 target="WSC_1300331"
target_connectivity_delete local_port="1:FC_Port:9:4"
fcaddress=50017380027F0190 target="WSC_6000639"

target_port_delete fcaddress=50017380027F0190 target="WSC_6000639"

target_port_delete target="WSC_6000639" fcaddress="50017380027F0183"
target_port_delete target="WSC_6000639" fcaddress="50017380027F0193"
target_delete target="WSC_6000639"
target_port_delete target="WSC_1300331" fcaddress="50017380014B0183"
target_port_delete target="WSC_1300331" fcaddress="50017380014B0193"
target_delete target="WSC_1300331"
Figure 4-60 Delete target XCLI commands

122 IBM XIV Storage System: Copy Services and Migration

4.12 Configuring Remote Mirroring
Configuration tasks differ depending on the nature of the coupling. Synchronous and
asynchronous mirroring are the two types of coupling supported. Refer to Chapter 5,
“Synchronous Remote Mirroring” on page 125, for specific configuration tasks related to
synchronous mirroring and Chapter 6, “Asynchronous Remote Mirroring” on page 149, for
specific configuration tasks related to asynchronous mirroring.

Chapter 4. Remote Mirroring 123

124 IBM XIV Storage System: Copy Services and Migration
 5

Chapter 5. Synchronous Remote Mirroring

This chapter describes the features of synchronous remote mirroring, the options that are
available, and procedures for setting it up and recovering from a disaster.

Note: GUI and XCLI illustrations included in this chapter were created with an early
version of the 10.2.1 code, available at the time of writing. There could be minor
differences with the code that was publicly released.

© Copyright IBM Corp. 2010. All rights reserved. 125

5.1 Synchronous mirroring configuration
The mirroring configuration process involves configuring volumes or CGs. When a pair of
volumes or consistency groups point to each other, it is referred to as a coupling.

We assume that the links between the local and remote XIV storage systems have already
been established, as discussed in 4.11.2, “Remote mirror target configuration” on page 118.

5.1.1 Volume mirroring setup and activation

Volumes or consistency groups that participate in mirror operations are configured in pairs.
These pairs are called peers. One peer is the source of the data to be replicated and the other
is the target. The source has the role of master and is the controlling entity in the mirror. The
target has the role of slave, which is controlled by operations performed by the master.

When initially configured, one volume is considered the source (master role and resides at
the primary system) and the other is the target (slave role and resides at the secondary
system). This designation is associated with the volume and its XIV system and does not
change. During various operations the role may change (master or slave), but one system is
always the primary and the other is always the secondary.

To create a mirror you can use the XIV GUI or the XCLI.

Using the GUI for volume mirroring setup

In the GUI select the primary XIV and select Remote Mirroring in the GUI, as shown in
Figure 5-1.

Figure 5-1 Selecting Remote Mirroring

To create a mirror:
1. Select Create Mirror, as shown in Figure 5-2, and specify the source volume or master for
the mirror pair (Figure 5-3 on page 127).

Figure 5-2 Selecting Create Mirror

There are also other way to create a mirror pair. If you are in the Volumes and Snapshots
list panel you can right-click a volume and select Create Mirror there.

126 IBM XIV Storage System: Copy Services and Migration

 The Create Mirror dialogue box is displayed (Figure 5-3).

Figure 5-3 Create Mirror parameters

2. Complete the following information:

– Sync Type
Select Sync as the sync type if for synchronous mirroring. (We discuss asynchronous
mirroring in Chapter 6, “Asynchronous Remote Mirroring” on page 149.)
– Master CG / Volume
This is the volume or consistency group at the local site to be mirrored. You can select
the volume or consistency groups from a list. The consistency groups are shown in
bold and they are at the end of the list.
– Target System
This is the XIV at the remote site that will contain the slave volumes. You can select the
target system from a list of known targets.
– Create slave
If you have not yet created target volumes at the remote XIV, you can check mark the
Create Slave option. In this case you also must select the storage pool where the
volume must be created at the target XIV. This pool must already exit at the target XIV.
The remote XIV will automatically create a target volume of the same size as the
source volume.
If selected, the slave volume will be created automatically. If left unselected you must
create the volume manually.
If you specified a consistency group instead of a volume this option is not available. A
slave consistency group must already exist at the remote site.
– Slave Pool
This is the storage pool on the XIV at the remote site that will contain the mirrored slave
volumes. This pool must already exit. This option is only available if you check marked
the Create Slave option.

Chapter 5. Synchronous Remote Mirroring 127

 – Slave CG / Volume
This is the name of the slave volume or consistency group. If you selected the Create
Slave option, the default is to use the same name as the source, but this can be
changed. If you did not check mark the Create Slave option, you can select the target
volume or a consistency group from a list.
If a target volume already exists in the remote XIV it must have exactly the same size
as the source volume, otherwise a mirror cannot be set up. In this case use the Resize
function of the XIV to adjust the capacity of the target volume to match the capacity of
the source volume.
Once mirroring is active, you can resize the source volume and the target volume will
be automatically resized accordingly.
3. Once all the appropriate entries have been completed, click Create.
A coupling is created and is in standby (inactive) mode, as shown in Figure 5-4. In this
state data is not yet copied from the source to the target volume.

Figure 5-4 Coupling on the primary XIV in standby (Inactive) mode

A corresponding coupling is automatically created on the secondary XIV, and it is also in

standby (Inactive) mode, as shown in Figure 5-5.

Figure 5-5 Coupling on the secondary XIV in standby (Inactive) mode

Repeat steps 1–3 to create additional couplings.

Using XCLI for volume mirroring setup

Tip: When working with the XCLI session or the XCLI command the windows look the
same and you could address the wrong XIV system with your command. Therefore, it
might be good to always first issue a config_get command to verify that you are talking to
the right XIV system.

To do this:
1. Open an XCLI session on the XIV at the local site (primary XIV) and run the
mirror_create command (Example 5-1).

Example 5-1 Create remote mirror coupling

>> mirror_create target="XIV MN00035" vol="itso_win2008_vol2"
slave_vol="itso_win2008_vol2" remote_pool="test_pool" create_slave=yes
Command executed successfully.

128 IBM XIV Storage System: Copy Services and Migration

 2. On the primary XIV, to list the couplings run the mirror_list command (Example 5-2).
Note the status of Initializing is used when the coupling is in standby (inactive) or is
initializing.

Example 5-2 Listing mirror couplings

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol1 no Initializing yes
itso_win2008_vol2 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol2 no Initializing yes

3. On the secondary XIV, to list the couplings run the mirror_list command, as shown in
Example 5-3. Note that the status of Initializing is used when the coupling is in standby
(inactive) or initializing.

Example 5-3 Newly created slave volumes

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol1 no Initializing yes
itso_win2008_vol2 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol2 no Initializing yes

Repeat steps 1–3 to create additional mirror couplings.

Activating the remote mirror coupling using the GUI

To activate the mirror, proceed as follows:
1. On the Primary XIV go the Remote Mirroring menu and highlight all the couplings that you
want to activate, right-click, and select Activate, as shown in Figure 5-6.

Figure 5-6 Activating a mirror coupling

Figure 5-7 shows the three couplings (itso_win2008_vol1, itso_win2008_vol2, and

itso_win2008_vol3) in various states.

Figure 5-7 Remote mirroring statuses on the primary XIV

Chapter 5. Synchronous Remote Mirroring 129

 2. On the Secondary XIV go to the Remote Mirroring menu to see the statuses of the
couplings (Figure 5-8). Note that due to the time lapse between Figure 5-7 on page 129
and Figure 5-8 being taken they do show different statuses.

Figure 5-8 Remote mirroring statuses on the secondary XIV

3. Repeat steps 1–2 until all required couplings are activated and are
synchronized/consistent.

Activating the remote mirror coupling using the XCLI

Proceed as follows:
1. On the primary XIV run the mirror_activate command (Example 5-4).

Example 5-4 Activating the mirror coupling

>> mirror_activate vol=itso_win2008_vol3
Command executed successfully.

2. On the primary XIV run the mirror_list command to see the status of the couplings
(Example 5-5).

Example 5-5 List remote mirror statuses on the primary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol1 yes Synchronized yes
itso_win2008_vol2 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol2 yes Synchronized yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol3 yes Synchronized yes

3. On the secondary XIV run the mirror_list command to see the status of the couplings
(Example 5-6).

Example 5-6 List remote mirror statuses on the secondary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol1 yes Consistent yes
itso_win2008_vol2 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol2 yes Consistent yes
itso_win2008_vol3 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol3 yes Consistent yes

4. Repeat steps 1–3 to activate additional couplings.

5.1.2 Consistency group setup and configuration

IBM XIV Storage System leverages its consistency group capability to allow for mirroring
numerous volumes at once.

Setting a consistency group to be mirrored is done by first creating a consistency group, then
setting it to be mirrored, and only then populating it with volumes. A consistency group must
be created at the primary XIV and a corresponding consistency group at the secondary XIV.
The names of the consistency groups can be different. When creating a consistency group,
you also must specify the storage pool.

130 IBM XIV Storage System: Copy Services and Migration

To create a mirrored consistency group first create a CG on the primary and secondary XIV
Storage System. Then select the CG at the primary and specify Create Mirror, as shown in
Figure 5-9.

Figure 5-9 Create Mirror CG

The Create Mirror dialog shown in Figure 5-10 is displayed. Be sure to specify the mirroring
parameters that match the volumes that will be part of that CG.

Figure 5-10 Sync mirrored CG

Now you can add mirrored volumes to this consistency group.

All volumes that you are going to add to the consistency group must be in that pool on the
primary XIV and in one pool at the secondary XIV. Adding a new volume pair to a mirrored
consistency group requires the volumes to be mirrored exactly as the other volumes within
this consistency group.

Important: All volumes that you want to add to a mirroring consistency group must be
defined in the same pool at the primary site and must be in one pool at the secondary site.

Chapter 5. Synchronous Remote Mirroring 131

 Adding a mirrored volume to a mirrored CG
Adding a volume to a CG requires that:
򐂰 The volume is on the same system as the consistency group.
򐂰 The volume belongs to the same storage pool as the consistency group.
򐂰 The command must be issued only on the master CG.
򐂰 The command must not be run during initialization of volume or CG.
򐂰 The volume mirroring settings must be identical to those of the CG:
– Mirroring type
– Mirroring role
– Mirroring status
– Mirroring target
– Target pool

Also, mirrors for volumes must be activated before volumes can be added to a mirrored
consistency group.

It is possible to add a mirrored volume to a non-mirrored consistency group and have this
volume retain its mirroring settings.

Removing a volume from a mirrored consistency group

When removing a volume from a mirrored consistency group, the corresponding peer volume
will be removed from the peer consistency group. Mirroring is retained with the same
configuration as the consistency group from which it was removed.

Synchronous mirroring and snapshot consistency group

A volume can be in only one consistency group. Because consistency groups can be used for
snapshot (see 1.3, “Snapshots consistency group” on page 23) and Remote Mirroring,
confusion can arise. Define separate and specific CG for snapshot and Remote Mirroring.

5.1.3 Coupling activation, deactivation, and deletion

Mirroring can be manually activated and deactivated per volume or CG pair. When it is
activated, the mirror is in active mode. When it is deactivated, the mirror is in inactive mode.
These modes have the following functions:
򐂰 Active
Mirroring is functioning and the data is being written to both the master and the slave
peers.
򐂰 Inactive
Mirroring is deactivated. The data is not being written to the slave peer, but writes to the
master volume are being recorded and can later be synchronized with the slave volume.
Inactive mode is used mainly when maintenance is performed at the secondary site or on
the secondary XIV. In this mode, the slave volumes do not generate alerts that the
mirroring has failed.

132 IBM XIV Storage System: Copy Services and Migration

 The mirror has the following characteristics:
򐂰 When a mirror is created, it is always initially in inactive mode.
򐂰 A Mirror can only be deleted when its is in inactive mode.
򐂰 Transitions between the two states can only be performed from the XIV with the master.
򐂰 In a DR situation, a role change changes the slave peers (at the secondary system) to a
master role (so that production can resume at the secondary). When the primary site is
recovered, and before the link is resumed, the master must be changed to a slave
(change_role).

Deletion
When a mirror pair (volume or consistency group) is inactive, the mirror relationship can be
deleted. When a mirror relationship has been deleted, the XIV forgets everything about the
relation. If you want to set up the mirror again, the XIV must do an initial copy again from the
source to the target.

Note that when the mirror is deleted, the slave volume becomes a normal volume again, but
the volume is locked, which means that it is write protected. To enable writing to the volume
go to the Volumes list panel. Right-click the volume and select Unlock.

The slave volume must also be formatted before it can be part of a new mirror. Formatting
also requires that all snapshots of that volume be deleted.

5.2 Disaster recovery

There are two broad categories of disaster, one that destroys the primary (local) site or the
data there and one that makes the primary site or the data there unavailable but that leaves
the data intact. However, within these broad categories there are a number of situations that
may exist. Some of these and the recovery procedures are considered below:
򐂰 A disaster that makes the XIV at the primary site unavailable but the site itself and the
servers there are still available
In this scenario the volumes/CG on the XIV at the secondary site can be switched to
master volumes/CG, servers at the primary site can be redirected to the XIV at the
secondary site, and normal operations can start again. When the XIV at the primary site is
recovered the data can be mirrored from the secondary site back to the primary site.
When it is synchronized the peer roles can be switched back to the master at the primary
site and the slave at the secondary site and the servers redirected back to the local site.
򐂰 A disaster that makes the entire primary site and data unavailable
In this scenario, the standby (inactive) servers at the secondary site are activated and
attached to the secondary XIV to continue normal operations. This requires changing the
role of the slave peers to become master peers.
After the primary site is recovered, the data at the secondary site can be mirrored back to
the primary site to become synchronized once again. If desired, a planned site switch can
then take place to resume production activities at the primary site. See 5.3, “Role reversal”
on page 134, for details related to this process.
򐂰 A disaster that breaks all links between the two sites but both site remain running
In this scenario the primary site continues to operate as normal. When the links are
reestablished the data at the primary site can be resynchronized with the secondary site.
See 5.4, “Resynchronization after link failure” on page 136, for more details.

Chapter 5. Synchronous Remote Mirroring 133

5.3 Role reversal
With synchronous mirroring, roles can be modified by either switching or changing.

Switching roles must be initiated on the master volume or consistency group when remote
mirroring is operational. As the task name implies, it switches the master role to the slave role
and at the same time the slave role to the master role.

Changing roles can be performed at any time (when a pair is active or inactive) for the slave,
and for the master when the coupling is inactive. A change role reverts only the role of that
peer.

5.3.1 Switching roles

Switching roles exchanges the roles of master and slave volumes or CGs. It can be
performed after the remote mirroring function is in operation and the pair is synchronized.
After switching roles, the master volume or CG becomes the slave volume or CG and vice
versa. There are two typical reasons for switching roles. These are:
򐂰 Drills/DR tests
Drills can be performed to test the functioning of the secondary site. In a drill, an
administrator simulates a disaster and tests that all procedures are operating smoothly
and that documentation is accurate.
򐂰 Scheduled maintenance
To perform maintenance at the primary site, operations can be switched to the secondary
site some time before the maintenance. This switchover cannot be performed if the master
and slave volumes or CG are not synchronized/consistent.

Normally, switching the roles requires shutting down the servers at the primary site first,
changing SAN zoning and XIV LUN masking to allow access to the secondary site volumes,
and then restarting the servers with access to the secondary (remote) XIV. However, in
certain clustered environments this takeover could be automated.

5.3.2 Change role

In a disaster at the primary site, a role change at the secondary site is the normal recovery
action.

Assuming that the primary site is down and the secondary site will become the main
production site, changing roles is performed at the remote (now production) site first. Later,
when the primary site is up again and communication is reestablished you also change the
role at the primary site to a slave to be able to establish remote mirroring from the secondary
site back to the normal production (primary) site.

134 IBM XIV Storage System: Copy Services and Migration

Changing the slave peer role
The role of the slave volume or consistency group can be changed to the master role, as
shown in Figure 5-11. After this changeover, the following is true:
򐂰 The slave volume or consistency group is now the master.
򐂰 The coupling has the status of unsynchronized.
򐂰 The coupling remains inactive, meaning that the remote mirroring is deactivated. This
ensures an orderly activation when the role of the peer on the other site is changed.

Figure 5-11 Change role of a slave consistency group

The new master volume or consistency group (at the secondary site) starts to accept write
commands from local hosts. Because coupling is not active, in the same way as for any
master volume, metadata maintain a record of which write operations must be sent to the
slave volume when communication resumes.

After changing the slave to the master, an administrator must change the original master to
the slave role before communication resumes. If both peers are left with the same role
(master), mirroring cannot be restarted.

Slave peer consistency

When the user is changing the slave volume or consistency group to a master volume or
master consistency group and a snapshot of the last consistent state exists that was
produced during the process of resynchronizing (as a result of a broken link, for instance), the
system reverts the slave to the last consistent snapshot.

Changing the master peer role

When coupling is inactive, the master volume or consistency group can change roles. After
such a change the master volume or consistency group becomes the slave volume or
consistency group.

Unsynchronized master becoming a slave volume or consistency group

When a master volume (or consistency group) is inactive, it is also in an unsynchronized
state, and it might have a backlog of uncommitted data. The uncommitted changes will
potentially be lost when the volume (CG) becomes a slave volume (CG), as this data must be
reverted to match the data on the peer volume, which is now the new master volume. In this
case, an event is created, summarizing the size of the changes that were lost. The
uncommitted data has now switched its semantics, and instead of representing updates that
the primary peer (former master, now slave) needs to update on the secondary peer (old
slave, new master), metadata now represents updates that must be replicated from the
secondary to the primary.

Upon re-establishing the connection, the primary volume or consistency group (current slave
volume/CG) updates the secondary volume/CG (new master volume/CG) with this

Chapter 5. Synchronous Remote Mirroring 135

 uncommitted data, and it is the responsibility of the secondary peer to synchronize these
updates to primary peer.

Reconnection when both sides have the same role

Situations where both sides are configured to the same role can only occur when one side
was changed. The roles must be changed to have one master and one slave (volume or
consistency group). Change the volume roles as appropriate on both sides before the link is
resumed.

If the link is resumed and both sides have the same role, the coupling will not become
operational. To solve this problem, the user must use the change role function on one of the
volumes and then activate the coupling.

5.4 Resynchronization after link failure

When synchronization between the peers has been interrupted, for example, by a failure of all
links or the pairs have been suspended by a command, you probably want to resume the
mirroring after the problems were solved.

Resynchronization can be performed in any direction given that one peer has the master role
and the other the slave role. When there is a temporary failure of all links from the primary XIV
to the secondary XIV, you re-establish the mirroring in the original direction after the links are
up again.

Also, if you suspended mirroring for a disaster recovery test at the secondary site, you might
want to reset the changes made to the secondary site during the tests and re-establish
mirroring from the primary to the secondary site.

If there was a disaster and production is now running on the secondary site, re-establish
mirroring first from the secondary site to the primary site and later on switch mirroring to the
original direction from the primary XIV to the secondary XIV.

In any case, the slave peers usually are in a consistent state up to the moment when
resynchronization starts. During the resynchronization process, the peers (volumes or
consistency group) are inconsistent. To preserve consistency, the XIV at the slave side
automatically creates a snapshot of the involved volumes or, in case of a consistency group, a
snapshot of the entire consistency group before transmitting any data to the slave volumes.

5.4.1 Last consistent snapshot

Before a resynchronization process is initiated, a snapshot of the slave volumes or
consistency groups is created. A snapshot is created to ensure the usability of the slave
volumes/CG in case of a primary site disaster during the resynchronization process. If the
master volume/CG is destroyed before resynchronization is completed, the slave volume/CG
might be inconsistent because it might have been only partially updated with the changes that
were made to the master volume. To handle this situation, the secondary XIV always creates
a snapshot of the last consistent slave volume/CG after reconnecting to the secondary XIV
and before starting the resynchronization process. No snapshot is created for couplings that
are in the initialization state. The snapshot is preserved until a volume pair is synchronized
again, or in case of remote mirror consistency groups, until all volumes of the consistency
group are synchronized.

136 IBM XIV Storage System: Copy Services and Migration

5.4.2 Last consistent snapshot timestamp
A timestamp is taken when the coupling between the master and slave volumes or
consistency group becomes non-operational. This timestamp specifies the last time that the
slave volumes/CG was consistent with the master (Figure 5-12).

If there is a disaster at the primary (master) site, the snapshot taken at the secondary site can
be used to restore the slave volumes to a consistent state, ready for production.

Important: You must delete the mirror relation at the secondary site before you can restore
the last consistent snapshot to the target volumes.

Figure 5-12 Snapshot during a resync

5.5 Synchronous mirror step-by-step scenario

In 5.1, “Synchronous mirroring configuration” on page 126, we explained the steps required to
set up, operate, and deactivate the mirror.

In this section, we go through a scenario to demonstrate synchronous mirroring. We assume

that all configuration has taken place for us to start configuring the remote mirroring
couplings. In particular, we assume that:
򐂰 A host server exists and has volumes assigned at the local site.
򐂰 Two XIV systems have been connected to each other over FC or iSCSI.
򐂰 A standby server exists at the remote site.

Note: When using the XCLI commands quotation marks (“ “) must be used to enclose
names that include spaces. If they are used for names without spaces the command still
works. The examples in this scenario contain a mixture of commands with and without
quotation marks.

This scenario discusses the following phases:

򐂰 Setup and configuration
Perform initial setup, activate coupling, write data to three volumes, and prove that the
data has been written and that the volumes are synchronized.
򐂰 Simulating a disaster at the local site
The link is broken between the two sites to simulate that the local site is unavailable, the
slave volumes are changed to master volumes, the standby server at the remote site has
LUNs mapped to it on the XIV at the remote site, and new data is written.

Chapter 5. Synchronous Remote Mirroring 137

 򐂰 The local site recovery
The old master volumes at the local site are changed to slave volumes and data is
mirrored back from the remote site to the local site.
򐂰 Failback to the local site
When the data is synchronized the volume roles are switched back to the original roles
(that is, master volumes at the local site and slave volumes at the remote site) and the
original production server (at the local site) is used.

5.5.1 Phase 1: setup and configuration

In our sample scenario, we have a Windows 2008 server with three LUNs at the local site and
communication has been configured between the XIVs at the local and remote sites.

After the couplings have been created and activated, as explained under 5.1, “Synchronous
mirroring configuration” on page 126, the environment will be as illustrated in Figure 5-13.

Local Remote
Site Site

Production
W indows 2008
Active Inactive Standby
W indows 2008
Server Server
Data Flow

FC Link

FC Link

Data Mirroring

FC Link

Primary Secondary
XIV XIV

Figure 5-13 Environment with remote mirroring activated

138 IBM XIV Storage System: Copy Services and Migration

5.5.2 Phase 2: disaster at local site
In this phase of the scenario we simulate a disaster at the local site. All communication has
been lost between the primary and secondary sites due to a complete power failure or a
disaster. This is depicted in Figure 5-14.

Local Remote
Site Site

Production Standby
Windows 2008 W indows 2008
Server Server
Data Flow

FC Link

FC Link

Primary Secondary
XIV XIV

Figure 5-14 Primary site disaster

Role changeover at the secondary site using the GUI

We now change roles for the slave volumes at the secondary site and make them master
volumes so that the standby server can write to them.
1. On the secondary XIV go to the Remote Mirroring menu and right-click a coupling and
select Change Role (Figure 5-15).

Figure 5-15 Remote mirror change role

Chapter 5. Synchronous Remote Mirroring 139

 Figure 5-15 on page 139 shows that the synchronization status is still consistent for the
couplings that are yet to be changed. This is because this is the last known state. When
the role is changed the coupling is automatically deactivated.

Role changeover at the secondary site using the XCLI

We now change roles for the slave volumes at the secondary site and make them master
volumes so that the standby server can write to them.
1. On the secondary XIV open an XCLI session and run the mirror_change_role command
(Example 5-7).

Example 5-7 Remote mirror change role

>> mirror_change_role vol=itso_win2008_vol2 new_role=master
Warning: ARE_YOU_SURE_YOU_WANT_TO_CHANGE_THE_PEER_ROLE_TO_MASTER Y/N: Y
Command executed successfully.

2. To view the status of the coupling run the mirror_list command, as shown in
Example 5-8.

Example 5-8 List mirror couplings

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol1 no Unsynchronized yes
itso_win2008_vol2 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol2 no Unsynchronized yes
itso_win2008_vol3 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol3 yes Consistent yes

Example 5-8 shows that the synchronization status is still consistent for one of the
couplings that is yet to be changed. This is because this reflects the last known state.
When the role is changed, the coupling is automatically deactivated.
3. Repeat steps 1–2 to change roles on other volumes.

140 IBM XIV Storage System: Copy Services and Migration

Map volumes on standby server and continue working
At this point we map the relevant mirrored volumes to the standby server. For details on how
to do this mapping refer to IBM XIV Storage System: Architecture, Implementation, and
Usage, SG24-7659. Once the volumes are mapped, we continue working as normal. This is
simulated by adding additional data to the server, as illustrated in Figure 5-16.

Figure 5-16 Additional data added to the standby server

Environment with production now at the secondary site

Figure 5-17 illustrates production at the secondary site.

Local Remote
Site Site

Production Standby
Windows 2008
Server
Active Windows 2008
Server
Data Flow

Data Flow

FC Link FC Link

Primary Secondary
XIV XIV

Figure 5-17 Production at secondary site

Chapter 5. Synchronous Remote Mirroring 141

5.5.3 Phase 3: recovery of the primary site
In this phase the primary site is recovered and communication between the primary and
secondary sites is restored. We assume that it was not totally damaged and that the data at
the local site is still there (so we can do a resync). Data is being written from the standby
server to the secondary XIV. At the primary site the original Windows 2008 production server
is now switched off, as illustrated in Figure 5-18.

Local Remote
Site Site

Production Standby
Windows 2008
Server
Down Active Windows 2008
Server

Data Flow
FC Link

FC Link

Mirroring Inactive

FC Link

Primary Secondary
XIV XIV

Figure 5-18 Primary site recovery

Role changeover at the primary site using the GUI

We are now going to change roles for the master volumes at the primary site and make them
slave volumes. Before doing this, ensure that the original production server is shut down.
1. On the primary XIV go to the Remote Mirroring menu. The synchronization status will
probably be inactive. Select one coupling (if you select several couplings, you cannot
change the role), right-click, and select Change Role, as shown in Figure 5-19.

Figure 5-19 Change master volumes to slave volumes on the primary XIV

142 IBM XIV Storage System: Copy Services and Migration

 2. You will be prompted to confirm the role change. Select OK to confirm (Figure 5-20).

Figure 5-20 Roll change confirmation

Once you have confirmed, the role is changed to slave, as shown in Figure 5-21.

Figure 5-21 New role as slave volume

3. Repeat steps 1–2 for all the volumes that must be changed.

Role changeover at the primary site using the XCLI

We now change roles for the master volumes at the primary site with the XCLI and make
them slave volumes. Before doing this, ensure that the original production server is shut
down.
1. On the primary XIV open an XCLI session and run the mirror_change_role command
(Example 5-9).

Example 5-9 Change master volumes to slave volumes on the primary XIV
>> mirror_change_role vol=itso_win2008_vol2
Warning: ARE_YOU_SURE_YOU_WANT_TO_CHANGE_THE_PEER_ROLE_TO_SLAVE Y/N: Y
Command executed successfully.

2. To view the status of the coupling run the mirror_list command, as shown in
Example 5-10.

Example 5-10 List mirror couplings

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Slave XIV MN00035 itso_win2008_vol1 no Inconsistent yes
itso_win2008_vol2 sync_best_effort Volume Slave XIV MN00035 itso_win2008_vol2 no Inconsistent yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol3 no Unsynchronized yes

3. Repeat steps 1–2 to change other couplings.

Chapter 5. Synchronous Remote Mirroring 143

 Reactivating the remote mirror coupling using the GUI
To reactivate the remote mirror coupling using the GUI:
1. On the secondary XIV go the Remote Mirroring menu and highlight all the couplings that
you want to activate. Right-click and select Activate, as illustrated in Figure 5-22 and
Figure 5-23.

Figure 5-22 Reactivating a mirror coupling

Figure 5-23 Synchronization status

2. On the primary XIV go to the Remote Mirroring menu to check the statuses of the
couplings (Figure 5-24). Note that due to the time lapse between Figure 5-23 and
Figure 5-24 being taken they do show different statuses.

Figure 5-24 Remote mirroring statuses on the secondary (local) XIV

3. Repeat steps 1–2 until all required couplings are reactivated and synchronized.

144 IBM XIV Storage System: Copy Services and Migration

 Reactivating the remote mirror coupling using the XCLI
To reactivate the remote mirror coupling using the XCLI:
1. On the secondary (remote) XIV run the mirror_activate command, as shown in
Example 5-11.

Example 5-11 Reactivating the mirror coupling

>> mirror_activate vol=itso_win2008_vol2
Command executed successfully.

2. On the secondary XIV run the mirror_list command to see the status of the couplings,
as illustrated in Example 5-12.

Example 5-12 List remote mirror statuses on the secondary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol1 yes Synchronized yes
itso_win2008_vol2 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol2 yes Synchronized yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol3 no Unsynchronized yes

3. On the primary XIV run the mirror_list command to see the status of the couplings, as
shown in Example 5-13.

Example 5-13 List remote mirror statuses on the primary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Slave XIV MN00035 itso_win2008_vol1 yes Consistent yes
itso_win2008_vol2 sync_best_effort Volume Slave XIV MN00035 itso_win2008_vol2 yes Consistent yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol3 no Unsynchronized yes

4. Repeat steps 1–3 to activate additional couplings.

Environment with remote mirroring reactivated

Figure 5-25 illustrates production at the secondary (remote) site.

Local Remote
Site Site

Production Standby
Windows 2008 Down Active Windows 2008
Server Server
Data Flow

FC Link

FC Link

Data Mirroring

FC Link

Primary Secondary
XIV XIV

Figure 5-25 Mirroring reactivated

Chapter 5. Synchronous Remote Mirroring 145

5.5.4 Phase 4: switching production back to the primary site
At this stage we have mirroring reactivated with production at the secondary site. We now
want to switch production back to the primary site. This involves doing the following:
򐂰 Shut down the servers.
򐂰 Switch peer roles.
򐂰 Switch from the standby server to the original production server.

Role switchover using the GUI

To switch over the role using the GUI:
1. At the secondary site, ensure that all the volumes for the standby server are synchronized
and shut down the servers.
2. On the secondary XIV go to the Remote Mirroring menu, highlight the required coupling,
and select Switch Roles (Figure 5-26).

Figure 5-26 Switch roles

3. You are prompted for confirmation. Select OK. Refer to Figure 5-27 and Figure 5-28 on
page 147.

Figure 5-27 Switch role confirmation

146 IBM XIV Storage System: Copy Services and Migration

 Figure 5-28 Switch role to slave volume on the secondary XIV

4. Go to the Remote Mirroring menu on the primary XIV and check the status of the coupling.
It must show the peer volume as a master volume (Figure 5-29).

Figure 5-29 Switch role to master volume on the primary XIV

5. Reassign volumes back to the production server at the primary site and power it on again.
Continue to work as normal. Figure 5-30 on page 148 shows that all the new data is now
back at the primary (local) site.

Role switchover using the XCLI

To switch over the role using the XCLI:
1. At the secondary site, ensure that all the volumes for the standby server are synchronized
and shut down the servers.
2. On the secondary XIV open an XCLI session and run the mirror_switch_roles command,
as shown in Example 5-14.

Example 5-14 Switch from master volume to slave volume on secondary XIV
>> mirror_switch_roles vol=itso_win2008_vol2
Command executed successfully.

3. On the secondary XIV, to list the mirror coupling run the mirror_list command
(Example 5-15).

Example 5-15 Mirror statuses on the secondary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol1 yes Consistent yes
itso_win2008_vol2 sync_best_effort Volume Slave XIV MN00019 itso_win2008_vol2 yes Consistent yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00019 itso_win2008_vol3 no Unsynchronized yes

4. On the primary XIV run the mirror_list command to list the mirror couplings, as shown
in Example 5-16.

Example 5-16 Mirror statuses on the primary XIV

>> mirror_list
Name Mirror Type Mirror Object Role Remote System Remote Peer Active Status Link Up
itso_win2008_vol1 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol1 yes Synchronized yes
itso_win2008_vol2 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol2 yes Synchronized yes
itso_win2008_vol3 sync_best_effort Volume Master XIV MN00035 itso_win2008_vol3 no Unsynchronized yes

Chapter 5. Synchronous Remote Mirroring 147

 5. Reassign volumes back to the production server at the primary (local) site and power it on
again. Continue to work as normal. Figure 5-30 shows that all the new data in now back at
the local site.

Figure 5-30 Production server with mirrored data reassigned at the local site

Environment back to its production state

The environment is now back to its production state with mirroring from the primary site to the
secondary site, as shown in Figure 5-31.

Local Remote
Site Site

Production Standby
W indows 2008
Server
Active Inactive W indows 2008
Server
Data Flow

FC Link

FC Link

Data Mirroring

FC Link

Primary Secondary
XIV XIV

Figure 5-31 Environment back to production state

148 IBM XIV Storage System: Copy Services and Migration

 6

Chapter 6. Asynchronous Remote Mirroring

This chapter describes the basic characteristics, options, and available interfaces for
asynchronous Remote Mirroring. It also includes step-by-step procedures for setting up and
removing the mirror.

Asynchronous mirroring is the volume or consistency group synchronization attained through

a periodic, recurring activity that takes a snapshot of a designated source and updates a
designated target with differences between that snapshot and the last replicated version of
the source. Unlike other implementations, XIV asynchronous mirroring supports multiple
consistency groups with different recovery point objectives. XIV asynchronous mirroring
supports multiple targets, 512 mirrored pairs, scheduling, event reporting, and statistics
collection.

Asynchronous mirroring enables replication between two XIV volumes or consistency groups
(CG) that does not suffer from the latency inherent to synchronous mirroring, thereby yielding
better system responsiveness and offering greater flexibility for implementing disaster
recovery solutions.

Note: GUI and XCLI illustrations included in this chapter were created with an early
version of the 10.2.1 code, available at the time of writing. There could be minor
differences with the code that was publicly released.

© Copyright IBM Corp. 2010. All rights reserved. 149

6.1 Asynchronous mirroring configuration
The mirroring configuration process involves configuring volumes and CGs. When a pair of
volumes or consistency groups point to each other, it is referred to as a mirror.

We assume that the links between the local and remote XIV storage systems have already
been established, as discussed in 4.11.2, “Remote mirror target configuration” on page 118.

6.1.1 Volume mirroring setup and activation

Volumes or consistency groups that participate in mirror operations are configured in pairs.
These pairs are called peers. One peer is the source of the data to be replicated and the other
is the target. The source has the role of master and is the controlling entity in the mirror. The
target has the role of slave, which is controlled by operations performed by the master.

When initially configured, one volume is considered the source (resides at the primary
system) and the other is the target (resides at the secondary system). This designation is
associated with the volume and its XIV system and does not change. During various
operations the role may change (master or slave) but one system is always the primary and
the other is always the secondary.

Asynchronous mirroring is initiated at defined intervals. This is the sync job schedule. A sync
job entails synchronization of data updates recorded on the master since the last successful
synchronization. The sync job schedule will be defined for both the primary and secondary
system peers in the mirror. This provides a schedule for each peer and will be used when the
peer takes on the role of master. The purpose of the schedule specification on the slave is to
set a default schedule for an automated failover scenario.

A schedule set as NEVER means that no sync jobs will be automatically scheduled. See 6.6,
“Detailed asynchronous mirroring process” on page 173.

The XIV GUI automatically creates schedules based on the RPO selected for the mirror being
created. The interval can be set in the Mirror Properties panel or must be explicitly specified
through the XCLI.

Tip: XIV allows you to set a specific RPO and schedule interval for each mirror coupling.

Slave volumes must be formatted before they are configured as part of a mirror. This means
that the volume must not have any snapshots and must be unlocked.

To create a mirror you can use the XIV GUI or the XCLI. Both methods are illustrated in the
following sections.

150 IBM XIV Storage System: Copy Services and Migration

Using the GUI for volume mirror setup
To create a mirror select the master peer (volume or CG) and select Create Mirror
(Figure 6-1).

Figure 6-1 Select volume to be mirrored

Then specify Sync Type as Async, select the slave peer (volume or CG), and specify an RPO
value. Set the Schedule Management field to XIV Internal to create automatic synchronization
using scheduled sync jobs, as shown in Figure 6-2.

Figure 6-2 Create Mirror

Chapter 6. Asynchronous Remote Mirroring 151

 The slave volume must be unlocked and created as formatted, which also means that it
cannot have any Snapshots.

When creating a mirror, the slave peer (volume or CG) can also be created automatically on
the target XIV System. To do this, select Create Slave and specify the slave pool name and
the slave volume or CG name, as shown in Figure 6-3.

Figure 6-3 Create Mirror and slave volume

If schedule type External is selected when creating a mirror, no sync jobs will run for this
mirror and the interval will be set to Never, as illustrated in Figure 6-4.

Figure 6-4 Mirror with external schedule

152 IBM XIV Storage System: Copy Services and Migration

When volumes are to be placed in a consistency group, they must all have the same mirroring
properties. Using the Mirror Properties panel, the Never interval can be changed to match the
other volumes created (Figure 6-5).

Figure 6-5 Mirror Properties

The Mirroring panel shows the current status of the mirrors. The synchronization of the mirror
must to be initiated manually using the Activate action, as seen in Figure 6-7 on page 155.

Notice, as shown in Figure 6-6, that the selected RPO is displayed for the mirrors created.

Figure 6-6 Mirroring status inactive

Note that mirrors of the sync type do not have an RPO value.

Chapter 6. Asynchronous Remote Mirroring 153

 Using XCLI for volume mirroring setup

Tip: When working with the XCLI session or the XCLI command the windows look the
same and you could address the wrong XIV system with your command. Therefore, it
might be good to always first issue a config_get command to verify that you are talking to
the right XIV system.

Example 6-1 illustrates the use of XCLI commands to set up a mirror volume.

Example 6-1 XCLI commands for mirror volume setup

-- Mirror HS1_1 (select slave volume)
schedule_create schedule="xiv_gui_schedule_30_1257877700437" interval=00:00:30
schedule_create schedule="xiv_gui_schedule_30_1257877700437" interval=00:00:30
mirror_create target="WSC_1300331" vol="HS1_1" slave_vol="HS1_1"
type=ASYNC_INTERVAL rpo=90 remote_rpo=90
schedule="xiv_gui_schedule_30_1257877700437"
remote_schedule="xiv_gui_schedule_30_1257877700437"

-- Mirror HS1_2 (create slave volume)

schedule_create schedule="xiv_gui_schedule_30_1257877742375" interval=00:00:30
schedule_create schedule="xiv_gui_schedule_30_1257877742375" interval=00:00:30
mirror_create target="WSC_1300331" vol="HS1_2" slave_vol="HS1_2" remote_pool="HS
Pool1" create_slave=yes type=ASYNC_INTERVAL rpo=90 remote_rpo=90
schedule="xiv_gui_schedule_30_1257877742375"
remote_schedule="xiv_gui_schedule_30_1257877742375"

-- Mirror HS1_3 (never schedule)

mirror_create target="WSC_1300331" vol="HS1_3" slave_vol="HS1_3"
type=ASYNC_INTERVAL rpo=90 remote_rpo=90 schedule="never" remote_schedule="never"

-- Mirror HS1_4 (Sync)

mirror_create target="WSC_1300331" vol="HS1_4" slave_vol="HS1_4"

-- Change Schedule HS1_3 (master)

schedule_create schedule="xiv_gui_schedule_30_1257878091625" interval=00:00:30
mirror_change_schedule vol="HS1_3" schedule="xiv_gui_schedule_30_1257878091625"

-- Change Schedule HS1_3 (slave)

schedule_create schedule="xiv_gui_schedule_30_1257878091625" interval=00:00:30
mirror_change_schedule vol="HS1_3" schedule="xiv_gui_schedule_30_1257878091625"

154 IBM XIV Storage System: Copy Services and Migration

 Activating the remote mirror coupling using the GUI
To activate the mirror, on the primary XIV go the Remote Mirroring menu and highlight all the
couplings that you want to activate, right-click, and select Activate, as shown in Figure 6-7.

Figure 6-7 Activate mirror

As seen in Figure 6-8, the Mirror panel now shows the status of the active mirrors as RPO
OK. All the async mirrors have the same mirroring status. Note that Sync Mirrors show the
status as synchronized.

Figure 6-8 Mirror status active

6.1.2 Consistency group configuration

IBM XIV Storage System leverages its consistency group capability to allow for mirroring
numerous volumes at once. The system creates snapshots of the master consistency groups
at user-configured intervals and synchronizes these point-in-time snapshots with the slave.
Setting the consistency group to be mirrored is done by first creating a consistency group,
then setting it to be mirrored, and only then populating it with volumes. A consistency group
must be created at the primary XIV and a corresponding consistency group at the secondary
XIV. The names of the consistency groups can be different. When creating a consistency
group, you also must specify the storage pool.

All volumes that you are going to add to the consistency group must be in that pool on the
primary XIV and in one pool at the secondary XIV. Adding a new volume pair to a mirrored

Chapter 6. Asynchronous Remote Mirroring 155

 consistency group requires the volumes to be mirrored exactly as the other volumes within
this consistency group.

Important: All volumes that you want to add to a mirroring consistency group must be
defined in the same pool at the primary site and must be in one pool at the secondary site.

It is possible to add a mirrored volume to a non-mirrored consistency group and have this
volume retain its mirroring settings.

To create a mirrored consistency group first create a CG on the primary and secondary XIV
Storage System. Then select the primary CG and specify Create Mirror (Figure 6-9).

Figure 6-9 Create mirrored CG

The consistency group must not contain any volume when you create the mirror, and be sure
to specify mirroring parameters that match the volumes that will be part of this CG, as shown
in Figure 6-10. The status of the new mirrored CG is now displayed in the Mirroring panel.

Figure 6-10 Async mirrored CG

156 IBM XIV Storage System: Copy Services and Migration

Adding a mirrored volume to a mirrored consistency group
The mirrored volume and the mirrored consistency group must have the following attributes:
򐂰 The volume is on the same system as the consistency group.
򐂰 The volume belongs to the same storage pool as the consistency group.
򐂰 Both the volume and the consistency group do not have outstanding sync jobs, either
scheduled or manual.
򐂰 The volume and consistency group have the same synchronization status.
򐂰 The volume’s and consistency group’s special snapshot (known as last_replicated
snapshot) have identical timestamps. This means that the volumes must have the same
schedule and at least one interval has passed since the creation of the mirrors.
For more information about asynchronous mirroring special snapshots refer to 6.5.4,
“Mirroring special snapshots” on page 172.

Also, mirrors for volumes must be activated before volumes can be added to a mirrored
consistency group. This activation results in the initial copy being completed and sync jobs
being run to create the special last_replicated snapshots (refer to Figure 6-7 on page 155).

As seen in Figure 6-11, the Mirror panel now shows the status of the active mirrors as RPO OK.
All the async mirrors and the mirrored CG have the same mirroring status. Note that sync
mirrors shows the status as synchronized.

Figure 6-11 Mirror status active

To add volumes to the mirrored CG the mirroring parameters must be identical, including the
last_replicated timestamps. The RPO and schedule will be changed to match the values set
for the mirrored consistency group. The volumes must have the same status (RPO OK). It is
possible that during the process the status may change or the last_replicated timestamp may
not be updated yet. If an error occurs verify the status and repeat the operation.

Chapter 6. Asynchronous Remote Mirroring 157

 Go to the Mirroring panel and verify the RPO and status for the volumes to be added to the
CG. Select each volume and specify to add to the consistency group (Figure 6-12).

Figure 6-12 Volumes and snapshots

Then specify the mirrored consistency group, as shown in Figure 6-13.

Figure 6-13 Select CG

The Mirroring panel now shows the consistency group as a group of volumes, all with the
same status for both the master CG (Figure 6-14) and the slave CG (Figure 6-15 on
page 159).

Figure 6-14 Master CG status

158 IBM XIV Storage System: Copy Services and Migration

Figure 6-15 Slave CG status

The Consistency Groups panel shows the last_replicated snapshots, and if the sync job is
currently running there will be a most_recent snapshot, as can be seen in Figure 6-16.

Figure 6-16 Mirrored CG: most_recent snapshot

Removing a volume from a mirrored consistency group

When removing a volume from a mirrored consistency group, the corresponding peer volume
will be removed from the peer consistency group. Mirroring is retained with the same
configuration as the consistency group from which it was removed. All ongoing consistency
groups’ sync jobs keep running.

Asynchronous mirroring and snapshot consistency groups

A volume can be in only one consistency group. Because consistency groups can be used for
snapshot and Remote Mirroring, confusion can arise. Define separate and specific CG for
snapshot and Remote Mirroring.

XCLI commands for consistency group configuration

Example 6-2 illustrates the use of XCLI commands for configuring consistency groups.

Example 6-2 XCLI commands for CG configuration

-- Activate async Mirrors
mirror_activate vol="HS1_1"
mirror_activate vol="HS1_2"
mirror_activate vol="HS1_3"

Chapter 6. Asynchronous Remote Mirroring 159

 -- Activate Mirror CG
mirror_activate cg="HS_Pool1_CG"

-- add volume to CG with changing RPO

mirror_change_schedule vol="HS1_3" schedule="xiv_gui_schedule_30_1258729268234"
schedule_delete schedule="xiv_gui_schedule_300_1258145098468"
mirror_change_schedule vol="HS1_3" schedule="xiv_gui_schedule_30_1258729268234"
mirror_change_rpo vol="HS1_3" rpo=90 remote_rpo=90
cg_add_vol cg="HS_Pool1_CG" vol="HS1_3"

-- Primary Mirror Status

>> mirror_list -t
local_peer_name,sync_type,current_role,target_name,remote_peer_name,active,sync_state,sched
ule_name,last_replicated_snapshot_time,specified_rpo
Name Mirror Type Role Remote System Remote Peer Active Status
Schedule Name Last Replicated RPO
HS1_1 async_interval Master WSC_1300331 HS1_1 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:41:30 0:01:30
HS1_2 async_interval Master WSC_1300331 HS1_2 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:41:30 0:01:30
HS1_3 async_interval Master WSC_1300331 HS1_3 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:41:30 0:01:30
HS_Pool1_CG async_interval Master WSC_1300331 HS_Pool1_CG yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:41:30 0:01:30

-- Secondary Mirror Status

>> mirror_list -t
local_peer_name,sync_type,current_role,target_name,remote_peer_name,active,sync_state,sched
ule_name,last_replicated_snapshot_time,specified_rpo
Name Mirror Type Role Remote System Remote Peer Active Status
Schedule Name Last Replicated RPO
HS1_1 async_interval Slave WSC_6000639 HS1_1 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:42:05 0:01:30
HS1_2 async_interval Slave WSC_6000639 HS1_2 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:42:05 0:01:30
HS1_3 async_interval Slave WSC_6000639 HS1_3 yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:42:05 0:01:30
HS_Pool1_CG async_interval Slave WSC_6000639 HS_Pool1_CG yes RPO OK
xiv_gui_schedule_30 2009-11-11 16:42:05 0:01:30

>> sync_job_list
Job Object Local Peer Source Target State Part of CG
Job Type
Volume HS1_1 last-replicated-HS1_1 most-recent-HS1_1 active yes
scheduled
Volume HS1_2 last-replicated-HS1_2 most-recent-HS1_2 active yes
scheduled
Volume HS1_3 last-replicated-HS1_3 most-recent-HS1_3 active yes
scheduled
CG HS_Pool1_CG last-replicated-HS1_1 most-recent-HS1_1 active no
scheduled

-- Remove from Mirrored Consistency Group

cg_remove_vol vol="HS1_3"

160 IBM XIV Storage System: Copy Services and Migration

6.1.3 Coupling activation, deactivation, and deletion
Mirroring can be manually activated and deactivated per volume or CG pair. When it is
activated, the mirror is in active mode. When it is deactivated, the mirror is in inactive mode.
These modes have the following functions:
򐂰 Active
Mirroring is functioning and the data is being written to the master and copied to the slave
peers at regular intervals.
򐂰 Inactive
Mirroring is deactivated. The data is not being replicated to the slave peer, but writes to the
master peer are being recorded and can later be replicated to the slave volume. Inactive
mode is used mainly when maintenance is performed on the secondary XIV system.

The mirror has the following characteristics:

򐂰 When a mirror is created, it is always initially in inactive mode.
򐂰 A mirror can only be deleted when its is in inactive mode.
򐂰 Transitions between the two states can only be performed from the XIV with the master.
򐂰 In a DR situation a role change changes the slave peers (at the secondary system) to a
master role (so that production can resume at the secondary). However, until the primary
site is recovered, the role of its volumes cannot be changed from master to slave. In this
case, both sides have the same role. When the primary site is recovered and before the
link is resumed, you must first change the role from master to slave at the primary (see
also 6.3, “Resynchronization after link failure” on page 169, and 6.4, “Disaster recovery”
on page 169).

The mirroring is terminated by deactivating the mirror and is required for the following actions:
򐂰 Terminating or deleting the mirroring
򐂰 Stopping the mirroring process
– For a planned network outage
– To reduce network bandwidth
– For a planned recovery test

The deactivation pauses a running sync job and no new sync jobs will be created as long as
the active state of the mirroring is not restored. However, the deactivation does not cancel the
status check by the master and the slave. The synchronization status of the deactivated
mirror is calculated as though the mirror was active.

Deactivating a mirror results in the synchronization status becoming RPO_Lagging when the
specified RPO time is exceeded. This means that the last-replicated snapshot is older than
the specified RPO.

Chapter 6. Asynchronous Remote Mirroring 161

 Change RPO and interval
The required RPO can be changed as illustrated in Figure 6-17 and the GUI selects a new
interval for the schedule. For example, as shown in Figure 6-18, the RPO was changed to 1
hour (01:00:00). The schedule selected is 15 min. This schedule can then be changed from
the Properties panel. There is a selection list of available intervals, as shown in Figure 6-19
on page 163.

Figure 6-17 Change RPO

Figure 6-18 New RPO value

162 IBM XIV Storage System: Copy Services and Migration

Figure 6-19 Change CG interval

Using XCLI commands to change RPO and interval

Example 6-3 illustrates the use of XCLI commands to change the RPO and interval.

Example 6-3 XCLI commands for changing RPO and interval

-- change RPO to 1 Hr and adjust schedule time
mirror_change_rpo cg="HS_Pool1_CG" rpo=3600 remote_rpo=3600
schedule_change schedule="xiv_gui_schedule_30_1258144725093" interval=00:15:00
schedule_rename schedule="xiv_gui_schedule_30_1258144725093"
new_name="xiv_gui_schedule_900_1258144924875"
---- on secondary
schedule_create schedule="xiv_gui_schedule_900_1258144924875" interval=00:15:00
mirror_change_schedule cg="HS_Pool1_CG"
schedule="xiv_gui_schedule_900_1258144924875"
schedule_delete schedule="xiv_gui_schedule_30_1258144725093"

-- change schedule time to 5 min

schedule_change schedule="xiv_gui_schedule_900_1258144924875" interval=00:05:00
schedule_rename schedule="xiv_gui_schedule_900_1258144924875"
new_name="xiv_gui_schedule_300_1258145098468"
---- on secondary
schedule_create schedule="xiv_gui_schedule_300_1258145098468" interval=00:05:00
mirror_change_schedule cg="HS_Pool1_CG"
schedule="xiv_gui_schedule_300_1258145098468"
schedule_delete schedule="xiv_gui_schedule_900_1258144924875"

Chapter 6. Asynchronous Remote Mirroring 163

 Deactivation on the master
To deactivate a mirror select Deactivate, as shown in Figure 6-20.

Figure 6-20 Mirror CG deactivate

The activation state changes to inactive, as shown in Figure 6-21. Subsequently, the
replication pauses (and records where it paused). Upon activation, the replication resumes.

Note that an ongoing sync job resumes upon activation. No new sync job will be created until
the next interval.

Figure 6-21 Mirror CG inactive

Deactivation on the slave

Deactivation is not available, regardless of the state of the mirror. However, the peer role can
be changed to master, which sets the status to inactive.

Note that for consistency group mirroring, deactivation pauses all running sync jobs
pertaining to the consistency group.

164 IBM XIV Storage System: Copy Services and Migration

Using XCLI commands for deactivation and activation
Example 6-4 shows XCLI commands for CG deactivation and activation.

Example 6-4 XCLI commands for CG deactivation and activation

-- Deactivate Mirrored CG
mirror_deactivate cg="HS_Pool1_CG"

-- Activate Mirrored CG
mirror_activate cg="HS_Pool1_CG"

Deletion
When a mirror pair (volume pairs or a consistency group) is inactive, the mirror relationship
can be deleted. When the mirror is deleted, the XIV forgets everything about the mirror. If you
want to set up the mirror again, the XIV must do an initial copy again from the source to the
target volume.

When the mirror is part of a consistency group the mirror first must be removed from the
mirrored CG and the last_replicated CG for the master and the slave must be disbanded. This
snapshot CG is recreated with only the current volumes after the next interval completes. The
last_replicated snapshots for the mirror can now be deleted, allowing a new mirror to be
created.

Note that when the mirror is deleted, the slave volume becomes a normal volume again, but
the volume is locked, which means that it is write protected. To enable writing to the volume
go to the Volumes list panel, select the volume, right-click it, and select Unlock.

The slave volume must also be formatted before it can be part of a new mirror. Formatting
also requires all snapshots of that volume to be deleted.

XCLI commands for mirror deletion

Example 6-5 illustrates the use of XCLI commands for mirror deletion.

Example 6-5 XCLI commands for mirror deletion

-- delete Mirror
cg_remove_vol vol="HS2_3"
mirror_deactivate vol="HS2_3"
mirror_delete vol="HS2_3"

-- Format slave volume

snap_group_disband snap_group="last-replicated-HS_Pool2_CG"
snapshot_delete snapshot="last-replicated-HS2_3"
vol_unlock vol="HS2_3"
vol_format vol="HS2_3"

-- Delete snapshots on Master

snap_group_disband snap_group="last-replicated-HS_Pool2_CG"
snapshot_delete snapshot="last-replicated-HS2_3"

Chapter 6. Asynchronous Remote Mirroring 165

6.2 Role reversal
Changing roles can be performed at any time (when a pair is active or inactive) for the slave,
and for the master when the mirror is inactive. A change role reverts only the role of that peer.
The single operation to switch roles is only available for synchronous mirrors. However, the
direction of the mirror can be reversed by following a process of multiple change role
operations.

Change role
In a disaster at the primary site, a role change at the secondary site is the normal recovery
action.

Assuming that the local site is down and that the remote site will become the main production
site, changing roles is performed at the remote (now production) site first. Later, when the
primary site is up again and communication is re-established, you also change the role at the
primary site to a slave to be able to establish mirroring from the secondary site back to the
primary site. This completes a switch role operation.

Changing the slave peer role

The role of the slave volume or consistency group can be changed to the master role, as
shown in Figure 6-22.

Figure 6-22 Change role of a slave consistency group

As shown in Figure 6-23, you are then prompted to confirm the role change (role reverse).

Figure 6-23 Verify change role

166 IBM XIV Storage System: Copy Services and Migration

After this changeover, the following is true:
򐂰 The slave volume or consistency group is now the master.
򐂰 The last_replicated snapshot is restored to the volumes
򐂰 The coupling has the status of inactive (Figure 6-24).

Figure 6-24 Slave becomes master

򐂰 The coupling remains in inactive mode (Figure 6-25). This means that remote mirroring is
deactivated. This ensures an orderly activation when the role of the peer on the other site
is changed.

Figure 6-25 Original master becomes inactive

The new master volume or consistency group starts to accept write commands from local
hosts. Because coupling is not active, in the same way as for any master volume, a log is
maintained of which write operations must be sent to the slave volume when communication
resumes.

After changing the slave to the master, an administrator also must change the original master
to the slave role before communication resumes (Figure 6-26). If both peers are left in the
same role (master), mirroring cannot be restarted.

Figure 6-26 Change role of a master consistency group

Chapter 6. Asynchronous Remote Mirroring 167

 Slave peer consistency
When the user is changing the slave volume or consistency group to a master volume or
master consistency group, they may not be in a consistent state. Therefore, the volumes are
automatically restored to the last_replicated snapshot.

Changing the master peer

When a peer role is changed from slave to master, then the mirror automatically becomes
inactive because both volumes are a master (Figure 6-25 on page 167). When coupling is
inactive, the master volume or consistency group can change roles. After such a change the
master volume or consistency group becomes the slave volume or consistency group
(Figure 6-27).

Figure 6-27 Original master becomes slave

Unsynchronized master becoming a slave volume or consistency group

When a master volume (or consistency group) is inactive, it is also not consistent with the
previous slave. Any changes made after the last replicated snapshot time will be lost when
the volume (CG) becomes a slave volume (CG). The data will be restored to the last
replicated snapshot to match the data on the peer volume, which is now the new master
volume.

Upon re-establishing the connection, the primary volume or consistency group (current slave
volume/CG) is updated from the secondary volume/CG (new master volume/CG) with data
that was written to the secondary volume after the last replicated snapshot timestamp.

Reconnection when both sides have the same role

Situations where both sides are configured to the same role can only occur when one side
was changed. The roles must be changed to have one master and one slave (volume or
consistency group). Change the volume roles as appropriate on both sides before the link is
resumed.

If the link is resumed and both sides have the same role, the coupling does not become
operational. The user must use the change role function on one of the volumes and then
activate the mirroring.

Peer reverts to last_replicated snapshot. See 6.5.4, “Mirroring special snapshots” on

page 172.

168 IBM XIV Storage System: Copy Services and Migration

 Using XCLI commands to change roles
Figure 6-28 shows an example of using XCLI commands to change roles.

-- Slave change role

mirror_change_role cg="HS_Pool1_CG"

-- Master change role

mirror_change_role cg="HS_Pool1_CG"

-- Activate new Master

mirror_activate cg="HS_Pool1_CG"

Figure 6-28 XCLI change roles

6.3 Resynchronization after link failure

When a link failure occurs, the primary system must start tracking changes to the mirror
source volumes so that these changes can be copied to the secondary once recovered.

When recovering from a link failure, the following steps are taken to synchronize the data:
򐂰 Asynchronous mirroring sync jobs proceed as scheduled. Sync jobs are restarted and a
new most_recent snapshot is taken. See 6.5.4, “Mirroring special snapshots” on
page 172.
򐂰 The primary system copies the changed data to the secondary volume. Depending on
how much data must be copied, this operation could take a long time, and the status
remains RPO_Lagging.

6.4 Disaster recovery

There are two broad categories of disaster:
򐂰 One that destroys the primary (local) site or destroys the data there
򐂰 One that makes the primary site or the data there unavailable, but that leaves the data
intact

However, within these broad categories there are a number of situations that may exist. Some
of these and the recovery procedures are considered below:
򐂰 A disaster that makes the XIV at the primary site unavailable, but leaves the site itself and
the servers there still available
In this scenario the volumes/CG on the XIV at the secondary site can be switched to
master volumes/CG, servers at the primary site can be redirected to the XIV at the
secondary site, and normal operations can start again. When the XIV at the primary site is
recovered the data can be mirrored from the secondary site back to the primary site. A full
initialization of the data is usually not needed.
Only changes that take place at the secondary site are transferred to the primary site. If
desired, a planned site switch can then take place to resume production activities at the
primary site. See 6.2, “Role reversal” on page 166, for details related to this process.

Chapter 6. Asynchronous Remote Mirroring 169

 򐂰 A disaster that makes the whole of the local site and data unavailable.
In this scenario, the standby (inactive) servers at the secondary site are activated and
attached to the secondary XIV to continue normal operations. This requires changing the
role of the slave peers to become master peers.
After the primary site is recovered, the data at the secondary site can be mirrored back to
the primary site. This most likely requires a full initialization of the primary site because the
local volumes may not contain any data. See 6.1, “Asynchronous mirroring configuration”
on page 150, for details related to this process.
When initialization completes the peer roles can be switched back to master at the primary
site and the slave at secondary site. The servers are then redirected back to the primary
site. See 6.2, “Role reversal” on page 166, for details related to this process.
򐂰 A disaster that breaks all links between the two sites but both sites remain running
In this scenario the primary site continues to operate as normal. When the links are
reestablished the data at the primary site can be resynchronized with the secondary site.
Only the changes since the previous last_replicated snapshot are sent to the remote site.

6.5 Mirroring process

This section explains the overall asynchronous mirroring process, from initialization to
outgoing operations. The asynchronous mirroring process generates snapshots of the master
at user-configured intervals and synchronizes these snapshots with the slave (see “Snapshot
life-cycle” on page 172).

6.5.1 Initialization process

The mirroring process starts with an initialization phase:
1. Read requests are served from the master. Upon each write operation to the master, the
master writes the data locally (primary site) and acknowledges the write operation.
2. Before any actual synchronization of a master can commence, a most_recent snapshot of
the master is created. This snapshot determines the scope of replication for the
initialization phase and the data to be replicated can be determined.

170 IBM XIV Storage System: Copy Services and Migration

3. The most_recent data is copied to the slave and a Last_Replicated snapshot of the slave
is taken (Figure 6-29).

Initialization Job completes

Initialization Sync Job

Master peer Slave peer
most_recent

last_replicated

Data to be replicated

Local Remote
site site

Figure 6-29 Initialization process completes

4. The most_recent snapshot on the master is renamed to last_replicated. This snapshot is

identical to the data in the last_replicated snapshot on the slave (Figure 6-30).

Master’s last_replicated snapshot created

Initialization phase ends

Master peer Slave peer

most_recent > last_replicated

las t_repl icated

last_replicated

Local Rem ote

sit e site

Figure 6-30 Ready for ongoing operation

5. Sync jobs can now be run to create periodic consistent copies of the master volumes or
consistency groups on the slave system. See 6.6, “Detailed asynchronous mirroring
process” on page 173.

Chapter 6. Asynchronous Remote Mirroring 171

6.5.2 Mirroring ongoing operation
Following the completion of the initialization phase, the master examines the synchronization
status at scheduled intervals and determines the scope of the synchronization. The following
process occurs whenever a synchronization is started:
1. A snapshot of the master is created.
2. The master calculates the differences between the master snapshot and the most recent
master snapshot that is synchronized with the slave.
3. The master establishes a synchronization process called a sync job that replicates the
differences from the master to the slave. Only data differences are replicated.

Details of this process can be found in 6.6, “Detailed asynchronous mirroring process” on
page 173.

6.5.3 Mirroring consistency groups

The synchronization status of the consistency group is determined by the status of all
volumes pertaining to this consistency group.
򐂰 The activation and deactivation of a consistency group affects all of its volumes.
򐂰 Role updates concerning a consistency group affect all of its volumes.
򐂰 It is impossible to directly activate, deactivate, or update the role of a given volume within a
consistency group.
򐂰 It is not possible to directly change the schedule of an individual volume within a
consistency group.

6.5.4 Mirroring special snapshots

The status of the synchronization process and the scope of the sync job are determined
through the use of the following two special snapshots:
򐂰 Most_recent snapshot
This snapshot is the most recent taken of the master system, either a volume or
consistency group. This snapshot is taken prior to the creation of a new sync job. This
entity is maintained on the master system only.
򐂰 Last_replicated snapshot
This is the most recent snapshot that has been fully synchronized with the slave system.
This snapshot is duplicated from the most_recent snapshot after the sync job is complete.
This entity is maintained on both the master and the slave systems.

Snapshot life-cycle
Throughout the sync job life cycle, the most_recent and last_replicated snapshots are created
and deleted to denote the completion of significant mirroring stages.

172 IBM XIV Storage System: Copy Services and Migration

 This mechanism bears the following characteristics and limitations:
򐂰 The last_replicated snapshots have two available time stamps:
– On the master system: the time that the last_replicated snapshot is copied from the
most_recent snapshot
– On the slave system: the time that the last_replicated snapshot is copied from the
master system
򐂰 No snapshot is created during the initialization phase.
򐂰 Snapshots are deleted only after newer snapshots are created.
򐂰 A failure in creating a last-replicated snapshot caused by space depletion is handled in a
designated process. See 6.8, “Pool space depletion” on page 182, for additional
information.
򐂰 Snapshots that are created by the snap now operation are retained and automatically
named by the system. This is identical to the last_replicated snapshot until a new sync job
runs.

Table 6-1 indicates which snapshot is created for a given sync job phase.

Table 6-1 Snapshots and sync job phases

Sync job Most_Recent Last_Replicated Details
phase snapshot snapshot

1 New interval Created on The most_recent snapshot is created

starts. the master only if there is no sync job running.
system

2 Calculate the The difference between the

differences. most_recent snapshot and the
last_replicated snapshot is transferred
from the master system to the slave
system.

3 The sync job is Created on the The last_replicated snapshot on the

complete. slave system slave system is created from the
snapshot that has just been mirrored.

4 Following the Created on the The last_replicated snapshot on the

creation of the master system master system is created from the
last_Replicated most_recent snapshot.
snapshot.

6.6 Detailed asynchronous mirroring process

After initialization is complete sync job schedules become active (unless schedule =never or
Type = external is specified for the mirror). This starts a specific process that replicates a
consistent set of data from the master to the slave. This process uses special snapshots to
preserve the state of the master and slave during the synchronization process. This allows
the changed data to be quantified and provides synchronous data points that can be used for
disaster recovery. See 6.5.4, “Mirroring special snapshots” on page 172.

Chapter 6. Asynchronous Remote Mirroring 173

 The sync job runs and the mirror status is maintained at the master system. If a previous sync
job is running, a new sync job will not start. The following actions are taken at the beginning of
each interval:
1. Most_recent snapshot is taken of the volume or consistency group:
a. Host I/O is halted.
b. The snapshot is taken to provide a consistent set of data to be replicated.
c. Host I/O resumes.
2. Changed data is copied to the slave:
a. The difference between the most_recent and last_replicated snapshots is determined.
b. This changed data is replicated to the slave.
Refer to Figure 6-31.

Sync job starts

The sync job data is
being replicated

Sync Job

Master peer most_recent Slave peer

last_replicated last_replicated

Data to be replicated

Local Remote
site site

Figure 6-31 Sync job starts

174 IBM XIV Storage System: Copy Services and Migration

3. A new last_replicated snapshot is created on the slave. This snapshot preserves the
consistent data for later recovery actions if needed. Refer to Figure 6-32.

Sync Job completed

…and a new last_replicated
snapshot is created that
represents the updated slave
peer’s state.

Master peer most_recent Slave peer

last_replicated last_replicated

Local Remote
site site

Figure 6-32 Sync job completes

4. The most recent_snapshot is renamed on the master (Figure 6-33):

a. The most recent data is now equivalent to the data on the slave.
b. Previous snapshots are deleted.
c. The most_recent snapshot is renamed to last_replicated.

New master last_replicated snapshot created

In one transaction - the master first deletes the current
last_replicated snapshot (8)
and the master then creates a new last_replicated
snapshot from the most_recent snapshot (9).

Interval sync process is now complete

The master and slave peers have an identical ‘restore
time point ‘ to which they can be reverted. This facilitates
among other things mirror peer switching .

most_recent > last_repli cated

Ma ster p ee r Sl av e p eer

last_replicated last_replicated

Local Remote
site site

Figure 6-33 New master’s last replicated snapshot

The next sync job can now be run at the next defined interval.

Chapter 6. Asynchronous Remote Mirroring 175

 Mirror synchronization status
Synchronization status is checked periodically and is independent of the mirroring process of
scheduling sync jobs. Refer to Figure 6-34 for a view of the synchronization states.

Example: RPO = Interval

Sync Job starts and
replicat es to Slave the
Master state at t 0

Master

Slave

Inter val Inte rval In terv al In terval In terva l

Interval
t0 t0’ t1 t2 t1’ t3
t4
tn

RPO_OK RPO_Lagging RPO_OK

If RPO is equal to or lower than the If RPO is higher than the difference
difference between the current time between the current time (when the
(when the check is run) and the check is calculated) and the timestamp of
timestamp of the the last_replicated_snapshot, then the
last_replicated_snapshot, then the status will be set to RPO_LAGGING
status will be set to RPO_OK

Figure 6-34 Synchronization states

The possible synchronization states are:

򐂰 Initialization
Synchronization does not start until the initialization completes.
򐂰 RPO_OK
Synchronization has completed within the specified sync job interval time (RPO).
򐂰 RPO_Lagging
Synchronization has completed but took longer than the specified interval time (RPO).

6.7 Asynchronous mirror step-by-step illustration

In the previous sections, the steps taken to set up, synchronize, and remove mirroring,
utilizing both the GUI and the XCLI were explained. In this section we provide an
asynchronous mirror step-by-step illustration.

6.7.1 Mirror initialization

At this point, we are continuing after the setup illustrated in 6.1, “Asynchronous mirroring
configuration” on page 150, which assumes that the Fibre Channel ports have been properly
defined as source and targets, the Ethernet switch has been updated to jumbo frames, and all
the physical paths are in place.

176 IBM XIV Storage System: Copy Services and Migration

 Mirrored volumes have been placed into a mirrored consistency group and the mirror has
been initialized and has a status of RPO OK. See Figure 6-35 and Figure 6-36.

Figure 6-35 Master status after setup

Figure 6-36 Slave status after setup

6.7.2 Remote backup scenario

One possible scenario related to the remote site is to provide a consistent copy of data that is
used as a periodic backup. This backup copy could be copied to tape or used for data-mining
activities that do not require the most current data.

This is accomplished by creating a duplicate of the last_replicated snapshot of the slave

consistency group. This new snapshot can then be mounted to hosts and backed up to tape
or used for other purposes.

Chapter 6. Asynchronous Remote Mirroring 177

 GUI steps to duplicate a snapshot group
From the Consistency Groups panel select Duplicate, as shown in Figure 6-37.

Figure 6-37 Duplicate last_replicated snapshot

A new snapshot is created with the same timestamp as the last_replicated snapshot
(Figure 6-38).

Figure 6-38 Duplicate snapshot

XCLI command to duplicate a snapshot group

Figure 6-39 illustrates the snap_group_duplicate command.

-- Duplicate last_replicated
snap_group_duplicate snap_group="last-replicated-HS_Pool1_CG"

Figure 6-39 XCLI to duplicate a snapshot group

178 IBM XIV Storage System: Copy Services and Migration

6.7.3 DR testing scenario
It is important to verify disaster recovery procedures. This can be accomplished by using the
remote volumes with hosts at the recovery site to verify that the data is consistent and that no
data is missing (due to volumes not being mirrored). This process is partly related to making
slave volumes available to the hosts, but it also includes processes external to the XIV system
commands. For example, the software available on the remote hosts and user access to
those hosts must also be verified. This example only covers the XIV system commands.

GUI steps for DR testing

The process begins by changing the role of the slave volumes to master volumes. This results
in the mirror being deactivated. The remote hosts can now access the remote volumes. See
Figure 6-40, Figure 6-41, and Figure 6-42 on page 180.

Figure 6-40 Change slave role to master

Figure 6-41 Verify change role

Chapter 6. Asynchronous Remote Mirroring 179

 Figure 6-42 New master volumes

After the testing is complete the remote volumes are returned to their previous slave role See
Figure 6-43, Figure 6-44, and Figure 6-45 on page 181.

Figure 6-43 Change role back to slave

Figure 6-44 Verify change role

180 IBM XIV Storage System: Copy Services and Migration

Figure 6-45 Slave role restored

Any changes made during the testing are removed by restoring the last_replicated snapshot,
and new updates from the local site will be transferred to the remote site when the mirror is
activated again (Figure 6-46 through Figure 6-48).

Figure 6-46 Activate mirror at local site

Figure 6-47 Master active

Figure 6-48 Slave active

Chapter 6. Asynchronous Remote Mirroring 181

 XCLI commands for DR testing
Figure 6-49 shows the steps and the corresponding XCLI commands required for DR testing.

-- Change Slave to Master

mirror_change_role cg="HS_Pool1_CG"

>> mirror_list -t
local_peer_name,sync_type,current_role,target_name,remote_peer_name,active
Name Mirror Type Role Remote System Remote Peer Active
HS1_1 async_interval Master WSC_6000639 HS1_1 no
HS1_2 async_interval Master WSC_6000639 HS1_2 no
HS1_3 async_interval Master WSC_6000639 HS1_3 no
HS_Pool1_CG async_interval Master WSC_6000639 HS_Pool1_CG no

-- Change back to Slave

mirror_change_role cg="HS_Pool1_CG"

>> mirror_list -t
local_peer_name,sync_type,current_role,target_name,remote_peer_name,active
Name Mirror Type Role Remote System Remote Peer Active
HS1_1 async_interval Slave WSC_6000639 HS1_1 no
HS1_2 async_interval Slave WSC_6000639 HS1_2 no
HS1_3 async_interval Slave WSC_6000639 HS1_3 no
HS_Pool1_CG async_interval Slave WSC_6000639 HS_Pool1_CG no

-- Activate Master on local Site

mirror_activate cg="HS_Pool1_CG"

>> mirror_list -t
local_peer_name,sync_type,current_role,target_name,remote_peer_name,active
Name Mirror Type Role Remote System Remote Peer Active
HS1_1 async_interval Slave WSC_6000639 HS1_1 yes
HS1_2 async_interval Slave WSC_6000639 HS1_2 yes
HS1_3 async_interval Slave WSC_6000639 HS1_3 yes
HS_Pool1_CG async_interval Slave WSC_6000639 HS_Pool1_CG yes

Figure 6-49 XCLI duplicate a snapshot group

6.8 Pool space depletion

The asynchronous mirroring process relies on special snapshots (most_recent,
last_replicated) that require and consume space from the pool snapshot reserve. An
adequate amount of snapshot space depends on the workload characteristics and the
intervals that you set for sync jobs. Observing your application over time allows you to
eventually fine tune the percentage of pool space to reserve for snapshots. Initially, a
conservative figure is to allocate 30% of the pool space to snapshots.

Tip: Set appropriate pool alert thresholds to be warned ahead of time and be able to take
proactive measures to avoid any serious pool space depletion situations.

182 IBM XIV Storage System: Copy Services and Migration

The XIV system has a sophisticated built-in process to cope with spool space depletion on
the slave or on the master before it eventually deactivates the mirror. If a pool does not have
enough free space to accommodate the storage requirements warranted by a new host write,
the system progressively deletes snapshots within that pool until enough space is made
available for successful completion of the write request.

Chapter 6. Asynchronous Remote Mirroring 183

184 IBM XIV Storage System: Copy Services and Migration
 7

Chapter 7. Data migration

This chapter introduces the XIV Storage System embedded data migration function, which is
used to migrate data from a non-XIV storage system to the XIV Storage System. The XIV
data migration function is included in the base XIV software and is very easy to deploy. This
chapter includes usage examples and troubleshooting information.

At a very high level, the steps to migrate to XIV using the XIV Data Migration function are:
1. Establish connectivity between source device and XIV. The source storage device must
have Fibre Channel connectivity with the XIV.
2. Collect configuration. Detail the configuration of the LUNs to be migrated.
3. Perform data migration:
– Stop/unconfigure all I/O from source-original LUNs.
– Start data migration in XIV.
– Activate new LUNs through XIV.
– Start all I/O on new XIV LUNs.

Note that all images of the XIV GUI shown in this chapter used Version 2.4.2 of the XIV GUI.
If you are still using an earlier version, certain panels may look slightly different.

© Copyright IBM Corp. 2010. All rights reserved. 185

7.1 Overview
Whatever the reason for your data migration, it is always desirable to avoid or minimize
disruption to your business applications. Whereas there are many options available for
migrating data from one storage system to another, the XIV Storage System includes a data
migration feature that enables the easy movement of data from an existing storage system to
the XIV Storage System. This feature enables the production environment to continue
functioning during the data transfer with only one brief period of downtime for your business
applications. Figure 7-1 illustrates a high-level view of what the data migration environment
could look like.

Module 9
2 4 XIV Port 4 Initiator Mode

1 3
1 5
Module 8
1 4

1 3

1 2
2 4
1 1
1 3 Non-XIV Disk System
1 0

Module 7
C h a n n e l1 D u a lP o rte d H o s tC h a n n e sl
2 1 Dvri e C h a n n e l 2 1 Ch 2 Ch1

OK

G /sb G b s/ DC
9 2 4
2
4
1 2
4
1
OK

ID /D a
i g

OK

2 4
8 AC

7 AC

Host Server 1 3
B OK
6 ID /D a
i g

1 2 1 2
DC 4 4 4 2
OK G b s/ G /sb

OK

Ch 1 Ch 2 1 2 D u a lP o rte d 1 2
5 H o s tC h a n n e sl D vri e C h a n n e l C h a n n e l2

Module 6
4

1
2 4
0

1 3
FC4
16 Module 5
2 4 XIV port 4 Initiator Mode

1 3
SAN Switch
Module 4
2 4
1 3

XIV Storage System

Figure 7-1 Data migration simple view

The IBM XIV Data Migration solution offers a smooth data transfer, because it:
򐂰 Requires only a single short outage to switch LUN ownership. This enables the immediate
connection of a host server to the XIV Storage System, providing the user with direct
access to all the data before it has been copied to the XIV Storage System.
򐂰 Synchronizes data between the two storage systems using transparent copying to the XIV
Storage System as a background process with minimal performance impact.
򐂰 Supports data migration from practically all storage vendors.
򐂰 Must be set up using Fibre Channel.
򐂰 Can be used to migrate SAN boot volumes.

The XIV Storage System manages the data migration by simulating host behavior. When
connected to the storage device containing the source data, XIV looks and behaves like a
SCSI initiator, which in common terms means that it acts like a host server. After the
connection is established, the storage device containing the source data believes that it is

186 IBM XIV Storage System: Copy Services and Migration

 receiving read requests from a host, when in fact it is the XIV Storage System doing a
block-by-block copy of the data, which the XIV is then writing onto an XIV volume.

During the background copy process, the host server is connected to the XIV Storage
System. The XIV Storage System handles all read and write requests from the host server,
even if the data is not resident on the XIV Storage System. In other words, during the data
migration, the data transfer is transparent to the host and the data is available for immediate
access.

It is important that the connections between the two storage systems remain intact during the
entire migration process. If at any time during the migration process the communication
between the storage systems fails, the process also fails. In addition, if communication fails
after the migration reaches synchronised status, writes from the host will fail if the source
updating option was chosen. The situation is further explained in the 7.2, “Handling I/O
requests” on page 187. The process of migrating data is performed at a volume level, as a
background process.

The data migration facility in XIV firmware revisions 10.1 and later supports the following:
򐂰 Up to four migration targets can be configured on an XIV (where a target is either one
controller in an active/passive storage device or one active/active storage device). XIV
firmware revision 10.2 increased the number of targets to 16. The target definitions are
used for both Remote Mirroring (RM) and data migration (DM). Both DM and RM functions
can be active at the same time. As already stated, an active/passive storage device with
two controllers uses two target definitions unless only one of the controllers is used for the
migration.
򐂰 The XIV can communicate with host LUN IDs ranging from 0 to 512 (in decimal). This
does not necessarily mean that the non-XIV disk system can provide LUN IDs in that
range. You may be restricted by the ability of the non-XIV storage controller to use only 16
or 256 LUN IDs depending on hardware vendor and device.
򐂰 Up to 4000 LUNs can be concurrently migrated.

Important: During the discussion in this chapter, the source system in a data migration
scenario is referred to as a target when setting up paths between the XIV Storage System
and the donor storage (the non-XIV storage). This terminology is also used in Remote
Mirroring, and both functions share the same terminology for setting up paths for
transferring data.

7.2 Handling I/O requests

The XIV Storage System handles all I/O requests for the host server during the data migration
process. All read requests are handled based on where the data currently resides. For
example, if the data has already been written to the XIV Storage System, it is read from that
location. However, if the data has not yet been migrated to the IBM XIV storage, the read
request comes from the host to the XIV Storage System, which in turn retrieves the data from
the source storage device and provides it to the host server.

The XIV Storage System handles all host server write requests and the non-XIV disk system
is now transparent to the host. All write requests are handled using one of two user-selectable
methods, chosen when defining the data migration. The two methods are known as source
updating and no source updating.

Chapter 7. Data migration 187

 An example of selecting which method to use is shown in Figure 7-2. The check box must be
selected to enable source updating, shown here as Keep Source Updated. Without this box
checked, changed data from write operations is only written to the XIV.

Figure 7-2 Keep Source Updated check box

Source updating
This method for handling write requests ensures that both storage systems (XIV and non-XIV
storage) are updated when a write I/O is issued to the LUN being migrated. By doing this the
source system remains updated during the migration process, and the two storage systems
remain in sync after the background copy process completes. Similar to synchronous Remote
Mirroring, the write commands are only acknowledged by the XIV Storage System to the host
after writing the new data to the local XIV volume, then writing to the source storage device,
and then receiving an acknowledgement from the non-XIV storage device.

An important aspect of selecting this option is that if there is a communication failure between
the target and the source storage systems or any other error that causes a write to fail to the
source system, the XIV Storage System also fails the write operation to the host. By failing
the update, the systems are guaranteed to remain consistent. Change management
requirements determine whether you choose to use this option.

No source updating
This method for handling write requests ensures that only the XIV volume is updated when a
write I/O is issued to the LUN being migrated. This method for handling write requests
decreases the latency of write I/O operations because write requests are only written to the
XIV volume and are not written to the non-XIV storage system. It must be clearly understood
that this limits your ability to back out a migration, unless you have another way of recovering
updates that were written to the volume being migrated after migration began. If the host is
being shutdown for the duration of the migration then this risk is mitigated.

Multi-pathing with data migrations

There are essentially two types of enterprise storage systems when it comes to multi-pathing:
򐂰 Active/active: These are storage systems where volumes can be active on all of the
storage system controllers at the same time (whether there are two controllers or more).
These systems support IO activity to any given volume down two or more paths. These
types of systems typically support load balancing capabilities between the paths with path
failover and recovery in the event of a path failure. The XIV is such a device and can utilize

188 IBM XIV Storage System: Copy Services and Migration

 this technology during data migrations. Examples of IBM products that are active/active
storage servers are the DS6000™, DS8000, ESS F20, ESS 800, and SVC. Note that the
DS6000 and SVC are examples of storage servers that have preferred controllers on a
LUN-by-LUN basis, but that if attached hosts ignore this preference, a potential
consequence is the risk of a small performance penalty.
If your non-XIV disk system supports active/active then you can carefully configure
multiple paths from XIV to non-XIV disk. The XIV load balances the migration traffic across
those paths and it automatically handles path failures.
򐂰 Active/passive: These are storage platforms where any given volume can be active on
only one controller at a time. These storage devices do not support I/O activity to any
given volume down multiple paths at the same time. Most support active volumes on one
or more controllers at the same time, but any given volume can only be active on one
controller at a time. An example of an IBM product that is an active/passive storage device
is the DS4700.

Migrating from an active/active storage device

If your non-XIV disk system supports active/active LUN access then you can configure
multiple paths from XIV to the non-XIV disk system. The XIV load balances the migration
traffic across these paths. This may lead to the temptation to configure more than two
connections or to increase the initialization speed to a very large value to speed up the
migration. However, the XIV only synchronizes one volume at a time per target (with four
targets, this means that four volumes could be being copied at once). This means that the
speed of the migration from each target is determined by the ability of the non-XIV storage
device to read from the LUN currently being migrated. Unless the non-XIV storage device has
striped the volume across multiple RAID arrays, the migration speed is unlikely to exceed
250–300 MBps (and could be much less), but this is totally dependant on the non-XIV storage
device.

Important: If multiple paths are created between an XIV and an active/active storage
device, the same SCSI LUN IDs must be used for each LUN on each path, or data
corruption may occur.

Migrating from an active/passive storage device

Because of the active/active nature of XIV, special considerations must be made when
migrating data from an active/passive storage device to XIV. A single path is configured
between any given non-XIV storage device controller and the XIV system. Many users decide
to perform migrations with the host applications offline, due to the single path.
򐂰 Define the target to the XIV per non-XIV storage controller (controller, not port). Define
only one path from that controller to the XIV. All volumes active on the controller can be
migrated using the defined target for that controller. For example, suppose the non-XIV
storage device contains two controllers (A and B):
– Define one target (called, for example, CX-A) with one path between the XIV and one
controller on the non-XIV storage device (for example, controller A). All volumes active
on this controller can be migrated by using this target. When defining the XIV initiator to
the controller, be sure to define it as not supporting fail-over. By doing so, volumes that
are passive on the A controller are not presented to the XIV. Check your non-XIV
storage device documentation for how to do this.
– Define another target (called, for example, CX-B) with one path between the XIV and
controller B. All volumes active down the B controller can be migrated to the XIV by
using this target. When defining the XIV initiator to the controller, be sure to define it as
not supporting failover. By doing so, volumes that are passive on the B controller are

Chapter 7. Data migration 189

 not presented to the XIV. Check your non-XIV storage device documentation for how to
do this.

Note: Certain examples shown in this chapter are from a DS4000 active/passive migration
with each DS4000 controller defined independently as a target to the XIV Storage System.
If you define a DS4000 controller as a target do not define the alternate controller as a
second port on the first target. Doing so causes unexpected issues such as migration
failure, preferred path errors on the DS4000, or very slow migration progress.

7.3 Data migration steps

The high-level steps required when migrating a volume from a non-XIV system to the IBM XIV
Storage System are:
1. Initial connection setup:
– Zone or cable the XIV to the non-XIV storage device.
– Define XIV to a non-XIV storage device (as a host).
– Define a non-XIV storage device to XIV (as a migration device).
2. Define the host being migrated to the XIV (using WWPNs).
3. Create and activate a data migration volume on XIV.
– Perform pre-migration tasks for the host being migrated:
• Back up your data.
• Shut down your host or application or unmount the file system.
• Zone host to XIV.
• Update host drivers.
– Define and test the data migration volume.
• On non-XIV storage, map volumes away from host and map them instead to XIV.
• On XIV, create data migration and test it.
4. Activate data migration.
– On XIV, activate data migration.
– On XIV, map volumes to host.
– Bring the host online and detect volumes.
5. Complete the data migration on XIV.
– On XIV, monitor the migration.
– On XIV, delete the migration.

Each step is further explained in the sections that follow.

7.3.1 Initial connection setup

For the initial connection setup, start by zoning or cabling XIV to the system being migrated.

190 IBM XIV Storage System: Copy Services and Migration

Zone or cable the XIV to the non-XIV storage device
Because the non-XIV storage device views the XIV as a host, the XIV must connect to the
non-XIV storage system as a SCSI initiator. Therefore, the physical connection from the XIV
must be from initiator ports on the XIV (which by default for Fibre Channel is port 4 on each
active interface module). The initiator ports on the XIV can be directly attached to the non-XIV
storage (without an intervening switch) or they may be fabric attached (it which case they will
need to be zoned to the non-XIV storage system). Two physical connections from two
separate modules on two separate fabrics are recommended for redundancy (although
redundant pathing will not be possible on active/passive controllers).

It is also possible that the host may be attached via one medium (such as iSCSI), whereas
the migration occurs via the other (such as Fibre Channel). The host-to-XIV connection
method and the data migration connection method are independent of each other.

Depending on the non-XIV storage device vendor and device, it may be easier to zone the
XIV to the ports where the volumes being migrated are already present. In this manner no
reconfiguration of the non-XIV storage device may be required. For example, in EMC
Symmetrix/DMX environments, it is easier to zone the fiber adapters (FAs) to the XIV where
the volumes are already mapped.

At the completion of this step you will have:

1. Run cables from port 4 on each selected XIV interface module to either a fabric switch or
directly to the non-XIV storage (if the non-XIV storage has free host ports and it supports
direct attach).
2. Zoned the XIV initiator ports (whose WWPNs end in 3) to the selected non-XIV storage
device host ports using single initiator zoning (each zone contains one initiator port and
one target port).

Chapter 7. Data migration 191

 Figure 7-3 depicts a fabric-attached configuration. It shows that module 4 port 4 is zoned to a
port on the non-XIV storage via fabric A. Module 7 port 4 is zoned to a port on the non-XIV
storage via fabric B.

Figure 7-3 Fabric attached

Define XIV to the non-XIV storage device (as a host)

Once the physical connection between the XIV and non-XIV storage device is complete, the
XIV initiator (WWPN) must be defined on the non-XIV storage device. The process to achieve
this is vendor and device dependent because you must use the non-XIV storage device
management interface. Therefore, refer to the non-XIV storage vendor’s documentation for
how to configure initiators to the storage device.

If you have already zoned the XIV to the non-XIV storage device, then the WWPNs of the XIV
initiator ports (that end in the number 3) will appear in the WWPN drop-down list. If they are
not there then you must manually add them (though this strongly suggests either the need to
map a LUN0, or that the SAN zoning has not been done correctly).

The XIV must be defined as a Linux or Windows host to the non-XIV storage device. If the
non-XIV device offers several variants of Linux, you can choose SuSE Linux or RedHat Linux
or Linux x86. This defines the correct SCSI protocol flags for communication between the XIV
and non-XIV storage device. The principal criterion is that the host type must start LUN
numbering with LUN ID 0. If the non-XIV storage device is active/passive, check to see
whether the host type selected affects LUN failover between controllers, such as DS4000
(see 7.12.5, “IBM DS3000/DS4000/DS5000” on page 227, for more details).

There may also be other vendor-dependant settings. Section 7.12, “Device-specific

considerations” on page 223, contains additional information.

192 IBM XIV Storage System: Copy Services and Migration

Define non-XIV storage device to XIV (as a migration target)
Once the physical connectivity is made and the XIV has been defined to the non-XIV storage
storage device, the non-XIV storage device must be defined on the XIV. This includes defining
the storage device object, defining the WWPN ports on the non-XIV storage device, and
defining the connectivity between the XIV and the non-XIV storage device.
1. In the XIV GUI go to the Remote  Migration Connectivity panel.
2. Right-click and choose Add Target, which brings up the menu shown in Figure 7-4. The
choices that must be configured are:
– Target Name: Type in a name of your own choice.
– Target Protocol: Choose FC from the pull-down menu.
Click Define.

Figure 7-4 Defining the non-XIV storage device

Tip: The data migration target is represented by an image of a generic rack. If you must
delete or rename the migration device do so by right-clicking the image of that rack.

3. On the dark shaded box that is part of the defined target, right-click and choose Add Port
(Figure 7-5).

Figure 7-5 Defining the target port

a. Enter the WWPN of the first (fabric A) port on the non-XIV storage device zoned to the
XIV. There is no drop-down menu of WWPNs, so you must manually type or paste in
the correct WWPN. Be careful not to make a mistake. It is not necessary to use full
colons to separate every second number. It makes no difference if you enter a WWPN
as 10:00:00:c9:12:34:56:78 or 100000c912345678.
b. Click Add.

Chapter 7. Data migration 193

 4. Enter another port (repeating step 3) for those storage devices that support active/active
multi-pathing. This could be the WWPN that is zoned to the XIV on a separate fabric.
5. Connect the XIV and non-XIV storage ports that are zoned to one another. This is done by
clicking and dragging from port 4 on the XIV to the port (WWPN) on the non-XIV storage
device to where the XIV is zoned. In Figure 7-6 the mouse started at module 9 port 4 and
has nearly reached the target port. The connection is currently colored blue and turns red
when the mouse connects to port 1 on the target.

Figure 7-6 Dragging a connection between XIV and migration target

194 IBM XIV Storage System: Copy Services and Migration

 In Figure 7-7 the connection from module 9 port 4 to port 1 on the non-XIV storage device is
currently active, as noted by the green color of the connecting line. This means that the
non-XIV storage system and XIV are connected and communicating (indicating that SAN
zoning was done correctly. The correct XIV initiator port was selected. The correct target
WWPN was entered and selected, and LUN 0 was detected on the target device). If there is
an issue with the path, the connection line is red.

Figure 7-7 Non-XIV storage device defined

Tip: Depending on the storage controller, ensuring that LUN0 is visible on the non-XIV
storage device down the controller path that you are defining helps ensure proper
connectivity between the non-XIV storage device and the XIV. Connections from XIV to
DS4000 or EMC DMX or Hitachi HDS devices require a real disk device to be mapped as
LUN0. However, the IBM ESS 800, for instance, does not need a LUN to be allocated to
the XIV for the connection to become active (turn green in the GUI). The same is true for
EMC CLARiiON.

7.3.2 Define the host being migrated to the XIV

Prior to performing data migrations and allocating the volumes to the hosts, the host must be
defined on the XIV. Volumes are then mapped to the hosts or clusters. If the host is to be a
member of a cluster, then the cluster must be defined first. However, a host can be moved
easily from or added to a cluster at any time. This also requires that the host be zoned to your
XIV target ports via the SAN fabric.
1. To define a cluster (optional):
a. In the XIV GUI go to the floating menu Host and Clusters  Host and Clusters.
b. Choose Add Cluster from the top menu bar.

Chapter 7. Data migration 195

 c. Name: Enter a cluster name in the provided space.
d. Click OK.
2. To define a host:
a. In the XIV GUI go to the floating menu Host and Clusters  Hosts and Clusters.
b. Choose Add Host from the top menu bar.
i. Name: Enter a host name.
ii. Cluster: If the host is part of a cluster, choose the cluster from the drop-down menu.
iii. Click Add.
iv. Select the host and right-click to bring up a menu, from which you choose Add
Port.
i. Port Type: Choose FC from the drop-down menu.
ii. Port Name: This is a drop-down menu of WWPNs that are logged into the XIV but
that have not been assigned to a host. WWPNs can be chosen from the drop-down
menu or entered manually.
iii. Click Add.
iv. Repeat the above steps to add all the HBAs of the host being defined.

7.3.3 Creating and activating a data migration volume

Perform the steps explained below.

Perform pre-migration tasks for the host being migrated

To perform pre-migration tasks:
1. Back up the volumes being migrated.
A full restorable backup must be created prior to any data migration activity. It is a best
practice to verify the backup and to verify that all the data is restorable and that there are
no backup media errors.
2. Shut down the application/host.
Before the actual migration can begin the application must be quiesced. This ensures that
the application data is in a consistent state. Because the host may need to be rebooted a
number of times prior to the application data being available again, also consider the
following steps:
– Set applications to not automatically start when the host operating system restarts.
– Stop file systems from being automatically remounted on boot. For UNIX®-based
operating systems consider commenting out all affected file system mount points in the
fstab or vfstab.

Note: In clustered environments you could work with only one node until the migration
is complete, so consider shutting down all other nodes in the cluster.

3. Perform a point-in-time copy of the volume on the non-XIV storage device (if that function
is available on the non-XIV storage). This point-in-time copy is a gold copy of the data that
is quiesced prior to starting the data migration process. Do this before changing any host
drivers or installing new host software, particularly if you are going to migrate boot from
SAN volumes.

196 IBM XIV Storage System: Copy Services and Migration

 4. Zone the host to XIV. The host must be directed (via SAN fabric zoning) to the XIV instead
of the non-XIV storage system. This is because the XIV is acting as a proxy between the
host and the non-XIV storage system. The host must no longer access the non-XIV
storage system once the data migration is activated. The host must perform all I/O through
the XIV.
5. Perform host administrative procedures. The host must be configured using the XIV host
attachment procedures. These may include removing any existing/non-XIV multi-pathing
software and installing the native multi-pathing drivers and recommended patches as
stated in the XIV Host Attachment Guides. Install the most current HBA driver and
firmware at this time. One or more reboots may be required. Documentation and other
software can be found here:
http://www.ibm.com/support/search.wss?q=ssg1*&tc=STJTAG+HW3E0&rs=1319&dc=D400&dtm

Define and test data migration volume

To do this:
1. Allocate the non-XIV volume to XIV.
The volumes being migrated to the XIV must be allocated via LUN mapping to the XIV.
The LUN ID presented to the XIV must be a decimal value from 0 to 512. If it uses
hexadecimal LUN numbers then the LUN IDs can range from 0x0 to 0x200, but must be
converted to decimal when entered into the XIV GUI. The XIV does not recognize a host
LUN ID above 512 (decimal). Figure 7-8 shows LUN mapping using a DS4700. It depicts
the XIV as a host called XIV_Migration_Host with four DS4700 logical drives mapped to
the XIV as LUN IDs 0 to 3.

Figure 7-8 Non-XIV LUNs defined to XIV

When mapping volumes to the XIV it is very important to note the LUN IDs allocated by
the non-XIV storage. The methodology to do this varies by vendor and device and is
documented in greater detail in 7.12, “Device-specific considerations” on page 223.

Important: You must unmap the volumes away from the host during this step, even if
you plan to power the host off during the migration. The non-XIV storage only presents
the migration LUNs to the XIV. Do not allow a possibility for the host to detect the LUNs
from both the XIV and the non-XIV storage.

2. Define data migration object/volume.

Once the volume being migrated to the XIV is allocated to the XIV, a new data migration
(DM) volume can be defined. The source volume from the non-XIV storage system and
XIV volume must be exactly the same size.

Chapter 7. Data migration 197

 Important: You cannot use the XIV data migration function to migrate data to a source
volume in an XIV remote mirror pair. If you need to do this, migrate the data first and
then create the remote mirror after the migration is completed.

If you want to manually create the volumes on the XIV, consult 7.5, “Manually creating the
migration volume” on page 207. Preferably, instead continue with the next step, discussed
in the following section

XIV volume automatically created

The XIV has the ability to determine the size of the non-XIV volume and create the XIV
quickly when the data migration object is defined. This method is easy, which helps avoid
potential issues when manually calculating the real block size of a volume.
1. In the XIV GUI go to the floating menu Remote  Migration.
2. Right-click and choose Define Data Migration. This brings up a panel like that shown in
Figure 7-9.
– Destination Pool: Choose the pool from the drop-down menu where the volume will be
created.
– Destination Name: Enter a user-defined name. This will be the name of the local XIV
volume.
– Source Target System: Choose the already defined non-XIV storage device from the
drop-down menu.

Important: If the non-XIV device is active/passive, then the source target system
must represent the controller (or service processor) on the non-XIV device that
currently owns the source LUN being migrated. This means that you must check,
from the non-XIV storage, which controller is presenting the LUN to the XIV.

Figure 7-9 Define Data Migration object/volume

– Source LUN: Enter the decimal value of the host LUN ID as presented to the XIV from
the non-XIV storage system. Certain storage devices present the LUN ID as hex. The
number in this field must be the decimal equivalent. Ensure that you do not accidentally
use internal identifiers that you may also see on the source storage systems

198 IBM XIV Storage System: Copy Services and Migration

 management panels. In Figure 7-8 on page 197, the correct values to use are in the
LUN column (numbered 0 to 3).
– Keep Source Updated: Check this if the non-XIV storage system source volume is to
be updated with writes from the host. In this manner all writes from the host will be
written to the XIV volume, as well as the non-XIV source volume, until the data
migration object is deleted.
Click Define and the migration appears as shown in Figure 7-10.

Figure 7-10 Defined data migration object/volume

3. Test the data migration object. Right-click to select the created data migration object and
choose Test Data Migration. If there are any issues with the data migration object the test
fails, reporting the issue found. See Figure 7-11.

Figure 7-11 Test Data Migration

Tip: If you are migrating volumes from an MSCS cluster that is still active, then testing a
migration may fail due to the reservations placed on the source LUN by MSCS. You must
bring the cluster down to get the test to succeed.

Chapter 7. Data migration 199

7.3.4 Create and activate a data migration volume on XIV
Once the data migration volume has been tested the process of the actual data migration can
begin. When data migration is initiated, the data is copied sequentially in the background from
the non-XIV storage system volume to the XIV. The host reads and writes data to the XIV
storage system without being aware of the background I/O being performed.

Note: Once activated, the data migration can be deactivated, but after deactivating the
data migration the host is no longer able to read or write to the migration volume and all
host I/O stops. Do not deactivate the migration with host I/O running. If you want to
abandon the data migration prior to completion consult the back-out process described in
section 7.10, “Backing out of a data migration” on page 219.

1. Activate the data migration.

Right-click to select the data migration object/volume and choose Activate. This begins
the data migration process where data is copied in the background from the non-XIV
storage system to the XIV. Activate all volumes being migrated so that they can be
accessed by the host. The host has read and write access to all volumes, but the
background copy occurs serially volume by volume. If two targets (such as non-XIV1 and
non-XIV2) are defined with four volumes each, two volumes are actively copied in the
background—one volume from non-XIV1 and another from non-XIV2. All eight volumes
are accessible by the hosts.
Figure 7-12 shows the menu choices when right-clicking the data migration. Note the Test
Data Migration, Delete Data Migration, and Activate menu items, as these are the
most-used commands.

Figure 7-12 Activate data migration

2. Allocate volumes to the host.

Once the data migration has been started, you can use the XIV GUI or XCLI to map the
migration volumes to the host. When mapping volumes to hosts on the XIV, LUN ID 0 is
reserved for XIV in-band communication. This means that the first LUN ID that you
normally use is LUN ID 1. This includes boot-from-SAN hosts. You may also choose to use
the same LUN IDs as were used on the non-XIV storage, but this is not mandatory.

Important: The host cannot read the data on the non-XIV volume until the data
migration has been activated. The XIV does not pass through (proxy) I/O for a migration
that is inactive. If you use the XCLI dm_list command to display the migrations, ensure
that the word Yes appears in the Active column for every migration.

200 IBM XIV Storage System: Copy Services and Migration

 3. Bring the host server online.
Once the volumes have been mapped to the host server, the host can be brought online.
When the host boots up successfully and volume visibility has been verified, the
application can be brought up and operations verified.

Note: In clustered environments, it is usually recommended that only one node of the
cluster be initially brought online after the migration is started, and that all other nodes
be offline until the migration is complete. Once complete, update all other nodes (driver,
host attachment package, and so on), as the primary node was during the initial outage
(see step 5 in “Perform pre-migration tasks for the host being migrated” on page 196).

4. Data migration progress.

Figure 7-13 shows the progress of the data migrations. The status bar can be toggled
between GB remaining, percent complete, and hours/minutes remaining. Figure 7-13
shows four data migrations, one of which has started background copy and three of which
have not. Only one migrations is being copied at this same time because there is only one
target (ITSO_DS4700_Ctrl_B).

Figure 7-13 Data migration progress

7.3.5 Complete the data migration on XIV

To complete the data migration, perform the following sequence of steps:
1. Synchronization
After all of a volume’s data has been copied, the data migration achieves synchronization
status. After synchronization is achieved, all read requests are served by the XIV Storage
System. If source updating was selected the XIV will continue to write data to both itself
and the outgoing storage system until the data migration is deleted. Figure 7-14 depicts a
completed migration.

Figure 7-14 Data migration complete

2. Delete data migration. Once the synchronization has been achieved, the data migration
object can be safely deleted without host interruption.

Chapter 7. Data migration 201

 Important: If this is an online migration, do not deactivate the data migration prior to
deletion, as this causes host I/O to stop and possibly causes data corruption.

Right-click to select the data migration volume and choose Delete Data Migration, as
shown in Figure 7-15. This can be done without host/server interruption.

Figure 7-15 Delete Data Migration

Deleting an inactive or unsynchronized data migration

For safety purposes, you cannot delete an inactive or unsynchronized data migration from the
Data Migration panel. An unfinished data migration can only be deleted by deleting the
relevant volume from the Volumes  Volumes & Snapshots section in the XIV GUI.

7.4 Command-line interface

All of the XIV GUI operation steps can be performed using the XIV command-line interface
(XCLI) either through direct command execution or through batch files containing numerous
commands. This is especially helpful in migration scenarios involving numerous LUNs. This
section lists the XCLI command equivalent of the GUI steps shown above. A full description of
all the XCLI commands can be found in the XCLI Users Guide available at the following IBM
Web site:
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/topic/com.ibm.help.xiv.doc/docs
/GC27-2213-02.pdf

Every command issued in the XIV GUI is logged in a text file with the correct syntax. This is
very helpful for creating scripts. If you are running the XIV GUI under Microsoft Windows, look
for a file titled guicommands_< todays date >.txt, which will be found in the following folder:
C:\Documents and Settings\ < Windows user ID >\Application Data\XIV\GUI10\logs

202 IBM XIV Storage System: Copy Services and Migration

All of the commands given on the next few pages are effectively in the order in which you
must execute them, starting with the commands to list all current definitions (which will also
be needed when you start to delete migrations).
򐂰 List targets.
Syntax target_list
򐂰 List target ports.
Syntax target_port_list
򐂰 List target connectivity.
Syntax target_connectivity_list
򐂰 List clusters.
Syntax cluster_list
򐂰 List hosts.
Syntax host_list
򐂰 List volumes.
Syntax vol_list
򐂰 List data migrations.
Syntax dm_list
򐂰 Define target (Fibre Channel only).
Syntax target_define target=<Name> protocol=FC xiv_features=no
Example target_define target=DMX605 protocol=FC xiv_features=no
򐂰 Define target port (Fibre Channel only).
Syntax target_port_add fcaddress=<non-XIV storage WWPN>
target=<Name>
Example target_port_add fcaddress=0123456789012345 target=DMX605
򐂰 Define target connectivity (Fibre Channel only).
Syntax target_connectivity_define
local_port=1:FC_Port:<Module:Port> fcaddress=<non-XIV
storage WWPN> target=<Name>
Example target_connectivity_define local_port=1:FC_Port:5:4
fcaddress=0123456789012345 target=DMX605
򐂰 Define cluster (optional).
Syntax cluster_create cluster=<Name>
Example cluster_create cluster=Exch01
򐂰 Define host (if adding host to a cluster).
Syntax host_define host=<Host Name> cluster=<Cluster Name>
Example host_define host=Exch01N1 cluster=Exch01
򐂰 Define host (if not using cluster definition).
Syntax host_define host=<Name>
Example host_define host=Exch01

Chapter 7. Data migration 203

 򐂰 Define host port (Fibre Channel host bus adapter port).
Syntax host_add_port host=<Host Name> fcaddress=<HBA WWPN>
Example host_add_port host=Exch01 fcaddress=123456789abcdef1
򐂰 Create XIV volume using decimal GB volume size.
Syntax vol_create vol=<Vol name> size=<Size> pool=<Pool Name>
Example vol_create vol=Exch01_sg01_db size=17 pool=Exchange
򐂰 Create XIV volume using 512 byte blocks.
Syntax vol_create vol=<Vol name> size_blocks=<Size in blocks>
pool=<Pool Name>
Example vol_create vol=Exch01_sg01_db size_blocks=32768
pool=Exchange
򐂰 Define data migration.
If you want the local volume to be automatically created:
Syntax dm_define target=<Target> vol=<Volume Name> lun=<Host LUN
ID as presented to XIV> source_updating=<yes|no>
create_vol=yes pool=<XIV Pool Name>
Example dm_define target=DMX605 vol=Exch01_sg01_db lun=5
source_updating=no create_vol=yes pool=Exchange
If the local volume was pre-created:
Syntax dm_define target=<Target> vol=<Pre-created Volume Name>
lun=<Host LUN ID as presented to XIV>
source_updating=<yes|no>
Example dm_define target=DMX605 vol=Exch01_sg01_db lun=5
source_updating=no
򐂰 Test data migration object.
Syntax dm_test vol=<DM Name>
Example dm_test vol=Exch_sg01_db
򐂰 Activate data migration object.
Syntax dm_activate vol=<DM Name>
Example dm_activate vol=Exch_sg01_db
򐂰 Map volume to host/cluster.
– Map to host:
Syntax map_vol host=<Host Name> vol=<Vol Name> lun=<LUN ID>
Example map_vol host=Exch01 vol=Exch01_sg01_db lun=1
– Map to cluster:
Syntax map_vol host=<Cluster Name> vol=<Vol Name> lun=<LUN ID>
Example map_vol host=Exch01 vol=Exch01_sg01_db lun=1
򐂰 Delete data migration object.
If the data migration is synchronized and thus completed:
Syntax dm_delete vol=<DM Volume name>
Example dm_delete vol=Exch01_sg01_db

204 IBM XIV Storage System: Copy Services and Migration

 If the data migration is not complete it must be deleted by removing the corresponding
volume from the Volume and Snapshot menu (or via the vol_delete command below).
򐂰 Delete volume (not normally needed).
Challenged volume delete (cannot be done via a script, as this command must be
acknowledged):
Syntax vol_delete vol=<Vol Name>
Example vol_delete vol=Exch_sg01_db
If you want to perform an unchallenged volume deletion:
Syntax vol_delete -y vol=<Vol Name>
Example vol_delete -y vol=Exch_sg01_db
򐂰 Delete target connectivity.
Syntax target_connectivity_delete
local_port=1:FC_Port:<Module:Port> fcaddress=<non-XIV
storage device WWPN> target=<Name>
Example target_connectivity_delete local_port=1:FC_Port:5:4
fcaddress=0123456789012345 target=DMX605
򐂰 Delete target port.
Fibre Channel
Syntax target_port_delete fcaddress=<non-XIV WWPN> target=<Name>
Example target_port_delete fcaddress=0123456789012345 target=DMX605
򐂰 Delete target.
Syntax target_delete target=<Target Name>
Example target_delete target=DMX605

7.4.1 Using XCLI scripts or batch files

In order to execute a XCLI batch job, it is best to use the XCLI (versus the XCLI Session).

Setting environment variables in Windows

You can remove the need to specify user and password information for every command by
making that information an environment variable. Example 7-1 shows how this is done using
a Windows command prompt. First the XIV_XCLIUSER variable is set to admin, then the
XIV_XCLIPASSWORD is set to adminadmin. Then both variables are confirmed as set. If
necessary, change the user ID and password to suit your setup.

Example 7-1 Setting environment variables in Microsoft Windows

C:\>set XIV_XCLIUSER=admin
C:\>set XIV_XCLIPASSWORD=adminadmin
C:\>set | find "XIV"
XIV_XCLIPASSWORD=adminadmin
XIV_XCLIUSER=admin

To make these changes permanent:

1. Right-click the My Computer icon and select Properties.
2. Click the Advanced tab.
3. Click Environment Variables.

Chapter 7. Data migration 205

 4. Click New for a new system variable.
5. Create the XIV_XCLIUSER variable with the relevant user name.
6. Click New again to create the XIV_XCLIPASSWORD variable with the relevant password.

Setting environment variables in UNIX

If your are using a UNIX-based operating system export the environment variables as shown
in Example 7-2 (which in this example is AIX®). In this example the user and password
variables are set to admin and adminadmin and then confirmed as being set.

Example 7-2 Setting environment variables in UNIX

root@dolly:/tmp/XIVGUI# export XIV_XCLIUSER=admin
root@dolly:/tmp/XIVGUI# export XIV_XCLIPASSWORD=adminadmin
root@dolly:/tmp/XIVGUI# env | grep XIV
XIV_XCLIPASSWORD=adminadmin
XIV_XCLIUSER=admin

To make these changes permanent update the relevant profile, making sure that you export
the variables to make them environment variables.

7.4.2 Sample scripts

With the environment variables set, a script or batch file like the one in Example 7-3 can be
run from the shell or command prompt in order to define the data migration pairings.

Example 7-3 Data migration definition batch file

xcli -m 10.10.0.10 dm_define vol=MigVol_1 target=DS4200_CTRL_A lun=4
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_2 target=DS4200_CTRL_A lun=5
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_3 target=DS4200_CTRL_A lun=7
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_4 target=DS4200_CTRL_A lun=9
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_5 target=DS4200_CTRL_A lun=11
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_6 target=DS4200_CTRL_A lun=13
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_7 target=DS4200_CTRL_A lun=15
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_8 target=DS4200_CTRL_A lun=17
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_9 target=DS4200_CTRL_A lun=19
source_updating=no create_vol=yes pool=test_pool
xcli -m 10.10.0.10 dm_define vol=MigVol_10 target=DS4200_CTRL_A lun=21
source_updating=no create_vol=yes pool=test_pool

With the data migration defined via the script or batch job above, an equivalent script or batch
job to execute the data migrations then must be run, as shown in Example 7-4.

Example 7-4 Activate data migration batch file

xcli -m 10.10.0.10 dm_activate vol=MigVol_1
xcli -m 10.10.0.10 dm_activate vol=MigVol_2
xcli -m 10.10.0.10 dm_activate vol=MigVol_3

206 IBM XIV Storage System: Copy Services and Migration

 xcli -m 10.10.0.10 dm_activate vol=MigVol_4
xcli -m 10.10.0.10 dm_activate vol=MigVol_5
xcli -m 10.10.0.10 dm_activate vol=MigVol_6
xcli -m 10.10.0.10 dm_activate vol=MigVol_7
xcli -m 10.10.0.10 dm_activate vol=MigVol_8
xcli -m 10.10.0.10 dm_activate vol=MigVol_9
xcli -m 10.10.0.10 dm_activate vol=MigVol_10

7.5 Manually creating the migration volume

The local XIV volume can be pre-created before defining the data migration object. This is not
the recommended option due to it being prone to manual calculation errors. This requires the
size of the source volume on the non-XIV storage device to be known in 512 byte blocks, as
the two volumes (source and XIV volume) must be exactly the same size. Finding the actual
size of a volume in blocks or bytes can be difficult, as certain storage devices do not show the
exact volume size. This may require you to rely on the host operating system to provide the
real volume size, but this is also not always reliable.

For an example of the process to determine exact volume size, consider ESS 800 volume
00F-FCA33 depicted in Figure 7-23 on page 217. The size reported by the ESS 800 Web GUI
is 10 GB, which suggests that the volume is 10,000,000,000 bytes in size (because the ESS
800 displays volume sizes using decimal counting). The AIX bootinfo -s hdisk2 command
reports the volume as 9,536 GiB, which is 9,999,220,736 bytes (because there are
1,073,741,824 bytes per GiB). Both of these values are too small. When the volume
properties are viewed on the volume information panel of the ESS 800 Copy Services GUI, it
correctly reports the volume as being 19,531,264 sectors, which is 10,000,007,168 bytes
(because there are 512 bytes per sector). If we created a volume that is 19,531,264 blocks in
size this will be correct. When the XIV automatically created a volume to migrate the contents
of 00F-FCA33 it did create it as 19,531,264 blocks. Of the three information sources that were
considered to manually calculate volume size, only one of them must have been correct.
Using the automatic volume creation eliminates this uncertainty.

If you are confident that you have determined the exact size, then when creating the XIV
volume, choose the Blocks option from the Volume Size drop-down menu and enter the size
of the XIV volume in blocks. If your sizing calculation was correct, this creates an XIV volume
that is the same size as the source (non-XIV storage device) volume. Then you can define a
migration:
1. In the XIV GUI go to the floating menu Remote  Migration.
2. Right-click and choose Define Data Migration (Figure 7-9 on page 198).
– Destination Pool: Choose the pool from the drop-down menu where the volume was
created.
– Destination Name: Chose the pre-created volume from the drop-down menu.
– Source Target System: Choose the already defined non-XIV storage device from the
drop-down menu.

Important: If the non-XIV device is active/passive, the source target system must
represent the controller (or service processor) on the non-XIV device that currently
owns the source LUN being migrated. This means that you must check from the
non-XIV storage, which controller is presenting the LUN to the XIV.

Chapter 7. Data migration 207

 – Source LUN: Enter the decimal value of the LUN as presented to the XIV from the
non-XIV storage system. Certain storage devices present the LUN ID as hex. The
number in this field must be the decimal equivalent.
– Keep Source Updated: Check this if the non-XIV storage system source volume is to
be updated with writes from the host. In this manner all writes from the host will be
written to the XIV volume, as well as the non-XIV source volume until the data
migration object is deleted.
Click Define.
3. Test the data migration object. Right-click to select the created data migration volume and
choose Test Data Migration. If there are any issues with the data migration object the test
fails reporting the issue that was found. See Figure 7-11 on page 199 for an example of
the panel.

If the volume that you created is too small or too large you will receive an error message when
you do a test data migration, as shown in Figure 7-16. If you try and activate the migration you
will get the same error message. You must delete the volume that you manually created on
the XIV and create a new correctly sized one. This is because you cannot resize a volume
that is in a data migration pair, and you cannot delete a data migration pair unless it has
completed the background copy. Delete the volume and then investigate why your size
calculation was wrong. Then create a new volume and a new migration and test it again.

Figure 7-16 XIV volume wrong size for migration

7.6 Changing and monitoring the progress of a migration

It is possible to speed up or slow down the migration process, as well as monitor its rate.

208 IBM XIV Storage System: Copy Services and Migration

7.6.1 Changing the synchronization rate
There is only one tunable parameter that determines the speed at which migration data is
transferred between the XIV and defined targets. There are two other tunable parameters that
apply to XIV Remote Mirroring (RM):
򐂰 max_initialization_rate
The rate (in MBps) at which data is transferred between the XIV and defined targets. The
default rate is 100 MBps and can be configured on a per-target basis. In other words, one
target can be set to 100 MBps while another is set to 50 MBps. In this example a total of
150 MBps (100+50) transfer rate is possible. If the transfer rate that you are seeing is
lower than the initialization rate, this may indicate that you are exceeding the capabilities of
the non-XIV disk system to operate at that rate. If the migration is not being done with
attached hosts off-line, consider dropping the initialization rate to a very low number
initially to ensure that there that the volume of migration I/O does not interfere with other
hosts using the non-XIV disk system. Then slowly increase the number while checking to
ensure that response times are not affected on other attached hosts. If you set the
max_initialization_rate to zero, then you will stop the background copy, but hosts will still
be able to access all activated migration volumes.
򐂰 max_syncjob_rate
This parameter (which is in MBps) is used in XIV remote mirroring for synchronizing
mirrored snapshots. It is not normally relevant to data migrations. However, the
max_initialization_rate cannot be greater than the max_syncjob_rate, which in turn cannot
be greater than the max_resync_rate. In general, there is no reason to ever increase this
rate.
򐂰 max_resync_rate
This parameter (which is in MBps) is again used for XIV remote mirroring only. It is not
normally relevant to data migrations. This parameter defines the resync rate for mirrored
pairs. Once remotely mirrored volumes are synchronized, a resync is required if the
replication is stopped for any reason. It is this resync where only the changes are sent
across the link that this parameter affects. The default rate is 300 MBps. There is no
minimum or maximum rate. However, setting the value to 400 or more in a 4 Gbps
environment does not show any increase in throughput. In general, there is no reason to
ever increase this rate.

Increasing the max_initialization_rate parameter may decrease the time required to migrate
the data. However, doing so may impact existing production servers on the non-XIV storage
device. By increasing the rate parameters, more outgoing disk resources will be used to serve
migrations and less for existing production I/O. Be aware of how these parameters affect
migrations as well as production. You could always choose to only set this to a higher value
during off-peak production periods.

Chapter 7. Data migration 209

 The rate parameters can only be set using XCLI, not via the XIV GUI. The current rate
settings are displayed by using the -x parameter, so run the target_list -x command. If the
setting is changed, the change takes place on the fly with immediate effect so there is no
need to deactivate/activate the migrations (doing so blocks host I/O). In Example 7-5 we first
display the target list and then confirm the current rates using the -x parameter. The example
shows that the initialization rate is still set to the default value (100 MBps). We then increase
the initialization rate to 200 MBps. We could then observe the completion rate, as shown in
Figure 7-13 on page 201, to see whether it has improved.

Example 7-5 Displaying and changing the maximum initialization rate

>> target_list
Name SCSI Type Connected
Nextrazap ITSO ESS800 FC yes
>> target_list -x target="Nextrazap ITSO ESS800"
<XCLIRETURN STATUS="SUCCESS" COMMAND_LINE="target_list -x target=&quot;Nextrazap ITSO
ESS800&quot;">
<OUTPUT>
<target id="4502445">
<id value="4502445"/>
<creator value="xiv_maintenance"/>
<creator_category value="xiv_maintenance"/>
<name value="Nextrazap ITSO ESS800"/>
<scsi_type value="FC"/>
<xiv_target value="no"/>
<iscsi_name value=""/>
<connected value="yes"/>
<port_list value="5005076300C90C21,5005076300CF0C21"/>
<num_ports value="2"/>
<system_id value="0"/>
<max_initialization_rate value="100"/>
<max_resync_rate value="300"/>
<max_syncjob_rate value="300"/>
<connectivity_lost_event_threshold value="30"/>
<xscsi value="no"/>
</target>
</OUTPUT>
</XCLIRETURN>
>> target_config_sync_rates target="Nextrazap ITSO ESS800" max_initialization_rate=200
Command executed successfully.

Important: Just because the initialization rate has been increased does not mean that the
actual speed of the copy increases. The outgoing disk system or the SAN fabric may well
be the limiting factor. In addition, you may cause host system impact by over-committing
too much bandwidth to migration I/O.

7.6.2 Monitoring migration speed

If you want to monitor the speed of the migration you can use the Data Migration panel, as
shown in Figure 7-13 on page 201. The status bar can be toggled between GB remaining,
percent complete, or hours/minutes remaining. However, if you change the
max_initialization_rate, you may want to see what affect this change has on the throughput
rate in MBps. To do this you will must use an external tool. This is because the performance
statistics displayed using the XIV GUI or using XIV Top do not include data migration I/O (the

210 IBM XIV Storage System: Copy Services and Migration

 back end copy). They do, however, show incoming I/O rates from hosts using LUNs that are
being migrated.

7.6.3 Monitoring migration via the XIV event log

The XIV event log can be used to confirm when a migration started and finished. From the
XIV GUI go to Monitor  Events. On the Events panel use the Type drop-down menu to
select dm and then click Filter. In Figure 7-17 the events for a single migration are displayed.
In this example the events must be read from bottom to top. You can sort the events by date
and time by clicking the Date column in the Events panel.

Figure 7-17 XIV Event GUI

7.6.4 Monitoring migration speed via the fabric

If you have a Brocade-based SAN, use the portperfshow command and verify the throughput
rate of the initiator ports on the XIV. If you have two fabrics you may need to connect to two
different switches. If multiple paths are defined between XIV and non-XIV disk system, the
XIV load balances across those ports. This means that you must aggregate the throughput
numbers from each initiator port to see total throughput. Example 7-6 shows the output of the
portperfshow command. The values shown are the combined send and receive throughput in
MBps for each port. In this example port 0 is the XIV Initiator port and port 1 is a DS4800 host
port. The max_initialization_rate was set to 50 MBps.

Example 7-6 Brocade portperfshow command

FB1_RC6_PDC:admin> portperfshow
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Total
======================================================================================
50m 50m 14m 14m 2.4m 848k 108k 34k 0 937k 0 27m 3.0m 0 949k 3.0m 125m

If you have a Cisco-based SAN, start Device Manager for the relevant switch and then select
Interface  Monitor  FC Enabled.

7.6.5 Monitoring migration speed via the non-XIV storage

The ability to display migration throughput varies by non-XIV storage device. For example, if
you are migrating from a DS4000 you could use the performance monitoring panels in the
DS4000 System Manager to monitor the throughput. In the DS4000 System Manager GUI, go
to Storage Subsystem  Monitor Performance. Display the volumes being migrated and
the throughput for the relevant controllers. You can then determine what percentage of I/O is

Chapter 7. Data migration 211

 being generated by the migration process. In Figure 7-18 you can see that one volume is
being migrated using a max_initialization_rate of 50 MBps. This represents the bulk of the I/O
being serviced by the DS4000 in this example.

Figure 7-18 Monitoring a DS4000 migration

7.7 Thick-to-thin migration

When the XIV migrates data from a LUN on a non-XIV disk system to an XIV volume, it reads
every block of the source LUN, regardless of contents. However, when it comes to writing this
data into the XIV volume, the XIV only writes blocks that contain data. Blocks that contain only
zeroes are not written and do not take any space on the XIV. This is called a thick-to-thin
migration, and it occurs regardless of whether you are migrating the data into a thin
provisioning pool or a regular pool.

While the migration background copy is being processed, the value displayed in the Used
column of the Volumes and Snapshots panel drops every time that empty blocks are
detected. When the migration is completed, you can check this column to determine how
much real data was actually written into the XIV volume. In Figure 7-19 the used space on the
Windows2003_D volume is 4 GB. However, the Windows file system using this disk shown in
Figure 7-21 on page 214 shows only 1.4 GB of data. This could lead you to conclude wrongly
that the thick-to-thin capabilities of the XIV do not work.

Figure 7-19 Thick-to-thin results

The reason that this has occurred is that when file deletions occur at a file-system level, the
data is not removed. The file system re-uses this effectively free space but does not write
zeros over the old data (as doing so generates a large amount of unnecessary I/O). The end
result is that the XIV effectively copies old and deleted data during the migration. It must be
clearly understood that this makes no difference to the speed of the migration, as these
blocks have to be read into the XIV cache regardless of what they contain.

If you are not planning to use the thin provisioning capability of the XIV, this is not an issue.
Only be concerned if your migration plan specifically requires you to be adopting thin
provisioning.

212 IBM XIV Storage System: Copy Services and Migration

 Writing zeros to recover space
One way to recover space before you start a migration is to use a utility to write zeros across
all free space. In a UNIX environment you could use a simple script like the one shown in
Example 7-7 to write large empty files across your file system. You may need to run these
commands many times to use all the empty space.

Example 7-7 Writing zeros across your file system

# The next command will write a 1 GB mytestfile.out
dd if=/dev/zero of=mytestfile.out bs=1000 count=1000000
# The next command will free the file allocation space
rm mytestfile.out

In a Windows environment you can use a Microsoft tool known as sdelete to write zeros
across deleted files. You can find this tool in the sysinternals section of Microsoft Technet.
Here is the current URL:
http://technet.microsoft.com/en-us/sysinternals/bb897443.aspx

If you instead choose to write zeros to recover space after the migration, you must initially
generate large amounts of empty files, which may initially appear to be counter-productive. It
takes several days for the used space value to decrease after the script or application is run.
This is because recovery of empty space runs as a background task.

7.8 Resizing the XIV volume after migration

Because of the way that XIV distributes data, the XIV allocates space in 17 GB portions
(which are exactly 17,179,869,184 bytes or 16 GiB). When creating volumes using the XIV
GUI this aspect of the XIV design becomes readily apparent when you enter a volume size
and it gets rounded up to the next 17 GB cutoff.

Chapter 7. Data migration 213

 If you chose to allow the XIV to determine the size of the migration volume, then you may find
that a small amount of extra space is consumed for every volume that was created. Unless
the volume sizes being used on the non-XIV storage device were created in multiples of
16 GiB, then it is likely that the volumes automatically created by the XIV will reserve more
XIV disk space than is actually made available to the volume. An example of the XIV volume
properties of such an automatically created volume is shown in Figure 7-20. In this example
the Windows2003_D drive is 53 GB in size, but the size on disk is 68 GB on the XIV.

Figure 7-20 Properties of a migrated volume

What this means is that we can resize that volume to 68 GB (as shown in the XIV GUI) and
make the volume 15 GB larger without effectively consuming any more space on the XIV. In
Figure 7-21 we can see that the migrated Windows2003_D drive is 53 GB in size
(53,678,141,440 bytes).

Figure 7-21 Windows D drive at 53 GB

214 IBM XIV Storage System: Copy Services and Migration

To resize a volume go to the Volumes  Volumes & Snapshots panel, right-click to select
the volume, then choose the Resize option. Change the sizing method drop-down from
Blocks to GB and the volume size is automatically moved to the next multiple of 17 GB. We
can also use XCLI commands, as shown in Example 7-8.

Example 7-8 Resize the D drive using XCLI

>> vol_resize vol=Windows2003_D size=68

Warning: ARE_YOU_SURE_YOU_WANT_TO_ENLARGE_VOLUME Y/N: Y

Command executed successfully.

Because this example is for a Microsoft Windows 2003 basic NTFS disk, we can use the
diskpart utility to extend the volume, as shown in Example 7-9.

Example 7-9 Expanding a Windows volume

C:\>diskpart
DISKPART> list volume

Volume ### Ltr Label Fs Type Size Status Info

---------- --- ----------- ----- ---------- ------- --------- --------
Volume 0 C NTFS Partition 34 GB Healthy System
Volume 4 D Windows2003 NTFS Partition 64 GB Healthy

DISKPART> select volume 4

Volume 4 is the selected volume.
DISKPART> extend
DiskPart successfully extended the volume

We can now confirm that the volume has indeed grown by displaying the volume properties.

In Figure 7-22 we can see that the disk is now 68 GB (68,713,955,328 bytes).

Figure 7-22 Windows 2003 D drive has grown to 64 GB

In terms of when to do the re-size, a volume cannot be resized while it is part of a data
migration. This means that the migration process must have completed and the migration for

Chapter 7. Data migration 215

 that volume must have been deleted before the volume can be resized. For this reason you
may choose to defer the resize until after the migration of all relevant volumes has been
completed. This also separates the resize change from the migration change. Depending on
the operating system using that volume, you may not get any benefit from doing this re-size.

7.9 Troubleshooting
This section lists common errors that are encountered during data migrations using the XIV
data migration facility.

7.9.1 Target connectivity fails

The connections (link line) between the XIV and non-XIV disks system on the migration
connectivity panel remain colored red or the link shows as down. There are several reasons
this can happen:
򐂰 On the Migration Connectivity panel, verify that the status of the XIV initiator port is OK
(Online). If not, check the connections between the XIV and the SAN switch.
򐂰 Verify that the Fibre Channel ports on the non-XIV storage device are enabled and online.
򐂰 Check whether SAN zoning is incorrect or incomplete. Verify the SAN fabric zoning
between the XIV and non-XIV storage device.
򐂰 Perhaps the XIV WWPN is not properly defined to the non-XIV storage device target port.
The XIV WWPN must be defined as a Linux or Windows host.
– If the XIV initiator port is defined as a Linux host to the non-XIV storage device, change
the definition to a Windows host. Delete the link (line connections) between the XIV
and non-XIV storage device ports and redefine the link. This is storage device
dependent and is caused by how the non-XIV storage device presents a pseudo
LUN-0 if a real volume is not presented as LUN 0.
– If the XIV initiator port is defined as a Windows host to the non-XIV storage device,
change the definition to a Linux host. Delete the link (line connections) between the
XIV and non-XIV storage device ports and redefine the link. This is storage device
dependent and is caused by how the non-XIV storage device presents a pseudo
LUN-0 if a real volume is not presented as LUN 0.
– If the above two attempts are not successful, assign a real disk/volume to LUN 0 and
present to the XIV. The volume assigned to LUN-0 can be a very small unused volume
or a real volume that will be migrated.
򐂰 Offline/Online the XIV Fiber channel port: Go to the Migration Connectivity panel, highlight
the port in question, right-click, and choose Configure. Choose No in the second row
drop-down menu (Enabled) and click Configure. Repeat the process, choosing Yes for
Enabled.

216 IBM XIV Storage System: Copy Services and Migration

7.9.2 Remote volume LUN is unavailable
This error typically occurs when defining a DM and the LUN ID specified in the Source LUN
field is not responding to the XIV. This can occur for several reasons:
򐂰 The LUN ID (host LUN ID or SCSI ID) specified is not allocated to the XIV on the ports
identified in the target definition (using the Migration Connectivity panel). You must log on
to the non-XIV storage device to confirm.
򐂰 The LUN ID is not allocated to the XIV on all ports specified in the target definition. For
example, if the target definition has two links from the non-XIV storage device to the XIV,
the volume must be allocated down both paths using the same LUN ID. The XIV looks for
the LUN ID specified on the first defined path. If it does not have access to the LUN it will
fail even if the LUN is allocated down the second path. The LUN must be allocated down
all paths as defined in the target definition. If two links are defined from the target
(non-XIV) storage device to the XIV, then the LUN must be allocated down both paths.
򐂰 Incorrect LUN ID: Do not confuse a non-XIV storage device's internal LUN ID with the
SCSI LUN ID (host LUN ID) that is presented to the XIV. This is a very common oversight.
The source LUN must be the LUN ID (decimal) as presented to the XIV.
򐂰 The Source LUN ID field is expecting a decimal number. Certain vendors present the LUN
ID in hex. This must be translated to decimal. Therefore, if LUN ID 10 is on a vendor that
displays its IDs in hex, the LUN ID in the DM define is 16 (hex 10). An example of a
hexadecimal LUN number is shown in Figure 7-23, taken from an ESS 800. In this
example you can see LUN 000E, 000F, and 0010. These are entered into the XIV data
migration definitions as LUNs 14, 15, and 16, respectively. See 7.12, “Device-specific
considerations” on page 223, for more details.
򐂰 The LUN ID allocated to the XIV has been allocated to an incorrect XIV WWPN. Make
sure that the proper volume is allocated to the correct XIV WWPNs.
򐂰 If multiple DM targets are defined, the wrong target may have been chosen when the DM
was defined.
򐂰 Sometimes when volumes are added after the initial connectivity is defined the volume is
not available. Go to the Migration Connectivity panel and delete the links between the XIV
and non-XIV storage device. Only delete the links. There is no need to delete anything
else. Once all links are deleted, recreate the links. Go back to the DM panel and recreate
the DM. (See item 5 under in “Define non-XIV storage device to XIV (as a migration
target)” on page 193).

Figure 7-23 ESS 800 LUN numbers

Chapter 7. Data migration 217

 򐂰 The volume on the source non-XIV storage device may not have been initialized or
low-level formatted. If the volume has data on it then this is not the case. However, if you
are assigning new volumes from the non-XIV storage device then perhaps these new
volumes have not completed the initialization process. On ESS 800 storage the
initialization process can be displayed from the Modify Volume Assignments panel. In
Figure 7-23 on page 217 the volumes are still 0% background formatted, so they will not
be accessible by the XIV. So for ESS 800, keep clicking Refresh Status on the ESS 800
Web GUI until the formatting message disappears.

7.9.3 Local volume is not formatted

This error occurs when a volume that already exists is chosen as the destination name and
has already been written either from a host or a previous DM process that has since been
removed from the DM panel. To get around this error do one of the following tasks:
򐂰 Use another volume as a migration destination.
򐂰 Delete the volume that you are trying to migrate to and then create it again.
򐂰 Go to the Volumes  Volumes and Snapshots panel. Right-click to select the volume
and choose Format. Warning: This deletes all data currently on the volume without
recovery. A warning message is displayed to challenge the request.

7.9.4 Host server cannot access the XIV migration volume

This error occurs if you attempt to read the contents of a volume on a non-XIV storage device
via an XIV data migration without activating the data migration. This happens if you perform
the migration without following the correct order of steps, in that the host will not attempt to
access the volume until the XIV shows that the migration is initializing and active (even if the
progress percentage only shows 0%) or fully synchronized.

7.9.5 Remote volume cannot be read

This error occurs when a volume is defined down the passive path on an active/passive
multi-pathing storage device. This can occur in several cases:
򐂰 Two paths were defined on a target (non-XIV storage device) that only supports
active/passive multi-pathing. XIV is an active/active storage device. Defining two paths on
any given target from an active/passive multi-pathing storage device is not supported.
Redefine the target with only one path. Another target can be defined with one connection
to the other controller. For example, if the non-XIV storage device has two controllers, but
the volume can only be active on one at time, controller A can be defined as one target on
the XIV and controller B can be defined as a different target. In this manner, all volumes
that are active on controller A can be migrated down the XIV A target and all volumes
active on the B controller can be migrated down the XIV B target.
򐂰 When defining the XIV initiator to an active/passive multi-pathing non-XIV storage device,
certain storage devices allow the initiator to be defined as not supporting failover. The XIV
initiator should be configured to the non-XIV storage device in this manner. When
configured as such, the volume on the passive controller is not presented to the initiator
(XIV). The volume is only presented down the active controller.

Refer to “Multi-pathing with data migrations” on page 188 and 7.12, “Device-specific
considerations” on page 223, for additional information.

218 IBM XIV Storage System: Copy Services and Migration

7.9.6 LUN is out of range
XIV currently supports migrating data from LUNs with a LUN ID less than 513 (decimal). This
is usually not an issue, as most non-XIV storage devices, by default, present volumes on an
initiator basis. For example, if there are three hosts connected to the same port on a non-XIV
storage device, each host can be allocated volumes starting at the same LUN ID. So for
migration purposes you must either map one host at a time (and then re-use the LUN IDs for
the next host) or use different sequential LUN numbers for migration. For example, if three
hosts each have three LUNs, each host’s LUNs mapped using LUN IDs 20, 21, and 22, then
for migration purposes, migrate them as LUN IDs 30, 31, 32 (first host); 33, 34, 35 (second
host); and 36, 37, 38 (third host). Then from the XIV you can again map them to each host as
LUN IDs 20, 21, and 22 (as they were from the non-XIV storage).

If migrating from an EMC Symmetrix or DMX there are special considerations. Refer to
7.12.2, “EMC Symmetrix and DMX” on page 224.

7.10 Backing out of a data migration

For change management purposes, you may be required to document a back-out procedure.
There are four possible points in the migration process where a back-out may occur.

7.10.1 Back-out prior to migration being defined on the XIV

If a data migration definition does not exist yet, then no action must be taken on the XIV. You
can simply zone the host server back to the non-XIV storage system and un-map the host
server’s LUNs away from the XIV and back to the host server, taking care to ensure that the
correct LUN order is preserved.

7.10.2 Back-out after a data migration has been defined but not activated
If the data migration definition exists but has not been activated, then you can follow the same
steps as described in 7.10.1, “Back-out prior to migration being defined on the XIV” on
page 219. To remove the inactive migration from the migration list you must delete the XIV
volume that was going to receive the migrated data.

7.10.3 Back-out after a data migration has been activated but is not complete
If the data migration shows in the GUI with a status of initialization or the XCLI shows it as
active=yes, then the background copy process has been started. If you deactivate the
migration in this state you will block any I/O passing through the XIV from the host server to
the migration LUN on the XIV and to the LUN on the non-XIV disk system. You must shut
down the host server or its applications first. After doing this you can deactivate the data
migration and then if desired you can delete the XIV data migration volume. Then restore the
original LUN masking and SAN fabric zoning and bring your host back up.

Important: If you chose to not allow source updating and write I/O has occurred after the
migration started, then the contents of the LUN on the non-XIV storage device will not
contain the changes from those writes. Understanding the implications of this is important
in a back-out plan.

Chapter 7. Data migration 219

7.10.4 Back-out after a data migration has reached the synchronised state
If the data migration shows in the GUI as having a status of synchronised, then the
background copy has completed. In this case back-out can still occur because the data
migration is not destructive to the source LUN on the non-XIV storage device. Simply reverse
the process by shutting down the host server or applications and restore the original LUN
masking and switch zoning settings. You may need to also reinstall the relevant host server
multi-path software for access to the non-XIV storage device.

Important: If you chose to not allow source updating and write I/O has occurred during the
migration or after it has completed, then the contents of the LUN on the non-XIV storage
device do not contain the changes from those writes. Understanding the implications of this
is important in a back-out plan.

7.11 Migration checklist

There are three separate stages to a migration cut over. First, prepare the environment for the
implementation of the XIV. Second, cut over your hosts. Finally, remove any old devices and
definitions as part of a clean up stage.

For site setup, the high-level process is:

1. Install XIV and cable it into the SAN.
2. Pre-populate SAN zones in switches.
3. Pre-populate the host/cluster definitions in the XIV.
4. Define XIV to non-XIV disk as a host.
5. Define non-XIV disk to XIV as a migration target and confirm paths.

Then for each host the high-level process is:

1. Update host drivers and then shut down the host.
2. Disconnect/un-zone the host from non-XIV storage and then zone the host to XIV.
3. Map the host LUNs away from the host instead of mapping them to the XIV.
4. Create XIV data migration (DM).
5. Map XIV DM volumes to the host.
6. Bring up the host.

When all data on the non-XIV disk system has been migrated, perform site clean up:
1. Delete all SAN zones related to the non-XIV disk.
2. Delete all LUNs on non-XIV disk and remove it from the site.

220 IBM XIV Storage System: Copy Services and Migration

 Table 7-1 shows the site setup checklist.

Table 7-1 Physical site setup

Task Completed Where to Task
Number perform

1 Site Install XIV.

2 Site Run fiber cables from SAN switches to XIV for host connections
and migration connections.

3 Non-XIV storage Select host ports on the non-XIV storage to be used for migration
traffic. These ports do not have to be dedicated ports. Run new
cables if necessary.

4 Fabric switches Create switch aliases for each XIV Fibre Channel port and any
new non-XIV ports added to the fabric.

5 Fabric switches Define SAN zones to connect hosts to XIV (but do not activate the
zones). You can do this by cloning the existing zones from host to
non-XIV disk and swapping non-XIV aliases for new XIV aliases.

6 Fabric switches Define and activate SAN zones to connect non-XIV storage to XIV
initiator ports (unless direct connected).

7 Non-XIV storage If necessary, create a small LUN to be used as LUN0 to allocate

to the XIV.

8 Non-XIV storage Define the XIV on the non-XIV storage device, mapping LUN0 to
test the link.

9 XIV Define non-XIV storage to the XIV as a migration target and add
ports. Confirm that links are green and working.

10 XIV Change the max_initialization_rate depending on the non-XIV

disk. You may want to start at a smaller value and increase it if no
issues are seen.

11 XIV Define all the host servers to the XIV (cluster first if using clustered
hosts). Use a host listing from the non-XIV disk to get the WWPNs
for each host.

12 XIV Create storage pools as required. Ensure that there is enough

pool space for all the non-XIV disk LUNs being migrated.

Once the site setup is complete, the host migrations can begin. Table 7-2 shows the host
migration check list. Repeat this check list for every host. Task numbers that are colored red
must be performed with the host application offline.

Table 7-2 Host Migration to XIV task list

Task Completed? Where to Task
number perform

1 Host From the host, determine the volumes to be migrated and their relevant LUN
IDs and hardware serial numbers or identifiers.

2 Host If the host is remote from your location, confirm that you can power the host
back on after shutting it down (using tools such as an RSA card or
BladeCenter® manager).

3 Non-XIV Get the LUN IDs of the LUNs to be migrated from non-XIV storage device.
Storage Convert from hex to decimal if necessary.

Chapter 7. Data migration 221

 Task Completed? Where to Task
number perform

4 Host Shut down the application.

5 Host Set the application to not start automatically at reboot. This helps when
performing administrative functions on the server (upgrades of drivers, patches,
and so on).

6 Host UNIX servers: Comment out disk mount points on affected disks in the mount
configuration file. This helps with system reboots while configuring for XIV.

7 Host Shut down affected servers.

8 Fabric Change the active zoneset to exclude the SAN zone that connects the host
server to non-XIV storage and include the SAN zone for the host server to XIV
storage. The new zone should have been created during site setup.

9 Non-XIV Unmap source volumes from the host server.

storage

10 Non-XIV Map source volumes to the XIV host definition (created during site setup).
storage

11 XIV Create data migration pairing (XIV volumes created on the fly).

12 XIV Test XIV migration for each volume.

13 XIV Start XIV migration and verify it. If you want, wait for migration to finish.

14 Host Boot the server. (Be sure that the server is not attached to any storage.)

15 Host If co-existence of non-XIV and XIV multi-path software is not allowed, remove
non-XIV multi-pathing software.

16 Host Install patches, update drivers, and HBA firmware as necessary.

17 Host Install the XIV Host Attachment Kit. (Be sure to note prerequisites.)

18 Host Shut down the host server.

19 XIV Map XIV volumes to the host server. (Use original LUN IDs.)

20 Host Boot the server.

21 Host Verify that the LUNs are available and that pathing is correct.

22 Host UNIX servers: Update mount points for new disks in the mount configuration file
if they have changed.

23 Host Start the application.

24 Host Set the application to start automatically if this was previously changed.

25 XIV Monitor the migration if it is not already completed.

26 XIV When the volume is synchronized delete the data migration (do not deactivate
the migration).

27 Non-XIV Un-map migration volumes away from XIV if you must free up LUN IDs.
Storage

28 XIV Consider re-sizing the migrated volumes to the next 17 GB boundary if the host
operating system is able to use new space on a re-sized volume.

29 Host If XIV volume was re-sized, use host procedures to utilize the extra space.

222 IBM XIV Storage System: Copy Services and Migration

Task Completed? Where to Task
number perform

30 Host If non-XIV storage device drivers and other supporting software were not
removed earlier, remove them when convenient.

When all the hosts and volumes have been migrated there are two site clean up tasks left, as
shown in Table 7-3.

Table 7-3 Site cleanup check list

Task number Completed? Where to perform Task

1 XIV Delete migration paths

and targets.

2 Fabric Delete all zones

related to non-XIV
storage including the
zone for XIV migration.

3 Non-XIV storage Delete all LUNs and

perform secure data
destruction if required.

7.12 Device-specific considerations

The XIV supports migration from practically any SCSI storage device that has Fibre Channel
interfaces. This section contains device-specific information, but it is not an exhaustive list.
Ensure that the following requirements are understood for your storage device:
LUN0 Do we need to specifically map a LUN to LUN ID zero? This
determines whether you will have a problem defining the paths.
LUN numbering Does the storage device GUI or CLI use decimal or hexadecimal LUN
numbering? This determines whether you must do a conversion when
entering LUN numbers into the XIV GUI.
Multipathing Is the device active/active or active/passive? This determines whether
you define the storage device as a single target or as one target per
internal controller or service processor.
Definitions Does the device have specific requirements when defining hosts?

Converting hexadecimal LUN IDs to decimal LUN IDs

When mapping volumes to the XIV it is very important to note the LUN IDs allocated by the
non-XIV storage. The methodology to do this varies by vendor and device. If the device uses
hexadecimal LUN numbering then it is also important to understand how to convert
hexadecimal numbers into decimal numbers, to enter into the XIV GUI.

Using a spreadsheet to convert hex to decimal

Microsoft Excel and Open Office both have a spreadsheet formula known as hex2dec. If, for
example, you enter a hexadecimal value into spreadsheet cell location A4, then the formula to
convert the contents of that cell to decimal is =hex2dec(A4). If this formula does not appear to
work in Excel then add the Analysis ToolPak (within Excel go to the Tools menu  Add ins 
Select Analysis ToolPak).

Chapter 7. Data migration 223

 Using Microsoft calculator to convert hex to decimal
Start the calculator with the following steps:
1. Selecting Program Files  Programs  Accessories  Calculator.
2. From the View drop-down menu change from Standard to Scientific.
3. Select Hex.
4. Enter a hexadecimal number and then select Dec.

The hexadecimal number will have been converted to decimal.

Given that the XIV supports migration from almost any storage device, it is impossible to list
the methodology to get LUN IDs from each one.

7.12.1 EMC CLARiiON

The following considerations were identified specifically for EMC CLARiiON:
򐂰 LUN0
There is no requirement to map a LUN to LUN ID 0 for the CLARiiON to communicate with
the XIV.
򐂰 LUN numbering
The EMC CLARiiON uses decimal LUN numbers for both the CLARiiON ID and the host
ID (LUN number).
򐂰 Multipathing
The EMC CLARiiON is an active/passive storage device. This means that each storage
processor (SP-A and SP-B) must be defined as a separate target to the XIV. You could
choose to move LUN ownership of all the LUNs that you are migrating to a specific SP and
simply define only that SP as a target. Moving a LUN away from the SP that owns it is
known as trespassing. Have only one path to each SP.
򐂰 Requirements when defining the XIV
If migrating from an EMC CLARiiON use the settings shown in Table 7-4 to define the XIV
to the CLARiiON. Ensure that Auto-trespass is disabled for every XIV initiator port
(WWPN) registered to the Clariion.
Table 7-4 Defining an XIV to the EMC CLARiiON
Initiator information Recommended setting

Initiator type CLARiiON Open

HBA type Host

Array CommPath Enabled

Failover mode 0

Unit serial number Array

7.12.2 EMC Symmetrix and DMX

The considerations discussed in this section were identified specifically for EMC Symmetrix
and DMX.

224 IBM XIV Storage System: Copy Services and Migration

LUN0
There is a requirement for the EMC Symmetrix or DMX to present a LUN ID 0 to the XIV in
order for the XIV Storage System to communicate with the EMC Symmetrix or DMX. In many
installations, the VCM device is allocated to LUN-0 on all FAs and is automatically presented
to all hosts. In these cases, the XIV connects to the DMX with no issues. However, in newer
installations, the VCM device is no longer presented to all hosts and therefore a real LUN-0 is
required to be presented to the XIV in order for the XIV to connect to the DMX. This LUN-0
can be a dummy device of any size that will not be migrated or an actual device that will be
migrated.

LUN numbering
The EMC Symmetrix and DMX, by default, does not present volumes in the range of 0 to 512
decimal. The Symmetrix/DMX presents volumes based on the LUN ID that was given the
volume when the volume was placed on the FA port. If a volume was placed on the FA with a
LUN ID of 90, this is how it is presented to the host by default. The Symmetrix/DMX also
presents the LUN IDs in hex. Thus, LUN ID 201 equates to decimal 513, which is greater than
512 and is outside of the XIV's range. There are two disciplines for migrating data from a
Symmetrix/DMX where the LUN ID is greater than 512 (decimal).

Re-map the volume

One way to migrate a volume with a LUN ID higher than 512 is to re-map the volume in one of
two ways:
򐂰 Map the volume to a free FA or an FA that has available LUN ID slots less than hex 200
(decimal 512). In most cases this can be done without interruption to the production
server. The XIV is zoned and the target defined to the FA port with the lower LUN ID.
򐂰 Re-map the volume to a lower LUN ID, one that is less than 200 hex. However, this
requires that the host be shut down while the change is taking place and is therefore not
the best option.

LUN-Offset
With EMC Symmetrix Enginuity code 68 - 71 code, there is an EMC method of presenting
LUN IDs to hosts other than the LUN ID given to the volume when placed on the FA. In the
Symmetrix/DMX world, a volume is given a unique LUN ID when configured on an FA. Each
volume on an FA must have a unique LUN ID. The default method (and a best practice of
presenting volumes to a host) is to use the LUN ID given to the volume when placed on the
FA. In other words, if 'vol1' was placed on an FA with an ID of 7A (hex (0x07a) decimal 122),
this is the LUN ID that is presented to the host. Using the lunoffset option of the symmask
command, a volume can be presented to a host (WWPN initiator) with a different LUN ID than
was assigned the volume when placed on the FA. Because it is done at the initiator level, the
production server can keep the high LUNs (above 128) while being allocated to the XIV using
lower LUN IDs (below 512 decimal).

Migrating volumes that were used by HP-UX

For HP-UX hosts attached to EMC Symmetrix there is a setting known as
Volume_Set_Addressing that can be enabled on a per-FA basis. This is required for HP-UX
host connectivity but is not compatible with any other host types (including XIV). If
Volume_Set_Addressing (also referred to as the V bit setting) is enabled on an FA, then the
XIV will not be able to access anything but LUN 0 on that FA. To avoid this issue, map the
HP-UX host volumes to a different FA that is not configured specifically for HP-UX. Then zone
the XIV migration port to this FA instead of the FA being used by HP-UX.

Multipathing
The EMC Symmetrix and DMX are active/active storage devices.

Chapter 7. Data migration 225

7.12.3 HDS TagmaStore USP
In this section we discuss HDS TagmaStore USP.

LUN0
There is a requirement for the HDS TagmaStore Universal Storage Platform (USP) to present
a LUN ID 0 to the XIV in order for the XIV Storage System to communicate with the HDS
device.

LUN numbering
The HDS USP uses hexadecimal LUN numbers.

Multipathing
The HDS USP is an active/active storage device.

7.12.4 HP EVA
The following requirements were determined after migration from a HP EVA 4400 and 8400.

LUN0
There is no requirement to map a LUN to LUN ID 0 for the HP EVA to communicate with the
XIV. This is because by default the HP EVA presents a special LUN known as the Console
LUN as LUN ID 0.

LUN numbering
The HP EVA uses decimal LUN numbers.

Multipathing
The HP EVA 4000/6000/8000 are active/active storage devices. For HP EVA 3000/5000, the
initial firmware release was active/passive, but a firmware upgrade to VCS Version 4.004
made it active/active capable. For more details see the following Web site:
http://h21007.www2.hp.com/portal/site/dspp/menuitem.863c3e4cbcdc3f3515b49c108973a8
01?ciid=aa08d8a0b5f02110d8a0b5f02110275d6e10RCRD

Requirements when connecting to XIV

Define the XIV as a Linux host.

To check the LUN IDs assigned to a specific host:

1. Log in into Command View EVA.h.
2. Select the storage on which you are working.
3. Click the Hosts icon.
4. Select the specific host.
5. Click the Presentation tab.
6. Here you will see the LUN name and the LUN ID presented.

To present EVA LUNs to XIV:

1. Create the host alias for XIV and add the XIV initiator ports that are zoned to EVA.
2. From the Command View EVA, select the active Vdisk that must be presented to XIV.
3. Click the Presentation tab.

226 IBM XIV Storage System: Copy Services and Migration

 4. Click Present.
5. Select the XIV host Alias created.
6. Click the Assign LUN button on top.
7. Specify the LUN ID that you want to specify for XIV. Usually this is the same as was
presented to the host when it was accessing the EVA.

7.12.5 IBM DS3000/DS4000/DS5000

The following considerations were identified specifically for DS4000 but apply for all models of
D3000, DS4000, and DS5000 (for purposes of migration they are functionally all the same).
For ease of reading, only the DS4000 is referenced.

LUN0
There is a requirement for the DS4000 to present a LUN on LUN ID 0 to the XIV to allow the
XIV to communicate with the DS4000. It may be easier to create a new 1 GB LUN on the
DS4000 just to satisfy this requirement. This LUN does not need to have any data on it.

LUN numbering
For all DS4000 models, the LUN ID used in mapping is a decimal value between 0 to 15 or 0
to 255 (depending on model). This means that no hex-to-decimal conversion is necessary.
Figure 7-8 on page 197 shows an example of how to display the LUN IDs.

Defining the DS4000 to the XIV as a target

The DS4000 is an active/passive storage device. This means that each controller on the
DS4000 must be defined as a separate target to the XIV. You must take note of which
volumes are currently using which controllers as the active controller.

Preferred path errors

The following issues can occur if you have misconfigured a migration from a DS4000. You
may initially notice that the progress of the migration is very slow. The DS4000 event log may
contain errors, such as the one shown in Figure 7-24. If you see the migration volume fail
between the A and B controllers, this means that the XIV is defined to the DS4000 as a host
that supports ADT/RDAC (which you should immediately correct) and that either the XIV
target definitions have paths to both controllers or that you are migrating from the wrong
controller.

Figure 7-24 DS4000 LUN fail over

Chapter 7. Data migration 227

 In Example 7-10 the XCLI commands show that the target called ITSO_DS4700 has two ports,
one from controller A (201800A0B82647EA) and one from controller B (201900A0B82647EA).
This is not the correct configuration and should not be used.

Example 7-10 Incorrect definition, as target has ports to both controllers

>> target_list
Name SCSI Type Connected
ITSO_DS4700 FC yes

>> target_port_list target=ITSO_DS4700_Ctrl_B

Target Name Port Type Active WWPN iSCSI Address iSCSI Port
ITSO_DS4700 FC yes 201800A0B82647EA 0
ITSO_DS4700 FC yes 201900A0B82647EA 0

Instead, two targets should have been defined, as shown in Example 7-11. In this example,
two separate targets have been defined, each target having only one port for the relevant
controller.

Example 7-11 Correct definitions for a DS4700

> target_list
Name SCSI Type Connected
ITSO_DS4700_ctrl_A FC yes
ITSO_DS4700_ctrl_B FC yes

>> target_port_list target=ITSO_DS4700_Ctrl_A

Target Name Port Type Active WWPN iSCSI Address iSCSI Port
ITSO_DS4700 FC yes 201800A0B82647EA 0

>> target_port_list target=ITSO_DS4700_Ctrl_A

Target Name Port Type Active WWPN iSCSI Address iSCSI Port
ITSO_DS4700 FC yes 201900A0B82647EA 0

Defining the XIV to the DS4000 as a host

Use the DS Storage Manager to check the profile of the DS4000 and select a host type for
which ADT is disabled or failover mode is RDAC. To display the profile from the DS Storage
Manager choose Storage Subsystem  View  Profile  All. Then go to the bottom of the
Profile panel. The profile may vary according to NVRAM version. In Example 7-12 select the
host type for which ADT status is disabled (Windows 2000).

Example 7-12 Earlier NVRAM versions

HOST TYPE ADT STATUS
Linux Enabled
Windows 2000/Server 2003/Server 2008 Non-Clustered Disabled

In Example 7-13 choose the host type that specifies RDAC (Windows 2000).

Example 7-13 Later NVRAM versions

HOST TYPE FAILOVER MODE
Linux ADT
Windows 2000/Server 2003/Server 2008 Non-Clustered RDAC

228 IBM XIV Storage System: Copy Services and Migration

 You can now create a host definition on the DS4000 for the XIV. If you have zoned the XIV to
both DS4000 controllers you can add both XIV initiator ports to the host definition. This
means that the host properties should look similar to Figure 7-25. After mapping your
volumes to the XIV migration host, you must take note of which controller each volume is
owned by. When you define the data migrations on the XIV, the migration should point to the
target that matches the controller that owns the volume being migrated.

Figure 7-25 XIV defined to the DS4000 as a host.

7.12.6 IBM ESS E20/F20/800

The following considerations were identified for ESS 800.

LUN0
There is no requirement to map a LUN to LUN ID 0 for the ESS to communicate with the XIV.

LUN numbering
The LUN IDs used by the ESS are in hexadecimal, so they must be converted to decimal
when entered as XIV data migrations. It is not possible to specifically request certain LUN
IDs. In Example 7-14 there are 18 LUNs allocated by an ESS 800 to an XIV host called
NextraZap_ITSO_M5P4. You can clearly see that the LUN IDs are hex. The LUN IDs given in
the right-hand column were added to the output to show the hex-to-decimal conversion
needed for use with XIV. An example of how to view LUN IDs using the ESS 800 Web GUI is
shown in Figure 7-23 on page 217.

Restriction: The ESS can only allocate LUN IDs in the range 0 to 255 (hex 00 to FF). This
means that only 256 LUNs can be migrated at one time.

Example 7-14 Listing ESS 800 LUN IDs using ESSCLI

C:\esscli -s 10.10.1.10 -u storwatch -p specialist list volumeaccess -d
"host=NextraZap_ITSO_M5P4"
Tue Nov 03 07:20:36 EST 2009 IBM ESSCLI 2.4.0

Volume LUN Size(GB) Initiator Host

------ ---- -------- ---------------- -------------------
100e 0000 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 0)
100f 0001 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 1)
1010 0002 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 2)

Chapter 7. Data migration 229

 1011 0003 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 3)
1012 0004 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 4)
1013 0005 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 5)
1014 0006 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 6)
1015 0007 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 7)
1016 0008 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 8)
1017 0009 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 9)
1018 000a 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 10)
1019 000b 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 11)
101a 000c 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 12)
101b 000d 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 13)
101c 000e 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 14)
101d 000f 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 15)
101e 0010 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 16)
101f 0011 10.0 5001738000230153 NextraZap_ITSO_M5P4 (LUN ID is 17)

Multipathing
The ESS 800 is an active/active storage device. You can define multiple paths from the XIV to
the ESS 800 for migration. Ideally, connect to more than one host bay in the ESS 800.
Because each XIV host port is defined as a separate host system, ensure that the LUN ID
used for each volume is the same. There is a check box on the Modify Volume Assignments
panel titled “Use same ID/LUN in source and target” that will assist you. Figure 7-28 on
page 236 shows a good example of two XIV host ports with the same LUN IDs.

Requirements when defining the XIV

Define each XIV host port to the ESS 800 as a Linux x86 host.

7.12.7 IBM DS6000 and DS8000

The following considerations were identified for DS6000 and DS8000.

LUN0
There is no requirement to map a LUN to LUN ID 0 for a DS6000 or DS8000 to communicate
with the XIV.

LUN numbering
The DS6000 and DS8000 use hexadecimal LUN IDs. These can be displayed using DSCLI
with the showvolgrp -lunmap xxx command, where xxx is the volume group created to assign
volumes to the XIV for data migration. Do not use the Web GUI to display LUN IDs.

Multipathing with DS6000

The DS6000 is an active/active storage device, but each controller has dedicated host ports,
whereas each LUN has a preferred controller. If I/O for a particular LUN is sent to host ports
of the non-preferred controller, the LUN will not fail over, but that I/O may experience a small
performance penalty. This may lead you to consider migrating volumes with even LSS
numbers (such as volumes 0000 and 0200) from the upper controller and volumes with odd
LSS numbers (such as volumes 0100 and 0300) from the lower controller. However, this is not
a robust solution. Define the DS6000 as a single target with one path to each controller.

230 IBM XIV Storage System: Copy Services and Migration

 Multipathing with DS8000
The DS8000 is an active/active storage device. You can define multiple paths from the XIV to
the DS8000 for migration. Ideally, connect to more than one I/O bay in the DS8000.

Requirements when defining the XIV

In Example 7-15 a volume group is created used a type of SCSI Map 256, which is the correct
type for a RedHat Linux host type. A starting LUN ID of 8 is chosen to show how hexadecimal
numbering is used. The range of valid LUN IDs for this volume group are 0 to FF (0 to 255 in
decimal). An extra LUN is then added to the volume group to show how specific LUN IDs can
be selected by volume. Two host connections are then created using the Red Hat Linux host
type. By using the same volume group ID for both connections, we ensure that the LUN
numbering used by each defined path will be the same.

Example 7-15 Listing DS6000 and DS8000 LUN IDs

dscli> mkvolgrp -type scsimap256 -volume 0200-0204 -LUN 8 migrVG
CMUC00030I mkvolgrp: Volume group V18 successfully created.
dscli> chvolgrp -action add -volume 0205 -lun 0E V18
CMUC00031I chvolgrp: Volume group V18 successfully modified.
dscli> showvolgrp -lunmap V18
Name migrVG
ID V18
Type SCSI Map 256
Vols 0200 0201 0202 0203 0204 0205
==============LUN Mapping===============
vol lun
========
0200 08 (comment: use decimal value 08 in XIV GUI)
0201 09 (comment: use decimal value 09 in XIV GUI)
0202 0A (comment: use decimal value 10 in XIV GUI)
0203 0B (comment: use decimal value 11 in XIV GUI)
0204 0C (comment: use decimal value 12 in XIV GUI)
- 0D
0205 0E (comment: use decimal value 14 in XIV GUI)
dscli> mkhostconnect -wwname 5001738000230153 -hosttype LinuxRHEL -volgrp V18 XIV_M5P4
CMUC00012I mkhostconnect: Host connection 0020 successfully created.
dscli> mkhostconnect -wwname 5001738000230173 -hosttype LinuxRHEL -volgrp V18 XIV_M7P4
CMUC00012I mkhostconnect: Host connection 0021 successfully created.
dscli> lshostconnect
Name ID WWPN HostType Profile portgrp volgrpID
===========================================================================================
XIV_M5P4 0020 5001738000230153 LinuxRHEL Intel - Linux RHEL 0 V18
XIV_M7P4 0021 5001738000230173 LinuxRHEL Intel - Linux RHEL 0 V18

7.13 Sample migration

Here is a specific example migration.

Chapter 7. Data migration 231

 Using XIV DM to migrate an AIX file system from ESS 800 to XIV
In this example we migrate a file system on an AIX host using ESS 800 disks to XIV. First we
select a volume group to migrate. In Example 7-16 we select a volume group called
ESS_VG1. The lsvg command shows that this volume group has one file system mounted on
/mnt/redbk. The df -k command shows that the file system is 20 GiB in size and is 46%
used.

Example 7-16 Selecting a file system

root@dolly:/mnt/redbk# lsvg -l ESS_VG1
ESS_VG1:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
loglv00 jfs2log 1 1 1 open/syncd N/A
fslv00 jfs2 20 20 3 open/syncd /mnt/redbk
root@dolly:/mnt/redbk# df -k
Filesystem 1024-blocks Free %Used Iused %Iused Mounted on
/dev/fslv00 20971520 11352580 46% 17 1% /mnt/redbk

We now determine which physical disks must be migrated. In Example 7-17 we use the lspv
commands to determine that hdisk3, hdisk4, and hdisk5 are the relevant disks for this VG.
The lsdev -Cc disk command confirms that they are located on an IBM ESS 2105. We then
use the lscfg command to determine the hardware serial numbers of the disks involved.

Example 7-17 Determine the migration disks

root@dolly:/mnt/redbk# lspv
hdisk1 0000d3af10b4a189 rootvg active
hdisk3 0000d3afbec33645 ESS_VG1 active
hdisk4 0000d3afbec337b5 ESS_VG1 active
hdisk5 0000d3afbec33922 ESS_VG1 active
root@dolly:~/sddpcm# lsdev -Cc disk
hdisk0 Available 11-08-00-2,0 Other SCSI Disk Drive
hdisk1 Available 11-08-00-4,0 16 Bit LVD SCSI Disk Drive
hdisk2 Available 11-08-00-4,1 16 Bit LVD SCSI Disk Drive
hdisk3 Available 17-08-02 IBM MPIO FC 2105
hdisk4 Available 17-08-02 IBM MPIO FC 2105
hdisk5 Available 17-08-02 IBM MPIO FC 2105

root@dolly:/mnt# lscfg -vpl hdisk3 | egrep "Model|Serial"

Machine Type and Model......2105800
Serial Number...............00FFCA33
root@dolly:/mnt# lscfg -vpl hdisk4 | egrep "Model|Serial"
Machine Type and Model......2105800
Serial Number...............010FCA33
root@dolly:/mnt# lscfg -vpl hdisk5 | egrep "Model|Serial"
Machine Type and Model......2105800
Serial Number...............011FCA33

232 IBM XIV Storage System: Copy Services and Migration

These volumes are currently allocated from an IBM ESS 800. In Figure 7-26 we use the ESS
Web GUI to confirm that the volume serial numbers match with those determined in
Example 7-17 on page 232. Note that the LUN IDs here are those used by ESS 800 with AIX
hosts (IDs 500F, 5010, and 5011). They are not correct for the XIV and will be changed when
we re-map them to the XIV.

Figure 7-26 LUNs allocated to AIX from the ESS 800

Because we now know the source hardware we can create connections between the ESS
800 and the XIV and the XIV and Dolly (our host server). First, in Example 7-18 we identify
the existing zones that connect Dolly to the ESS 800. We have two zones, one for each AIX
HBA. Each zone contains the same two ESS 800 HBA ports.

Example 7-18 Existing zoning on the SAN Fabric

zone: ESS800_dolly_fcs0
10:00:00:00:c9:53:da:b3
50:05:07:63:00:c9:0c:21
50:05:07:63:00:cd:0c:21
zone: ESS800_dolly_fcs0
10:00:00:00:c9:53:da:b2
50:05:07:63:00:c9:0c:21
50:05:07:63:00:cd:0c:21

We now create two new zones. The first zone connects the initiator ports on the XIV to the
ESS 800. The second and third zones connects the target ports on the XIV to Dolly (for use
after the migration). These are shown in Example 7-19. All six ports on the XIV clearly must
have been cabled into the SAN fabric.

Example 7-19 New zoning on the SAN Fabric

zone: ESS800_nextrazap
50:05:07:63:00:c9:0c:21
50:05:07:63:00:cd:0c:21
50:01:73:80:00:23:01:53
50:01:73:80:00:23:01:73
zone: nextrazap_dolly_fcs0
10:00:00:00:c9:53:da:b3
50:01:73:80:00:23:01:41
50:01:73:80:00:23:01:51

Chapter 7. Data migration 233

 zone: nextrazap_dolly_fcs1
10:00:00:00:c9:53:da:b2
50:01:73:80:00:23:01:61
50:01:73:80:00:23:01:71

We then create the migration connections between the XIV and the ESS 800. An example of
using the XIV GUI to do this was shown in “Define target connectivity (Fibre Channel only).”
on page 203. In Example 7-20 we use the XCLI to define a target, then the ports on that
target, then the connections between XIV and the target (ESS 800). Finally, we check that the
links are active=yes and up=yes. We can use two ports on the ESS 800 because it is an
active/active storage device.

Example 7-20 Connecting ESS 800 to XIV for migration using XCLI
>> target_define protocol=FC target=ESS800 xiv_features=no
Command executed successfully.
>> target_port_add fcaddress=50:05:07:63:00:c9:0c:21 target=ESS800
Command executed successfully.
>> target_port_add fcaddress=50:05:07:63:00:cd:0c:21 target=ESS800
Command executed successfully.
>> target_connectivity_define local_port=1:FC_Port:5:4
fcaddress=50:05:07:63:00:c9:0c:21 target=ESS800
Command executed successfully.
>> target_connectivity_define local_port=1:FC_Port:7:4
fcaddress=50:05:07:63:00:cd:0c:21 target=ESS800
Command executed successfully.
>> target_connectivity_list
Target Name Remote Port FC Port IP Interface Active Up
ESS800 5005076300C90C21 1:FC_Port:5:4 yes yes
ESS800 5005076300CD0C21 1:FC_Port:7:4 yes yes

We now define the XIV as a host to the ESS 800. In Figure 7-27 we have defined the two
initiator ports on the XIV (with WWPNs that end in 53 and 73) as Linux (x86) hosts called
Nextra_Zap_5_4 and NextraZap_7_4.

Figure 7-27 Define the XIV to the ESS 800 as a host

234 IBM XIV Storage System: Copy Services and Migration

Finally, we can define the AIX host to the XIV as a host using the XIV GUI or XCLI. In
Example 7-21 we use the XCLI to define the host and then add two HBA ports to that host.

Example 7-21 Define Dolly to the XIV using XCLI

>> host_define host=dolly
Command executed successfully.
>> host_add_port fcaddress=10:00:00:00:c9:53:da:b3 host=dolly
Command executed successfully.
>> host_add_port fcaddress=10:00:00:00:c9:53:da:b2 host=dolly
Command executed successfully.

Once the zoning changes have been done and connectivity and correct definitions confirmed
between XIV to ESS and XIV to AIX host, we take an outage on the volume group and related
file systems that are going to be migrated. In Example 7-22 we unmount the file system, vary
off the volume group, and then export the volume group. Finally, we rmdev the hdisk devices.

Example 7-22 Removing the non-XIV file system

root@dolly:/# umount /mnt/redbk
root@dolly:/# varyoffvg ESS_VG1
root@dolly:/# exportvg ESS_VG1
root@dolly:/# rmdev -dl hdisk3
hdisk3 deleted
root@dolly:/# rmdev -dl hdisk4
hdisk4 deleted
root@dolly:/# rmdev -dl hdisk5
hdisk5 deleted

If the Dolly host no longer needs access to any LUNs on the ESS 800 we remove the SAN
zoning that connects Dolly to the ESS 800. In Example 7-18 on page 233 this was the zone
called ESS800_dolly_fcs0.

Chapter 7. Data migration 235

 We now allocate the ESS 800 LUNS to the XIV, as shown in Figure 7-28, where volume
serials 00FFCA33, 010FCA33, and 011FCA33 have been unmapped from the host called
Dolly and remapped to the XIV definitions called NextraZap_5_4 and NextraZap_7_4. We do
not allow the volumes to be presented to both the host and the XIV. Note that the LUN IDs in
the Host Port column are correct for use with XIV because they start with zero and are the
same for both NextraZap Initiator ports.

Figure 7-28 LUNs allocated to the XIV

We now create the DMs and run a test on each LUN. The XIV GUI or XCLI could be used. In
Example 7-23 the commands to create, test, and activate one of the three migrations is
shown. We must run each command for hdisk3 and hdisk4 also.

Example 7-23 Creating one migration

> dm_define target="ESS800" vol=”dolly_hdisk3” lun=0 source_updating=yes create_vol=yes pool=AIX
Command executed successfully.
> dm_test vol=”dolly_hdisk3”
Command executed successfully.
> dm_activate vol=”dolly_hdisk3”
Command executed successfully.

After we create and activate all three migrations, the Migration panel in the XIV GUI looks as
shown in Figure 7-29. Note that the remote LUN IDs are 0, 1, and 2, which must match the
LUN numbers seen in Figure 7-28.

Figure 7-29 Migration has started

236 IBM XIV Storage System: Copy Services and Migration

 Now that the migration has been started we can map the volumes to the AIX host definition
on the XIV, as shown in Figure 7-30, where the AIX host is called Dolly.

Figure 7-30 Map the XIV volumes to the host

Now we can bring the volume group back online. Because this AIX host was already using
SDDPCM, we can install the XIVPCM (the AIX host attachment kit) at any time prior to the
change. In Example 7-24 we confirm that SDDPCM is in use and that the XIV definition file
set is installed. We then run cfgmgr to detect the new disks. We then confirm that the disks
are visible using the lsdev -Cc disk command.

Example 7-24 Rediscovering the disks

root@dolly:~# lslpp -L | grep -i sdd
devices.sddpcm.53.rte 2.2.0.4 C F IBM SDD PCM for AIX V53
root@dolly:/# lslpp -L | grep 2810
disk.fcp.2810.rte 1.1.0.1 C F IBM 2810XIV ODM definitions
root@dolly:/# cfgmgr -l fcs0
root@dolly:/# cfgmgr -l fcs1
root@dolly:/# lsdev -Cc disk
hdisk1 Available 11-08-00-4,0 16 Bit LVD SCSI Disk Drive
hdisk2 Available 11-08-00-4,1 16 Bit LVD SCSI Disk Drive
hdisk3 Available 17-08-02 IBM 2810XIV Fibre Channel Disk
hdisk4 Available 17-08-02 IBM 2810XIV Fibre Channel Disk
hdisk5 Available 17-08-02 IBM 2810XIV Fibre Channel Disk

A final check before bringing the volume group back ensures that the Fibre Channel pathing
from the host to the XIV is set up correctly. We can use the AIX lspath command against
each hdisk, as shown in Example 7-25. Note that in this example the host can connect to
port 2 on each of the XIV modules 4, 5, 6, and 7 (which is confirmed by checking the last two
digits of the WWPN).

Example 7-25 Using the lspath command

root@dolly:~/# lspath -l hdisk5 -s available -F"connection:parent:path_status:status"
5001738000230161,3000000000000:fscsi1:Available:Enabled
5001738000230171,3000000000000:fscsi1:Available:Enabled
5001738000230141,3000000000000:fscsi0:Available:Enabled
5001738000230151,3000000000000:fscsi0:Available:Enabled

Chapter 7. Data migration 237

 We can also use a script provided by the XIV Host Attachment Kit for AIX, called xiv_devlist.
An example of the output is shown in Example 7-26.

Example 7-26 Using xiv_devlist

root@dolly:~# xiv_devlist
XIV devices
===========
Device Vol Name XIV Host Size Paths XIV ID Vol ID
------------------------------------------------------------------------------
hdisk3 dolly_hdisk3 dolly 10.0GB 4/4 MN00023 8940
hdisk4 dolly_hdisk4 dolly 10.0GB 4/4 MN00023 8941
hdisk5 dolly_hdisk5 dolly 10.0GB 4/4 MN00023 8942

Non-XIV devices
===============
Device Size Paths
-----------------------------------
hdisk1 N/A 1/1
hdisk2 N/A 1/1

We can also use the XIV GUI to confirm connectivity by going to the Hosts and Clusters 
Host Connectivity panel. An example is shown in Figure 7-31, where the connections match
those seen in Example 7-25 on page 237.

Figure 7-31 Host connectivity panel

Having confirmed that the disks have been detected and that the paths are good, we can now
bring the volume group back online. In Example 7-27 we import the VG, confirm that the
PVIDs match those seen in Example 7-17 on page 232, and then mount the file system.

Example 7-27 Bring the VG back online

root@dolly:/# /usr/sbin/importvg -y'ESS_VG1' hdisk3
ESS_VG1
root@dolly:/# lsvg -l ESS_VG1
ESS_VG1:
LV NAME TYPE LPs PPs PVs LV STATE MOUNT POINT
loglv00 jfs2log 1 1 1 closed/syncd N/A
fslv00 jfs2 20 20 3 closed/syncd /mnt/redbk
root@dolly:/# lspv
hdisk1 0000d3af10b4a189 rootvg active
hdisk3 0000d3afbec33645 ESS_VG1 active
hdisk4 0000d3afbec337b5 ESS_VG1 active
hdisk5 0000d3afbec33922 ESS_VG1 active
root@dolly:/# mount /mnt/redbk
root@dolly:/mnt/redbk# df -k
Filesystem 1024-blocks Free %Used Iused %Iused Mounted on
/dev/fslv00 20971520 11352580 46% 17 1% /mnt/redbk

238 IBM XIV Storage System: Copy Services and Migration

Once the sync is complete it is time to delete the migrations. Do not leave the migrations in
place any longer than they need to be. We can use multiple selection to perform the deletion,
as shown in Figure 7-32, taking care to delete and not deactivate the migration.

Figure 7-32 Deletion of the synchronized data migration

Now at the ESS 800 Web GUI we can un-map the three ESS 800 LUNs from the Nextra_Zap
host definitions. This frees up the LUN IDs to be reused for the next volume group migration.

After the migrations are deleted, a final suggested task is to re-size the volumes on the XIV to
the next 17 GB cutoff. In this example we migrate ESS LUNs that are 10 GB in size. However,
the XIV commits 17 GB of disk space because all space is allocated in 17 GB portions. For
this reason it is better to resize the volume on the XIV GUI from 10 GB to 17 GB so that all the
allocated space on the XIV is available to the operating system. This presumes that the
operating system can tolerate a LUN size growing, which in the case of AIX is true.

We must unmount any file systems and vary off the volume group before we start. Then we
go to the volumes section of the XIV GUI, right-click to select the 10 GB volume, and select
the Resize option. The current size appears. In Figure 7-33 the size is shown in 512 byte
blocks because the volume was automatically created by the XIV based on the size of the
source LUN on the ESS 800. If we multiply 19531264 by 512 bytes we get 10,000,007,168
bytes, which is 10 GB.

Figure 7-33 Starting volume size in blocks

We change the sizing methodology to GB and the size immediately changes to 17 GB, as
shown in Figure 7-34. If the volume was already larger than 17 GB, then it will change to the
next interval of 17 GB. For example, a 20 GB volume shows as 34 GB.

Figure 7-34 Size changed to GB

Chapter 7. Data migration 239

 We then get a warning message. The volume is increasing in size. Click OK to continue.

Now the volume is really 17 GB and no space is being wasted on the XIV. The new size is
shown in Figure 7-35.

Figure 7-35 Resized volumes

Vary on the VG again to update AIX that the volume size has changed. In Example 7-28 we
import the VG, which detects that the source disks have grown in size. We then run the chvg
-g command to grow the volume group, then confirm that the file system can still be used.

Example 7-28 Importing larger disks

root@dolly:~# /usr/sbin/importvg -y'ESS_VG1' hdisk3
0516-1434 varyonvg: Following physical volumes appear to be grown in size.
Run chvg command to activate the new space.
hdisk3 hdisk4 hdisk5
ESS_VG1
root@dolly:~# chvg -g ESS_VG1
root@dolly:~# mount /mnt/redbk
root@dolly:/mnt/redbk# df -k
Filesystem 1024-blocks Free %Used Iused %Iused Mounted on
/dev/fslv00 20971520 11352580 46% 17 1% /mnt/redbk

We can now resize the file system to take advantage of the extra space. In Example 7-29 the
original size of the file system in 512 byte blocks is shown.

Example 7-29 Displaying the current size of the file system

Change/Show Characteristics of an Enhanced Journaled File System

Type or select values in entry fields.

Press Enter AFTER making all desired changes.

[Entry Fields]
File system name /mnt/redbk
NEW mount point [/mnt/redbk]
SIZE of file system
Unit Size 512bytes
Number of units [41943040]

240 IBM XIV Storage System: Copy Services and Migration

We change the number of 512 byte units to 83886080 because this is 40 GB in size, as
shown in Example 7-30.

Example 7-30 Growing the file system

SIZE of file system
Unit Size 512bytes
+
Number of units [83886080]

The file system has now grown. In Example 7-31 we can see the file system has grown from
20 GB to 40 GB.

Example 7-31 Displaying the enlarged file system

root@dolly:~# df -k
/dev/fslv00 41943040 40605108 4% 7 1% /mnt/redbk

Chapter 7. Data migration 241

242 IBM XIV Storage System: Copy Services and Migration
 8

Chapter 8. SVC migration with XIV

This chapter discusses data migration considerations for the XIV Storage System when used
in combination with the IBM SAN Volume Controller (SVC). It presumes that you have an
existing SVC and that you are replacing back-end disk controllers with a new XIV or simply
adding an XIV as a new managed disk controller.

The combination of SVC and XIV allows a client to benefit from the high-performance grid
architecture of the XIV while retaining the business benefits delivered by the SVC (such as
higher performance via disk aggregation, multivendor and multi-device copy services, and
data migration functions).

The order of the sections in this chapter address each of the requirements of an
implementation plan in the order in which they arise. This chapter does not, however, discuss
physical implementation requirements (such as power requirements), as they are already
addressed in the book IBM XIV Storage System: Architecture, Implementation, and Usage,
SG24-76599, found here:
http://www.redbooks.ibm.com/Redbooks.nsf/RedbookAbstracts/sg247659.html?Open

© Copyright IBM Corp. 2010. All rights reserved. 243

8.1 Steps to take when using SVC migration with XIV
There are six considerations when placing a new XIV behind an SVC:
򐂰 “XIV and SVC interoperability” on page 244
򐂰 “Zoning setup” on page 245
򐂰 “Volume size considerations for XIV with SVC” on page 248
򐂰 “Using an XIV for SVC quorum disks” on page 253
򐂰 “Configuring an XIV for attachment to SVC” on page 255
򐂰 “Data movement strategy overview” on page 259

8.2 XIV and SVC interoperability

Because SVC-attached hosts do not communicate directly with the XIV, there are only two
interoperability considerations:
򐂰 8.2.1, “Firmware versions” on page 244
򐂰 8.2.2, “Copy functions” on page 245

8.2.1 Firmware versions

The SVC and XIV both have minimum firmware requirements. Whereas the versions given
here are current at the time of writing, they may have since changed. Confirm them by visiting
the IBM Systems Storage Interoperation Center (SSIC) at:
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp

SVC firmware
The first SVC firmware version that supported XIV was 4.3.0.1. However, the SVC cluster
should be on at least SVC firmware Version 4.3.1.4 or more preferably the most recent level
available from IBM. You can display the SVC firmware version by viewing the cluster
properties in the SVC GUI or by using the svcinfo lscluster command specifying the name
of the cluster. The SVC in Example 8-1 is on SVC code level 4.3.1.5.

Example 8-1 Displaying the SVC cluster code level using SVC CLI
IBM_2145:SVCSTGDEMO:admin> svcinfo lscluster SVCSTGDEMO
code_level 4.3.1.5 (build 9.16.0903130000)

XIV firmware
The XIV should be on at least XIV firmware Version 10.0.0.a. The XIV firmware version is
shown on the All Systems front page of the XIV GUI. The XIV in Figure 8-1 is on version
10.0.1.b (circled on the upper right in red).

Figure 8-1 Figure 8-1Version 10.0.1.b

244 IBM XIV Storage System: Copy Services and Migration

 The XIV firmware version can also be displayed by using an XCLI command as shown in
Example 8-2, where the example machine is on XIV firmware Version 10.0.1.b.

Example 8-2 Displaying the XIV firmware version

xcli -m 10.0.0.1 -u admin -p adminadmin version_get
Version
10.0.1.b

Note that an upgrade from XIV 10.0.x.x code levels to 10.1.x.x code levels is not concurrent
(meaning that the XIV is unavailable for I/O during the upgrade).

8.2.2 Copy functions

The XIV has many advanced copy and remote mirror capabilities, but for XIV volumes being
used as SVC MDisks (including image mode VDisk/MDisks), none of these functions can be
used. If copy and mirror functions are needed, they should be performed using the equivalent
functional capabilities in the SVC (such as SVC FlashCopy and SVC Metro and Global
Mirror). This is because XIV copy functions are not aware of un-destaged write cache data
resident in the SVC cache. Whereas it is possible to disable SVC write-cache (when creating
VDisks), this method is not supported by IBM for VDisks resident on XIV.

8.2.3 TPC with XIV and SVC

XIV code levels 10.1.0.a and later support the use of Tivoli Storage Productivity Center (TPC)
via an embedded SMI-S agent in the XIV. Subsequently, if you want to use TPC in conjunction
with XIV then you your XIV must be code level 10.1.0.a (or later). TPC itself must be Version
4.1 (or later). Refer to the “Recommended Software Levels for SAN Volume Controller”
documentation for your SVC code level to identify the latest recommended TPC for Disk
software version via the following Web site:
http://www.ibm.com/storage/support/2145

8.3 Zoning setup

One of the first tasks when implementing XIV is to add the XIV to the SAN fabric so that the
SVC cluster can communicate with the XIV over the Fibre Channel. The XIV can have up to
24 Fibre Channel host ports. Each XIV reports a single World Wide Node Name (WWNN)
that is the same for every XIV Fibre Channel host port. Each port also has a unique and
persistent World Wide Port Name (WWPN), which means that we can potentially zone 24
unique WWPNS from an XIV to an SVC cluster. However, the current SVC firmware has a
requirement that one SVC cluster cannot detect more than 16 WWPNs per WWNN, so at this
time there is no value in zoning more than 16 ports to the SVC. Because the XIV can have up
to six interface modules with four ports per module, it is better to use just two ports on each
module (allowing up to 12 ports total).

Chapter 8. SVC migration with XIV 245

 When a partially populated XIV has a hardware upgrade to add usable capacity, more data
modules are added. At particular points in the upgrade path, the XIV will get more usable
Fibre Channel ports. In each case, we use half the available ports to communicate with an
SVC cluster (we do this to facilitate growth as modules are added). Depending on the total
usable capacity of the XIV, not all interface modules have active Fibre Channel ports.
Table 8-1 shows which modules will have active ports as capacity grows. You can also see
how many XIV ports we zone to the SVC as capacity grows.

Table 8-1 XIV host ports as capacity grows

XIV modules Total usable Total XIV XIV host Active Inactive
capacity (TB) host ports ports to zone interface interface
to an SVC modules modules
cluster

6 27.26 8 4 4:5 6

9 43.09 16 8 4:5:7:8 6:9

10 50.29 16 8 4:5:7:8 6:9

11 54.65 20 10 4:5:7:8:9 6

12 61.74 20 10 4:5:7:8:9 6

13 66.16 24 12 4:5:6:7:8:9

14 73.24 24 12 4:5:6:7:8:9

15 79.11 24 12 4:5:6:7:8:9

Another way to view the activation state of the XIV interface modules is shown in Table 8-2.
As additional capacity is added to an XIV, additional XIV host ports become available. Where
a module is shown as inactive, this refers only to the host ports, not the data disks.

Table 8-2 XIV host ports as capacity grows

Module 27 TB 43 TB 50 TB 54 TB 61 TB 66 TB 73 TB 79 TB

Module 9 Not present Inactive Inactive Active Active Active Active Active
host ports

Module 8 Not present Active Active Active Active Active Active Active
host ports

Module 7 Not present Active Active Active Active Active Active Active
host ports

Module 6 Inactive Inactive Inactive Inactive Inactive Active Active Active

host ports

Module 5 Active Active Active Active Active Active Active Active

host ports

Module 4 Active Active Active Active Active Active Active Active

host ports

8.3.1 Capacity on demand

If the XIV has the Capacity on Demand (CoD) feature, then all Fibre Channel interface ports
are present and active (usable) at the time of install, regardless of how much usable capacity
has been purchased.

246 IBM XIV Storage System: Copy Services and Migration

8.3.2 Determining XIV WWPNs
XIV WWPNs are in the format 50:01:73:8x:xx:xx:RR:MP which break out as follows:
5 The WWPN format (1, 2, or 5, where XIV is always format 5)
0:01:73:8 The IEEE OID for IBM (formerly registered to XIV)
x:xx:xx Determined by IBM manufacturing and unique for every XIV rack
RR Rack ID (starts at 01)
M Module ID (ranges from 4 to 9)
P Port ID (0 to 3, although port numbers are 1–4)

91 93
 Module 9
90 92

91
81 93
83
 Module 8
90
80 92
82

71 73
 Module 7
70 72

61 63
 Module 6
60 62

91
51 93
53
 Module 5
90
50 92
52

41 43
 Module 4
40 42

Port 2 Port 4
Port 1 Port 3

Figure 8-2 XIV WWPN determination

In Figure 8-2, the MP value (module/port, which make up the last two digits of the WWPN) is
shown in each small box. The diagram represents the patch panel found at the rear of the XIV
rack.

To display the XIV WWPNs use the back view on the XIV GUI or the XCLI fc_port_list
command.

In the output example shown in Example 8-3 the four ports in module 4 are listed.

Example 8-3 Listing XIV Fibre Channel host ports

fc_port_list
Component ID Status Currently Functioning WWPN
1:FC_Port:4:4 OK yes 5001738000350143
1:FC_Port:4:3 OK yes 5001738000350142
1:FC_Port:4:2 OK yes 5001738000350141
1:FC_Port:4:1 OK yes 5001738000350140

Chapter 8. SVC migration with XIV 247

8.3.3 Hardware dependencies
There are two Fibre Channel HBAs in each XIV interface module. From a physical
perspective:
򐂰 Ports 1 and 2 are on the left-hand HBA (viewed from the rear).
򐂰 Ports 3 and 4 are on the right-hand HBA (viewed from the rear).

From a configuration perspective:

򐂰 Ports 1, 2, and 3 are in SCSI target mode by default.
򐂰 Port 4 is set to SCSI initiator mode by default (for XIV replication and data migration).

For availability and performance use ports 1 and 3 for SVC and general host traffic. If you
have two fabrics, place port 1 in the first fabric and port 3 in the second fabric.

8.3.4 Sharing an XIV with another SVC cluster or non-SVC hosts

It is possible to share XIV host ports between an SVC cluster and non-SVC hosts, or between
two different SVC clusters. Simply zone the XIV host ports 1 and 3 on each XIV module to
both SVC and non-SVC hosts as required.

You can instead choose to use ports 2 and 4, although in principle these are reserved for data
migration and remote mirroring. For that reason port 4 on each module is by default in initiator
mode. If you want to change the mode of port 4 to target mode, you can do so easily from the
XIV GUI or XCLI. However, you may also need an RPQ from IBM. Contact your IBM XIV
representative to discuss this.

8.3.5 Zoning rules

The XIV-to-SVC zone should simply contain all the XIV ports in that fabric and all the SVC
ports in that fabric. In other words one big zone. This recommendation is relatively unique to
SVC. If you zone individual hosts directly to the XIV (instead of via SVC), then you should
always use single-initiator zones where each switch zone contains just one host (initiator)
HBA WWPN and up to six XIV host port WWPNs.

For SVC, ensure that the following rules are followed:

򐂰 With current SVC firmware levels, no more than 16 WWPNs from a single WWNN should
be zoned to an SVC cluster. Because the XIV has only one WWNN, this means that no
more than 16 XIV host ports should be zoned to a specific SVC cluster. If you use the
recommendations in Table 8-1 on page 246 this restriction should not be an issue.
򐂰 All nodes in an SVC cluster must be able to see the same set of XIV host ports. Operation
in a mode where two nodes see a different set of host ports on the same XIV will result in
the controller showing on the SVC as degraded and the system error log will request a
repair action. If the one big zone per fabric rule is followed, then this requirement is met.

8.4 Volume size considerations for XIV with SVC

There are several considerations when migrating data onto XIV using SVC. Volume sizes is
clearly an important one.

248 IBM XIV Storage System: Copy Services and Migration

8.4.1 SCSI queue depth considerations
The SVC uses one XIV host port as a preferred port for each MDisk (assigning them in a
round-robin fashion). A best practice is to therefore configure sufficient volumes on the XIV to
ensure that:
򐂰 Each XIV host port will receive closely matching I/O levels.
򐂰 The SVC will utilize the deep queue depth of each XIV host port.

Ideally, the number of MDisks presented by the XIV to the SVC should be a multiple of the
number of XIV host ports, from one to four. However, there is good math to support this.

The XIV can handle a queue depth of 1400 per Fibre Channel host port and a queue depth of
256 per mapped volume per host port:target port:volume tuple. However, the SVC sets the
following internal limits:
򐂰 The maximum queue depth per MDisk is 60.
򐂰 The maximum queue depth per target host port on an XIV is 1000.

Based on this knowledge, we can determine an ideal number of XIV volumes to map to the
SVC for use as MDisks by using this algorithm:
Q = ((P x C) / N) / M

This breaks out as follows:

Q The calculated queue depth for each MDisk
P The number of XIV host ports (unique WWPNs) visible to the SVC
cluster (should be 4, 8, 10, or 12 depending on the number of modules
in the XIV)
N The number of nodes in the SVC cluster (2, 4, 6, or 8)
M The number of volumes from the XIV to the SVC cluster (detected as
MDisks)
C 1000 (which is the maximum SCSI queue depth that an SVC will use
for each XIV host port)

If a 2-node SVC cluster is being used with 12 ports on IBM XIV System and 48 MDisks, this
yields a queue depth that as follows:
Q = ((12 ports*1000)/2 nodes)/48 MDisks = 125

Because 125 is greater than 60, the SVC uses a queue depth of 60 per MDisk. If a 4-node
SVC cluster is being used with 12 host ports on the IBM XIV System and 48 MDisks, this
yields a queue depth that as follows:
Q = ((12 ports*1000)/4 nodes)/48 MDisks = 62

Because 62 is greater than 60, the SVC uses a queue depth of 60 per MDisk.

Chapter 8. SVC migration with XIV 249

 This leads to the following recommended volume sizes and quantities for 2-node and 4-node
SVC clusters.

Table 8-3 XIV volume size and quantity recommendation

Modules Total usable XIV Volume Volume Ratio of Approximate
capacity host size quantity volumes left
(TB) ports (GB) to XIV host over space
ports (TB)

6 27.26 4 1632 16 4 1.14

9 43.09 8 1632 26 3.3 0.65

10 50.29 8 1632 30 3.7 1.32

11 54.65 10 1632 33 3.3 0.79

12 61.74 10 1632 37 3.7 1.35

13 66.16 12 1632 40 3.3 0.87

14 73.24 12 1632 44 3.7 1.42

15 79.11 12 1632 48 4.0 0.76

If you have a 6-node or 8-node cluster, the formula suggests that you must use much larger
XIV volumes. However, currently available SVC firmware does not support an MDisk larger
than 2 TB, so it is simpler to continue to use the 1632 GB volume size. When using 1632 GB
volumes, there is leftover space. That space could be used for testing or for non-SVC
direct-attach hosts. If you map the remaining space to the SVC as an odd sized volume then
VDisk striping is not balanced, meaning that I/O is not be evenly striped across all XIV host
ports.

Tip: If you only provision part of the usable space of the XIV to be allocated to the SVC,
then the calculations above no longer work. You should instead size your MDisks to ensure
that at least two (and up to four) MDisks are created for each host port on the XIV.

8.4.2 XIV volume sizes

All volume sizes shown on the XIV GUI use decimal counting (109), meaning that 1 GB =
1,000,000,000 bytes. However, a GB using binary counting (using 230 bytes, more accurately
referred to as a GiB) counts 1 GiB as 1,073,741,824 bytes (ideally called a GiB to differentiate
it from a GB where size is calculated using decimal counting).
򐂰 By default the SVC uses MiB and GiB (binary counting method) when displaying MDisk
and VDisk sizes. However, the SVC still uses the term MB in the SVC GUI and MB or GB
in the SVC CLI output when displaying volume and disk sizes (the SVC CLI displays
capacity in whatever unit it decides is the most human readable).
򐂰 By default the XIV uses GB (decimal counting method) in the XIV GUI and CLI output
when displaying volume sizes.

250 IBM XIV Storage System: Copy Services and Migration

It also must be clearly understood that a volume created on an XIV is created in 17 GB
increments, which are not exactly 17 GB. In fact, the size of an XIV 17 GB volume can be
described in four ways:
GB 17 GB (decimal), as shown in the XIV GUI, but actually rounded down
to the nearest GB (see the number of bytes below).
GiB 16 GiB (binary counting where 1 GiB = 230 bytes). This is exactly 16
GiB.
Bytes 17,179,869,184 bytes.
Blocks 33,554,432 blocks (each block being 512 bytes).

Thus, XIV is using binary sizing when creating volumes, but displaying it in decimal and then
rounding it down.

The recommended volume size for XIV volumes presented to the SVC is 1632 GB (as viewed
on the XIV GUI). There is nothing special about this volume size, it simply divides nicely to
create on average four XIV volumes per XIV host port (for queue depth purposes).

The size of a 1632 GB volume (as viewed on the XIV GUI) can be stated in four ways:
GB 1632 GB (decimal), as shown in the XIV GUI, but rounded down to the
nearest GB (see the number of bytes below).
GiB 1520 GiB (binary counting where 1 GiB = 230 bytes). This is exactly
1520 GiB.
Bytes 1,632,087,572,480 bytes.
Blocks 3,187,671,040 blocks (each block being 512 bytes).

Note that the SVC reports each MDisk presented by XIV as 1520 GiB. Figure 8-3 shows what
the XIV reports.

Figure 8-3 An XIV volume sized for use with SVC

If you right-click the volume in the XIV GUI and display properties, you will be able to see that
this volume is 3,187,671,040 blocks. If you multiply 3,187,671,040 by 512 (because there are
512 bytes in a SCSI block) you will get 1,632,087,572,480 bytes. If you divide that by
1,073,741,824 (the number of bytes in a binary GiB), then you will get 1520 GiB, which is
exactly what the SVC reports for the same volume (MDisk), as shown in Example 8-4.

Example 8-4 An XIV volume mapped to the SVC

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -bytes

id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 1632087572480 0000000000000007 XIV

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk
id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 1520.0GB 0000000000000007 XIV

Chapter 8. SVC migration with XIV 251

8.4.3 Creating XIV volumes that are exactly the same size as SVC VDisks
To create an XIV volume that is exactly the same size as an existing SVC VDisk you can use
the process documented in 8.10.1, “Create image mode destination volumes on the XIV” on
page 267. This is only for a transition to or from image mode.

8.4.4 SVC 2TB volume limit

The XIV can create volumes of any size up to the entire capacity of the XIV. However, in the
current release of SVC firmware (including release 5.1), the largest MDisk that an SVC can
detect is 2 TiB in size (which is 2048 GiB). To create this volume on the XIV, create a volume
sized 2199 GB (because 2199 GB = 2048 GiB). However, the recommended volume size for
SVC is 1632 GB (1520 GiB).

In Figure 8-4 there are three volumes that will be mapped to the SVC. The first volume is
2199 GB (2 TiB), but the other two are larger than that.

Figure 8-4 XIV volumes larger than 2 TiB

When presented to the SVC, the SVC reports all three as being 2 TiB (2048 GiB), as shown
in Example 8-5.

Example 8-5 2 TiB volume size limit on SVC

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk
id name status mode capacity
9 mdisk9 online unmanaged 2048.0GB
10 mdisk10 online unmanaged 2048.0GB
11 mdisk11 online unmanaged 2048.0GB

Because there was no benefit in using larger volume sizes do not follow this example. Always
ensure that volumes presented by the XIV to the SVC are 2199 GB or smaller (when viewed
on the XIV GUI or XCLI).

8.4.5 MDisk group creation

All volumes presented by the XIV to the SVC are represented on the SVC as MDisks and
should be placed into one managed disk group. All VDisks created in this managed disk
group should be created as striped and striped across all MDisks in the group. This ensures
that we stripe SVC host I/O evenly across all the XIV host ports.

8.4.6 SVC MDisk group extent sizes

SVC MDisk groups have a fixed extent size. This extent size affects the maximum size of an
SVC cluster. When migrating SVC data from other disk technology to XIV, change the extent
size at the same time. This not only allows for larger sized SVC clusters, but also ensures that

252 IBM XIV Storage System: Copy Services and Migration

 the data from each extent best utilizes the striping mechanism in the XIV. Because the XIV
divides each volume into 1 MB partitions, the MDisk group extent size in MB should exceed
the maximum number of disks that are likely to exist in a single XIV footprint. For many
customers this means that an extent size of 256 MB is acceptable (because 256 MB covers
256 disks where a single XIV rack has only 180 disks). However, strongly consider using an
extent size of 1024 MB because this covers the possibility of a 5-rack XIV with 900 disks.

In terms of the available SVC extent sizes and the effect on maximum SVC cluster size, see
Table 8-4.

Table 8-4 SVC extent size and cluster size

MDisk group Maximum SVC
extent size cluster size

16 MB 64 TB

32 MB 128 TB

64 MB 256 TB

128 MB 512 TB

256 MB 1024 TB

512 MB 2048 TB

1024 MB 4096 TB

2048 MB 8192 TB

8.5 Using an XIV for SVC quorum disks

The SVC cluster uses three MDisks as quorum disks. It uses a small area on each of these
MDisks to store important SVC cluster management information. If you are replacing non-XIV
disk storage with XIV, ensure that you relocate the quorum disks before removing the MDisks.
Review the tip at the following Web site:
http://www.ibm.com/support/docview.wss?uid=ssg1S1003311

To determine whether removing a managed disk controller requires quorum disk relocation,
run a script to find the MDisks that are being used as quorum disks, as shown in
Example 8-6. This script can be run safely without modification. Example 8-6 shows two
MDisks on the DS6800 and one MDisk on the DS4700.

Example 8-6 Identifying the quorum disks

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -nohdr | while read id name status mode
mdisk_grp_id mdisk_grp_name capacity ctrl_LUN controller_name mdisk_UID; do
svcinfo lsmdisk $id | while read key value; do if [ "$key" == "quorum_index" ];
then if [ "$value" != "" ]; then echo "Quorum index $value : mdisk $id ($name),
status=$status, controller=$controller_name"; fi; fi; done; done
Quorum index 0 : mdisk 0 (mdisk0), status=online, controller=DS6800_1
Quorum index 1 : mdisk 1 (mdisk1), status=online, controller=DS6800_1
Quorum index 2 : mdisk 2 (mdisk2), status=online, controller=DS4700

Chapter 8. SVC migration with XIV 253

 If your SVC uses firmware Version 5.1 or later, we simply use the svcinfo lsquorum
command, as shown in Example 8-7.

Example 8-7 Using the svcinfo lsquorum command on SVC code level 5.1 and later
IBM_2145:mycluster:admin>svcinfo lsquorum
quorum_index status id name controller_id controller_name active
0 online 0 mdisk0 0 DS6800_1 yes
1 online 1 mdisk1 1 DS6800_1 no
2 online 2 mdisk2 2 DS4700 no

To move the quorum disk function, we specify three MDisks that will become quorum disks.
Depending on your MDisk group extent size, each selected MDisk must have between 272
MB and 1024 MB of free space. Execute the svctask setquorum commands before you start
migration. If all available MDisk space has been allocated to VDisks then you will not be able
to use that MDisk as a quorum disk. Table 8-5 shows the amount of space needed on each
MDisk.

Table 8-5 Quorum disk space requirements for each of the three quorum MDisks
Extent size (in MB) Number of extents Amount of space per MDisk
needed by quorum needed by quorum

16 17 272 MB

32 9 288 MB

64 5 320 MB

128 3 384 MB

256 2 512 MB

1024 1 1024 MB

2048 1 2048 MB

In Example 8-8 there are three free MDisks. They are 1520 GiB in size (1632 GB).

Example 8-8 New XIV MDisks detected by SVC

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged
id name status mode capacity
9 mdisk9 online unmanaged 1520.0 GB
10 mdisk10 online unmanaged 1520.0 GB
11 mdisk11 online unmanaged 1520.0 GB

In Example 8-9 the MDisk group is created using an extent size of 1024 MB.

Example 8-9 Creating an MDIsk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name XIV -mdisk 9:10:11 -ext 1024
MDisk Group, id [4], successfully created

254 IBM XIV Storage System: Copy Services and Migration

 In Example 8-10 the MDisk group has 4,896,262,717,440 free bytes (1520 GiB x 3).

Example 8-10 Listing the free capacity of the MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp -bytes 4

free_capacity 4896262717440

All three MDisks are set to be quorum disks, as shown in Example 8-11.

Example 8-11 Setting XIV MDisks as quorum disks

IBM_2145:SVCSTGDEMO:admin>svctask setquorum -quorum 0 mdisk9
IBM_2145:SVCSTGDEMO:admin>svctask setquorum -quorum 1 mdisk10
IBM_2145:SVCSTGDEMO:admin>svctask setquorum -quorum 2 mdisk11

The MDisk group has now lost free space, as shown in Example 8-12.

Example 8-12 Listing free space in the MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp -bytes 4
free_capacity 4893041491968

This means that free capacity fell by 3,221,225,472 bytes, which is 3 GiB or 1 GiB per quorum
MDisk.

Note: In this example all three quorum disks were placed on a single XIV. This may not be
an ideal configuration. The Web tip referred to at the start of this section has more details
about best practice, but in short you should try and use more than one managed disk
controller if possible.

8.6 Configuring an XIV for attachment to SVC

First we must configure the XIV.

8.6.1 XIV setup steps

The XIV GUI is remarkably easy to use, so we do not reproduce a series of XIV GUI images.
This section provides the setup steps using the XIV XCLI. They are reproduced mainly to
show the flow of commands rather than to indicate a preference for XCLI over the XIV GUI.
1. Define the SVC cluster to the XIV as in Example 8-13. An SVC cluster consists of several
nodes, with each SVC node being defined as a separate host.

Example 8-13 Define the SVC Cluster to the XIV

cluster_create cluster="SVC_Cluster1"

Chapter 8. SVC migration with XIV 255

 2. Define the SVC nodes to the XIV (as members of the cluster), as shown in Example 8-14.
By defining each node as a separate host, we can get more information about individual
SVC nodes from the XIV performance statistics display.

Example 8-14 Define the SVC nodes to the XIV

host_define host="SVC_Node1" cluster="SVC_Cluster1"
host_define host="SVC_Node2" cluster="SVC_Cluster1"

3. Add the SVC host ports to the host definition of the first SVC node, as shown in
Example 8-15.

Example 8-15 Define the WWPNs of the first SVC node

host_add_port host="SVC_Node1" fcaddress="5005076801101234"
host_add_port host="SVC_Node1" fcaddress="5005076801201234"
host_add_port host="SVC_Node1" fcaddress="5005076801301234"
host_add_port host="SVC_Node1" fcaddress="5005076801401234"

4. Add the SVC host ports to the host definition of the second SVC node, as shown in
Example 8-16.

Example 8-16 Define the WWPNs of the second SVC node

host_add_port host="SVC_Node2" fcaddress="5005076801105678"
host_add_port host="SVC_Node2" fcaddress="5005076801205678"
host_add_port host="SVC_Node2" fcaddress="5005076801305678"
host_add_port host="SVC_Node2" fcaddress="5005076801405678"

5. Repeat steps 3 and 4 for each SVC I/O group. If you only have two nodes then you only
have one I/O group.
6. Create a storage pool. In Example 8-17 the command shown creates a pool with 8160 GB
of space and no snapshot space. The total size of the pool is determined by the volume
size that you choose to use. We do not need snapshot space because we cannot use XIV
snapshots with SVC MDisks.

Example 8-17 Create a pool on the XIV

pool_create pool="SVCDemo" size=8160 snapshot_size=0

Important: You must not use XIV thin provisioning pools with SVC. You must only use
regular pools. The command shown in Example 8-17 creates a regular pool (where the
soft size is the same as the hard size. This does not stop you from using thin
provisioned VDisks on the SVC.

7. Create the volumes in the pool, as shown in Example 8-18.

Example 8-18 Create XIV volumes for use by the SVC

vol_create size=1632 pool="SVCDemo" vol="SVCDemo_1"
vol_create size=1632 pool="SVCDemo" vol="SVCDemo_2"
vol_create size=1632 pool="SVCDemo" vol="SVCDemo_3"
vol_create size=1632 pool="SVCDemo" vol="SVCDemo_4"
vol_create size=1632 pool="SVCDemo" vol="SVCDemo_5"

256 IBM XIV Storage System: Copy Services and Migration

 8. Map the volumes to the SVC cluster using available LUN IDs (starting at zero), as shown
in Example 8-19.

Example 8-19 Map XIV volumes to the SVC cluster

map_vol cluster="SVC_Cluster1" vol="SVCDemo_1" lun="0"
map_vol cluster="SVC_Cluster1" vol="SVCDemo_2" lun="1"
map_vol cluster="SVC_Cluster1" vol="SVCDemo_3" lun="2"
map_vol cluster="SVC_Cluster1" vol="SVCDemo_4" lun="3"
map_vol cluster="SVC_Cluster1" vol="SVCDemo_5" lun="4"

Important: Only map volumes to the SVC cluster (not to individual nodes in the
cluster). This ensures that each SVC node sees the same LUNs with the same LUN
IDs. You must not allow a situation where two nodes in the same SVC cluster have
different LUN mappings.

Tip: The XIV GUI normally reserves LUN ID 0 for in-band management. The SVC
cannot take advantage of this, but is not affected either way. In Example 8-19 we
started the mapping with LUN ID 0, but if you used the GUI you will find that by default
you start with LUN ID 1.

9. If necessary, change the system name for XIV so that it matches the controller name used
on the SVC. In Example 8-20 we use the config_get command to determine the machine
type and serial number. Then we use the config_set command to set the system_name.
Whereas the XIV allows a long name with spaces, SVC can only use 15 characters with
no spaces.

Example 8-20 Setting the XIV system name

>> config_get
machine_serial_number=6000081
machine_type=2810
system_name=XIV 6000081
timezone=-39600
ups_control=yes
>> config_set name=system_name value="XIV_28106000081"

The XIV configuration tasks are now complete.

8.6.2 SVC setup steps

Assuming that the SVC is zoned to the XIV, we now switch to the SVC and run the following
SVC CLI commands:
1. Detect the XIV volumes:
svctask detectmdisk

Chapter 8. SVC migration with XIV 257

 2. List the newly detected MDisks, as shown in Example 8-21, where there are five free
MDisks. They are 1520 GiB in size (1632 GB).

Example 8-21 New XIV MDisks detected by SVC

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged
id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 1520.0GB 0000000000000000 controller2
10 mdisk10 online unmanaged 1520.0GB 0000000000000001 controller2
11 mdisk11 online unmanaged 1520.0GB 0000000000000002 controller2
12 mdisk12 online unmanaged 1520.0GB 0000000000000003 controller2
13 mdisk13 online unmanaged 1520.0GB 0000000000000004 controller2

3. Create an MDisk group, as shown in Example 8-22, where an MDisk group is created
using an extent size of 1024 MB.

Example 8-22 Create the MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name XIV -mdisk 9:10:11:12:13 -ext 1024
MDisk Group, id [4], successfully created

Important: Adding a new managed disk group to the SVC may result in the SVC
reporting that you have exceeded the virtualization license limit. Whereas this does not
affect operation of the SVC, you continue to receive this error message until the
situation is corrected (by either removing the MDisk Group or increasing the
virtualization license). If the non-XIV disk is not being replaced by the XIV then ensure
that an additional license has been purchased. Then increase the virtualization limit
using the svctask chlicense -virtualization xx command (where xx specifies the
new limit in TB).

4. Relocate quorum disks if required as documented in “Using an XIV for SVC quorum disks”
on page 253.
5. Rename the controller from its default name. A managed disk controller is given a name
by the SVC such as controller0 or controller1 (depending on how many controllers have
already been detected). Because the XIV can have a system name defined for it, aim to
closely match the two names. Note, however, that the controller name used by SVC
cannot have spaces and cannot be more than 15 characters long. In Example 8-23
controller number 2 is renamed to match the system name used by the XIV itself (which
was set in Example 8-20 on page 257).

Example 8-23 Rename the XIV controller definition at the SVC

IBM_2145:SVCSTGDEMO:admin>svcinfo lscontroller
id controller_name ctrl_s/n vendor_id product_id_low product_id_high
0 controller0 13008300000 IBM 1750500
1 controller1 NETAPP LUN
2 controller2 IBM 2810XIV- LUN-0

IBM_2145:SVCSTGDEMO:admin>svctask chcontroller -name "XIV_28106000081" 2

258 IBM XIV Storage System: Copy Services and Migration

 6. Rename all the SVC MDisks from their default names (such as mdisk9 and mdisk10) to
match the volume names used on the XIV. An example of this is shown in Example 8-24
(limited to just two MDisks). You can match the ctrl_LUN_# value to the LUN ID assigned
when mapping the volume to the SVC (for reference also see Example 8-18 on page 256).
Be aware that the ctrl_LUN field displays LUN IDs using hexadecimal numbering, whereas
the XIV displays them using decimal numbering. This means that XIV LUN ID 10 displays
as ctrl_LUN ID A.

Example 8-24 Rename the MDisks

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mdisk_grp_id=4
id name status mode mdisk_grp_id capacity ctrl_LUN_# controller_name
9 mdisk9 online managed 4 1520.0GB 0000000000000000 XIV_28106000081
10 mdisk10 online managed 4 1520.0GB 0000000000000001 XIV_28106000081

IBM_2145:SVCSTGDEMO:admin>svctask chmdisk -name SVCDemo_1 mdisk9

IBM_2145:SVCSTGDEMO:admin>svctask chmdisk -name SVCDemo_2 mdisk10
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mdisk_grp_id=4
id name status mode mdisk_grp_id capacity ctrl_LUN_# controller_name
9 SVCDemo_1 online managed 4 1520.0GB 0000000000000000 XIV_28106000081
10 SVCDemo_2 online managed 4 1520.0GB 0000000000000001 XIV_28106000081

Now we must follow one of the migration strategies, as described in the 8.7, “Data movement
strategy overview” on page 259.

8.7 Data movement strategy overview

There are three possible data movement strategies that we detail in this section and in
subsequent sections.

8.7.1 Using SVC migration to move data

You can use standard SVC migration to move data from MDisks presented by a non-XIV disk
controller to MDisks resident on the XIV. This process does not require a host outage, but
does not allow the MDisk group extent size to be changed. At a high level, the process is as
follows:
1. We start with existing VDisks in an existing MDisk Group. We must confirm the extent size
of that MDisk group. We call this the source MDisk group.
2. We create 1632 GB sized volumes on the XIV and map these to the SVC.
3. We detect these new MDisks and use them to create an MDisk group. We call this the
target Mdisk group. The target MDisk group must use the same extent size as the source
MDisk group.
4. We migrate each VDisk from the source MDisk group to the target MDisk group.
5. When all the VDisks are migrated we can choose to delete the source MDisks and the
source MDisk group (in preparation for removing the non-XIV storage).

We discuss this method in greater depth in 8.8, “Using SVC migration to move data to XIV” on
page 261.

Chapter 8. SVC migration with XIV 259

8.7.2 Using VDisk mirroring to move the data
We can use the VDisk copy (mirror) function introduced in SVC firmware Version 4.3 to create
two copies of the data, one in the source MDisk group and one in the target MDisk group. We
then remove the VDisk copy in the source MDisk group and retain the VDisk copy present in
the target MDisk group. This process does not require a host outage and allows us to move to
a larger MDisk group extent size. However, it also uses additional SVC cluster memory and
CPU while the multiple copies are managed by the SVC.

At a high level the process is as follows:

1. We start with existing VDisks in an existing MDisk Group. The extent size of that MDisk
group is not relevant. We call this MDisk group the source MDisk group.
2. We create 1632 GB sized volumes on the XIV and map these to the SVC.
3. We detect these XIV MDisks and create an MDisk group using an extent size of 1024 MB.
We call this MDisk group the target Mdisk group.
4. For each VDisk in the source MDisk group, we create a VDisk copy in the target MDisk
group.
5. When the two copies are in sync we remove the VDisk copy that exists in the source
MDisk group (which is normally copy 0 since it existed first, as opposed to copy 1, which
we created for migration purposes).
6. When all the VDisks have been successfully copied from the source MDisk group to the
target MDisk group, we can choose to delete the source MDisks and the source MDisk
group (in preparation for removing the non-XIV storage) or split the VDisk copies and
retain copy 0 for as long as necessary.

We discuss this method in greater depth in 8.9, “Using VDisk mirroring to move the data” on
page 263.

8.7.3 Using SVC migration with image mode

This migration method is used when:
򐂰 The extent size must be changed but VDisk mirroring cannot be used, perhaps because
the SVC nodes are already constrained for CPU and memory. Because the SVC must be
on 4.3 code to support XIV (SVC code level 4.3 being the level that brought in VDisk
mirroring), having downlevel SVC firmware is not a valid reason.
򐂰 We want to move the VDisks from one SVC cluster to a different one.
򐂰 We want to move the data away from the SVC without using XIV migration.

In these cases we can migrate the VDisks to image mode and take an outage to do the
relocation and extent re-size. There will be a host outage, although it can kept very short
(potentially in the order of seconds or minutes).

At a high level the process is as follows:

1. We start with existing VDisks in an existing MDisk group. Possibly the extent size of this
MDisk group is small (say 16 MB). We call this the source MDisk group.
2. We create XIV volumes that are the same size (or larger) than the existing VDisks. This
may need extra steps, as the XIV volumes must be created using 512 byte blocks. We
map these specially sized volumes to the SVC.

260 IBM XIV Storage System: Copy Services and Migration

 3. We migrate each VDisk to image mode using these new volumes (presented as
unmanaged MDisks). The new volumes move into the source MDisk group as image
mode MDisks and the VDisks become image mode VDisks.
4. We can now remove all the Image Mode MDisks from the source MDisk group. This is the
disruptive part of this process. They are now unmanaged MDisks, but the data on these
volumes is intact. We could at this point map these volumes to a different SVC cluster or
we could remove them from the SVC altogether (in which case the process is complete).
5. We create a new managed disk group that contains only the image mode VDisks, but
using the recommended extent size (1024 MB) and present the VDisks back to the hosts.
We call this the transition MDisk group. The host downtime is now over.
6. We create another new managed disk group using free space on the XIV, using the same
large extent size (1024 MB). We call this the target MDisk group.
7. We migrate the image mode VDisks to managed mode VDisks, moving the data from the
transition MDisk group created in step 5 to the target MDisk Group created in step 6. The
MDisks themselves are already on the XIV.
8. When the process is complete, we can delete the source MDisks and the source MDisk
group (which represent space on the non-XIV storage controller) and the transitional XIV
volumes (which represent space on the XIV).
9. We can then use the transitional volume space on the XIV to create more 1632 GB
volumes to present to the SVC. These can be added into the existing MDisk group or used
to create a new one.

This method is detailed in greater depth in 8.10, “Using SVC migration with image mode” on
page 267.

8.8 Using SVC migration to move data to XIV

This process migrates data from a source MDisk Group to a target Mdisk group using the
same Mdisk group extent size. These is no interruption to host I/O.

8.8.1 Determine the required extent size and VDisk candidates

We must determine the extent size of the source MDisk group. In Example 8-25 MDisk Group
ID 1 is the source group and has an extent size of 256.

Example 8-25 Listing MDisk groups

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp
id name status mdisk_count vdisk_count capacity extent_size free_capacity
0 MDG_DS6800 online 2 0 399.5GB 16 399.5GB
1 Source_GRP online 1 1 50.0GB 256 45.0GB

We then must identify the VDisks that we are migrating. We can filter by MDisk Group ID, as
shown in Example 8-26, where there is only one VDisk that must be migrated.

Example 8-26 Listing VDisks filtered by MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk -filtervalue mdisk_grp_id=1
id name status mdisk_grp_id mdisk_grp_name capacity type
5 migrateme online 1 Source_GRP 5.00GB striped

Chapter 8. SVC migration with XIV 261

8.8.2 Create the MDisk group
We must create volumes on the XIV and map them to the SVC cluster. Presuming that we
have done this, we then detect them on the SVC, as shown in Example 8-27.

Example 8-27 Detecting new MDisks

IBM_2145:SVCSTGDEMO:admin>svctask detectmdisk
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskcandidate
id
9
10
11
12
13
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged
id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 1520.0GB 0000000000000007 XIV
10 mdisk10 online unmanaged 1520.0GB 0000000000000008 XIV
11 mdisk11 online unmanaged 1520.0GB 0000000000000009 XIV
12 mdisk12 online unmanaged 1520.0GB 000000000000000A XIV
13 mdisk13 online unmanaged 1520.0GB 000000000000000B XIV

We then create an Mdisk group called XIV_Target using the new XIV MDisks, with the same
extent size as the source group. In Example 8-28 it is 256.

Example 8-28 Creating an MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name XIV_Target -mdisk 9:10:11:12:13 -ext 256
MDisk Group, id [2], successfully created

We confirm the new MDisk group is present. In Example 8-29 we are filtering by using the
new ID of 2.

Example 8-29 Checking the newly created MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp -filtervalue id=2
id name status mdisk_count vdisk_count capacity extent_size free_capacity
2 XIV_Target online 5 1 7600.0GB 256 7600.0GB

8.8.3 Migration
Now we are ready to migrate the VDisks. In Example 8-30 we migrate VDisk 5 into MDisk
group 2 and then confirm that the migration is running.

Example 8-30 Migrating a VDisk

IBM_2145:SVCSTGDEMO:admin>svctask migratevdisk -mdiskgrp 2 -vdisk 5
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmigrate
migrate_type MDisk_Group_Migration
progress 0
migrate_source_vdisk_index 5
migrate_target_mdisk_grp 2
max_thread_count 4
migrate_source_vdisk_copy_id 0

262 IBM XIV Storage System: Copy Services and Migration

 When the lsmigrate command returns no output, the migration is complete. Once all Vdisks
have been migrated out of the Mdisk group, we can remove the source MDisks and then
remove the source Mdisk group, as shown in Example 8-31.

Example 8-31 Removing non-XIV MDisks and MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mdisk_grp_id=1
id name status mode capacity ctrl_LUN_# controller_name
8 mdisk8 online managed 50.0GB 00000000000000070 DS6800_1
IBM_2145:SVCSTGDEMO:admin>svctask rmmdisk -mdisk 8 1
IBM_2145:SVCSTGDEMO:admin>svctask rmmdiskgrp 1

Important: Scripts that use VDisk names or IDs will not be affected by the use of VDisk
migration, as the VDisk names and IDs do not change.

8.9 Using VDisk mirroring to move the data

This process mirrors data from a source MDisk group to a target Mdisk group using a different
extent size and with no interruption to the host.

8.9.1 Determine the required extent size and VDisk candidates

We must determine the source MDisk group. In Example 8-32 MDisk group ID 1 is the
source.

Example 8-32 Listing the MDisk groups

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp
id name status mdisk_count vdisk_count capacity extent_size free_capacity
0 MDG_DS68 online 2 0 399.5GB 16 399.5GB
1 Source online 1 1 50.0GB 256 45.0GB

We then must identify the VDisks that we are migrating. In Example 8-33 we filter by ID.

Example 8-33 Filter VDisks by MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk -filtervalue mdisk_grp_id=1
id name status mdisk_grp_id mdisk_grp_name capacity type
5 migrateme online 1 Source_GRP 5.00GB striped

Chapter 8. SVC migration with XIV 263

8.9.2 Create the MDisk group
We must create volumes on the XIV and map them to the SVC cluster. Presuming that we
have done this, we then detect them on the SVC, as shown in Example 8-34.

Example 8-34 Detecting new MDisks and creating an MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask detectmdisk
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskcandidate
id
9
10
11
12
13
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged
id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 1520.0GB 0000000000000007 XIV
10 mdisk10 online unmanaged 1520.0GB 0000000000000008 XIV
11 mdisk11 online unmanaged 1520.0GB 0000000000000009 XIV
12 mdisk12 online unmanaged 1520.0GB 000000000000000A XIV
13 mdisk13 online unmanaged 1520.0GB 000000000000000B XIV

We then create an MDisk group called XIV_Target using the new XIV MDisks (with the same
extent size as the source group, in this example 256), as shown in Example 8-35.

Example 8-35 Creating an MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name XIV_Target -mdisk 9:10:11:12:13 -ext 256
MDisk Group, id [2], successfully created

We confirm that the new MDisk group is present. In Example 8-36 we are filtering by using
the new ID of 2.

Example 8-36 Checking the newly created MDisk group

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdiskgrp -filtervalue id=2
id name status mdisk_count vdisk_count capacity extent_size free_capacity
2 XIV_Target online 5 1 7600.0GB 256 7600.0GB

8.9.3 Set up the IO group for mirroring

The IO group requires reserved memory for mirroring. First check to see whether this has
been done. In Example 8-37 it has not been setup yet on I/O group 0.

Example 8-37 Checking the I/O group for mirroring

IBM_2145:SVCSTGDEMO:admin>svcinfo lsiogrp 0
id 0
name io_grp0
node_count 2
vdisk_count 6
host_count 2
flash_copy_total_memory 20.0MB
flash_copy_free_memory 20.0MB
remote_copy_total_memory 20.0MB

264 IBM XIV Storage System: Copy Services and Migration

 remote_copy_free_memory 20.0MB
mirroring_total_memory 0.0MB
mirroring_free_memory 0.0MB

We must assign space for mirroring. Assigning 20 MB will support 40 TB of mirrors. In

Example 8-38 we do this on I/O group 0 and confirm that it is done.

Example 8-38 Setting up the I/O group for VDisk mirroring

IBM_2145:SVCSTGDEMO:admin>svctask chiogrp -size 20 -feature mirror 0
IBM_2145:SVCSTGDEMO:admin>svcinfo lsiogrp 0
id 0
name io_grp0
node_count 2
vdisk_count 6
host_count 2
flash_copy_total_memory 20.0MB
flash_copy_free_memory 20.0MB
remote_copy_total_memory 20.0MB
remote_copy_free_memory 20.0MB
mirroring_total_memory 20.0MB
mirroring_free_memory 20.0MB

8.9.4 Create the mirror

Now we create the mirror. In Example 8-39 we create a mirror copy of VDisk 5 into MDisk
group 2. Remember Mdisk group 2 has a different extent size than Mdisk group 1.

Example 8-39 Creating the VDisk mirror

IBM_2145:SVCSTGDEMO:admin>svctask addvdiskcopy -mdiskgrp 2 5
Vdisk [5] copy [1] successfully created

In Example 8-40 we can see the two copies (and also that they are not yet in sync).

Example 8-40 Monitoring mirroring progress

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdiskcopy 5
vdisk_id vdisk_name copy_id status sync primary mdisk_grp_id mdisk_grp_name capacity
5 migrateme 0 online yes yes 1 SOURCE_GRP 5.00GB
5 migrateme 1 online no no 2 XIV_Target 5.00GB

In Example 8-41 we display the progress percentage for a specific VDisk.

Example 8-41 Checking the VDisk sync

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisksyncprogress 5
vdisk_id vdisk_name copy_id progress estimated_completion_time
5 migrateme 0 100
5 migrateme 1 30 090831110818

Chapter 8. SVC migration with XIV 265

 In Example 8-42 we display the progress of all out-of-sync mirrors. If a mirror has reached
100% it is not listed unless we specify that particular VDisk.

Example 8-42 Displaying all VDisk mirrors

IBM_2145:SVCCLUSTER_DC1:admin>svcinfo lsvdisksyncprogress
vdisk_id vdisk_name copy_id progress estimated_completion_time
21 arielle_8 1 42 091105193656
24 mitchell_17 1 83 091105185432
32 sharon_1 1 3 091106083130

If copying is going too slowly, you could choose set a higher syncrate when you create the
copy.

You can also increase the syncrate from the default value of 50 (which equals 2 MBps) to 100
(which equals 64 MBps). This change affects the VDisk itself and isvalid for any future copies.
Example 8-43 shows the syntax.

Example 8-43 Changing the VDisk sync rate

IBM_2145:SVCSTGDEMO:admin>svctask chvdisk -syncrate 100 5

Once the estimated completion time passes, we can confirm that the copy process is
complete for VDisk 5. In Example 8-44 the sync is complete.

Example 8-44 VDisk sync completed

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisksyncprogress 5
vdisk_id vdisk_name copy_id progress estimated_completion_time
5 migrateme 0 100
5 migrateme 1 100

8.9.5 Validating a VDisk copy

If you want to confirm that the data between the two VDisk copies is the same, you can run a
validate. This compares the two copies. The command itself completes immediately, but the
validate runs in the background. In Example 8-45 a validate against VDisk 5 is started and
then monitored until it is complete. This validation step is not mandatory and is normally only
needed if an event occurred that makes you doubt the validity of the mirror. It is documented
here in case you want to add an extra layer of certainty to your change.

Example 8-45 Validating a VDisk mirror

IBM_2145:SVCSTGDEMO:admin>svctask repairvdiskcopy -validate 5
IBM_2145:SVCSTGDEMO:admin>svcinfo lsrepairvdiskcopyprogress 5
vdisk_id vdisk_name copy_id task progress est_completion
5 migrateme 0 validate 57 091103155927
5 migrateme 1 validate 57 091103155927
IBM_2145:SVCSTGDEMO:admin>svcinfo lsrepairvdiskcopyprogress 5
vdisk_id vdisk_name copy_id task progress est_completion
5 migrateme 0
5 migrateme 1

266 IBM XIV Storage System: Copy Services and Migration

8.9.6 Removing the VDisk copy
Now that the sync is complete, we can remove copy 0 from the VDisk so that the VDisk
continues to use only copy 1 (which should be on the XIV). We have two methods of
achieving this. We can either split the copies or we can just remove one copy.

Removing a VDisk copy

In Example 8-46, we remove copy 0 from VDisk 5. This effectively discards the VDisk copy on
the source MDisk group. This is simple and quick but has one disadvantage, which is that you
must mirror the data back if you decide to back out the change.

Example 8-46 Removing VDisk copy

IBM_2145:SVCSTGDEMO:admin>svctask rmvdiskcopy -copy 0 5

Splitting the VDisk copies

In Example 8-47 we split the VDisk copies, moving copy 0 (which is on the source MDisk
group) to become a new unmapped VDisk. This means that copy 1 (which is on the target XIV
MDisk group) continues to be accessed by the host as VDisk 5. The advantage of doing this
is that the original VDisk copy remains available if we decide to back out (although it may no
longer be in sync once we split the copies). An additional step is needed to discreetly delete
the new VDisk that was created when we performed the split.

Example 8-47 Splitting the VDisk copies

IBM_2145:SVCSTGDEMO:admin>svctask splitvdiskcopy -copy 0 -name mgrate_old 5
Virtual Disk, id [6], successfully created

Important: Scripts that use VDisk names or IDs should not be affected by the use of VDisk
mirroring, as the VDisk names and IDs do not change. However, if you choose to split the
VDisk copies and continue to use copy 0, it will be a totally new VDisk with a new name
and a new ID.

8.10 Using SVC migration with image mode

This process converts VDisks on non-XIV storage to image mode MDisks on the XIV that can
then be reassigned to a different SVC or released from the SVC altogether. Because of this
extra step, the XIV requires sufficient space to hold both transitional volumes (for image mode
MDisks) and the final destination volumes (for managed mode MDisks).

8.10.1 Create image mode destination volumes on the XIV

On the XIV we must create one new volume for each SVC VDisk that we are migrating (which
must be the same size as the source VDisk, or larger). These are to allow transition of the
VDisk to image mode. To do this, we must determine the size of the VDisk so that we can
create a matching XIV volume.

Chapter 8. SVC migration with XIV 267

 When an SVC volume is created we normally specify a size in GiB (binary GB). For instance,
Example 8-48 creates a 10 GiB Vdisk in Mdisk group 1.

Example 8-48 Create a transitional VDisk

svctask mkvdisk -mdiskgrp 1 -iogrp 0 -name migrateme -size 10 -unit gb

Now to make a matching XIV volume we can either make an XIV volume that is larger than
the source VDisk or one that is exactly the same size. The easy solution is to create a larger
volume. Because the XIV creates volumes in 16 GiB portions (that display in the GUI as
rounded decimal 17 GB chunks), we could create a 17 GB LUN using the XIV and then map
it to the SVC (in this example the SVC host is defined by the XIV as svcstgdemo) and use the
next free LUN ID, which in Example 8-49 is LUN ID 12 (it is different every time).

Example 8-49 XIV commands to create transitional volumes

vol_create size=17 pool="SVC_MigratePool" vol="ImageMode"
map_vol host="svcstgdemo" vol="ImageMode" lun="12"

The drawback of using a larger volume size is that we eventually end up using extra space.
So it is better to create a volume that is exactly the same size. To do this we must know the
size of the VDisk in bytes (by default the SVC shows the VDisk size in GiB, even though it
says GB). In Example 8-50 we first choose to display the size of the VDisk in GB.

Example 8-50 Displaying a VDisk size in GB

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk -filtervalue mdisk_grp_id=1
id name status mdisk_grp_id mdisk_grp_name capacity
6 migrateme online 1 MGR_MDSK_GRP 10.00GB

Example 8-51 displays the size of the same VDisk in bytes.

Example 8-51 Displaying a VDisk size in bytes

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk -filtervalue mdisk_grp_id=1 -bytes
id name status mdisk_grp_id mdisk_grp_name capacity
6 migrateme online 1 MGR_MDSK_GRP 10737418240

Now that we know the size of the source VDisk in bytes, we can divide this by 512 to get the
size in blocks (there are always 512 bytes in a standard SCSI block). 10,737,418,240 bytes
divided by 512 bytes per block is 20,971,520 blocks. This is the size that we use on the XIV to
create our image mode transitional volume.

Example 8-52 shows an XCLI command run on an XIV to create a volume using blocks.

Example 8-52 Create an XIV volume using blocks

vol_create size_blocks=20971520 pool="SVC_MigratePool" vol="ImageBlocks"

268 IBM XIV Storage System: Copy Services and Migration

 The XIV GUI volume creation panel is shown in Figure 8-5. (We must change the GB
drop-down to blocks.)

Figure 8-5 Creating an XIV volume using blocks

Having created the volume, on the XIV we now map it to the SVC (using the XIV GUI or
XCLI).

Then, on the SVC, we can detect it as an unmanaged MDisk using the svctask detectmdisk
command.

8.10.2 Migrate the VDisk to image mode

We now migrate the source VDisk to image mode using the MDisk that we created for
transition. These examples show an MDisk that is 16 GiB (17 GB on the XIV GUI). This
example also shows what will eventually happen if you do not exactly match sizes.

In Example 8-53 we first identify the source VDisk number (by listing VDisks per MDisk
group) and then identify the candidate MDisk (by looking for unmanaged MDisks).

Example 8-53 Identifying candidates

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk -filtervalue mdisk_grp_id=1
id name status mdisk_grp_id mdisk_grp_name capacity
5 migrateme online 1 MGR_MDSK_GRP 10.00GB

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged

id name status mode capacity ctrl_LUN_# controller_name
9 mdisk9 online unmanaged 16.0GB 00000000000000C XIV

In Example 8-53 we identified a source VDisk(5) sized 10 GiB and a target MDisk(9) sized
16 GiB.

Now we migrate the VDisk into image mode without changing MDisk groups (we stay in
group 1, which is where the source VDisk is currently located). The target MDisk must be
unmanaged to be able to do this. If we migrate to a different MDisk group, the extent size of
the target group must be the same as the source group. The advantage of using the same
group is simplicity, but it does mean that the MDisk group contains MDisks from two different
controllers (which is not the best option for normal operations). Example 8-54 shows the
command to start the migration.

Example 8-54 Migrate a VDisk to image mode

svctask migratetoimage -vdisk 5 -mdisk 9 -mdiskgrp 1

Chapter 8. SVC migration with XIV 269

 In Example 8-55, we monitor the migration and wait for it to complete (no response means
that it is complete). We then confirm that the MDisk shows as in image mode and the VDisk
shows as image type.

Example 8-55 Monitoring the migration

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmigrate
IBM_2145:SVCSTGDEMO:admin>
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk
id name status mode mdisk_grp_id mdisk_grp_name capacity
9 mdisk9 online image 1 MGR_MDSK_GRP 16.0GB
IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk
id name status mdisk_grp_id mdisk_grp_name capacity type
5 migrateme online 1 MGR_MDSK_GRP 10.00GB image

We must confirm that the VDisk is in image mode or data loss will occur in the next step. At
this point we must take an outage.

8.10.3 Outage step

At the SVC we un-map the volume (which disrupts the host) and then remove the VDisk. At
the host we must have unmounted the volume (or shut down the host) to ensure that any data
cached at the host has been flushed to the SVC. However, at the SVC itself, if there is still
write data in cache for this VDisk, then you will get a not empty message. You can check
whether this is the case by displaying the fast_write_state for the VDisk with an svcinfo
lsvdisk command. You must wait for the data to flush out of cache, which may take several
minutes.

The commands shown in Example 8-56 apply to a host whose Host ID is 2 and the VDisk ID
is 5.

Example 8-56 Removing the source VDisk

IBM_2145:SVCSTGDEMO:admin>svctask rmvdiskhostmap -host 2 5
IBM_2145:SVCSTGDEMO:admin>svctask rmvdisk 5
IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk
id name status mode mdisk_grp_id mdisk_grp_name capacity
9 mdisk9 online unmanaged 16.0GB

The MDisk is now unmanaged (even though it contains customer data) and could be mapped
to a different SVC cluster or simply mapped directly to a non-SVC host.

8.10.4 Bring the VDisk online

We now create a new managed disk group with an extent size of 1024, but no MDisks. We
could have done this earlier, but it is a very quick step. In Example 8-57 we create MDisk
group 2.

Example 8-57 Creating a new MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name image1024 -ext 1024
MDisk Group, id [2], successfully created

270 IBM XIV Storage System: Copy Services and Migration

 We now use the unmanaged MDisk to create an image mode VDisk in the new MDisk group
and map it to the relevant host. Notice in Example 8-58 that the host ID is 2 and the VDisk
number changed to 10.

Example 8-58 Creating the image mode VDisk

IBM_2145:SVC:admin>svctask mkvdisk -mdiskgrp 2 -iogrp 0 -vtype image -mdisk 9 -name migrated
Virtual Disk, id [10], successfully created
IBM_2145:SVC:admin>svctask mkvdiskhostmap -host 2 10
Virtual Disk to Host map, id [2], successfully created

We can now reboot the host (or scan for new disks) and the LUN will return with data intact.

Important: The VDisk ID and VDisk names were both changed in this example. Scripts
that use the VDisk name or ID (such as those used to automatically create flashcopies)
must be changed to reflect the new name and ID.

8.10.5 Migration from image mode to managed mode

We now must migrate the VDisks from image mode on individual image mode MDisks to
striped mode VDisks in a managed mode MDisk group.

First we create a new managed disk group using volumes on the XIV intended to be used as
the final destination. In Example 8-59, five volumes, each 1632 GB, were created on the XIV
and mapped to the SVC. These are detected as 1520 GiB (because 1632 GB on the XIV GUI
equals 1520 GiB on the SVC GUI). At a certain point the MDisks must also be renamed from
the default names given by the SVC using the svctask chmdisk -name command.

Example 8-59 Listing free MDisks

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmdisk -filtervalue mode=unmanaged
id name status mode capacity ctrl_LUN_# controller
14 mdisk14 online unmanaged 1520.0GB 0000000000000007 XIV
15 mdisk15 online unmanaged 1520.0GB 0000000000000008 XIV
16 mdisk16 online unmanaged 1520.0GB 0000000000000009 XIV
17 mdisk17 online unmanaged 1520.0GB 000000000000000A XIV
18 mdisk18 online unmanaged 1520.0GB 000000000000000B XIV

We create a MDisk group using an extent size of 1024 MB with the five free MDisks. In
Example 8-60 MDisk group 3 is created.

Example 8-60 Creating target MDisk group

IBM_2145:SVCSTGDEMO:admin>svctask mkmdiskgrp -name XIV_Target -mdisk 14:15:16:17:18 -ext 1024
MDisk Group, id [3], successfully created

We then migrate the image mode VDisk (in our case VDisk 5) into the new MDisk group (in
our case group 3), as shown in Example 8-61.

Example 8-61 Migrating the VDisk

IBM_2145:SVCSTGDEMO:admin>svctask migratevdisk -mdiskgrp 3 -vdisk 5

Chapter 8. SVC migration with XIV 271

 The VDisk moves from being in image mode in MDisk group 1 to being in managed mode in
MDisk group 3. Notice in Example 8-62 that it is now 16 GB instead of 10 GB. This is because
we migrated it initially onto a 16 GB image mode MDisk. We should have created a 10 GB
image mode MDisk.

Example 8-62 VDisk space usage

IBM_2145:SVCSTGDEMO:admin>svcinfo lsvdisk
id name status mdisk_grp_id mdisk_grp_name capacity
5 migrated online 3 XIV_Target 16.00GB

In Example 8-63, we monitor the migration and wait for it to complete (no response means
that it is complete).

Example 8-63 Checking that the migrate is complete

IBM_2145:SVCSTGDEMO:admin>svcinfo lsmigrate
IBM_2145:SVCSTGDEMO:admin>

We can clean up the transitional MDisk (which should now be unmanaged), as shown in
Example 8-64.

Example 8-64 Removing the transitional MDisks and MDisk groups

IBM_2145:SVCSTGDEMO:admin>svctask rmmdisk -mdisk 9 2
IBM_2145:SVCSTGDEMO:admin>svctask rmmdiskgrp 2

8.10.6 Remove image mode MDisks

We can then unmap and delete the transition volume on the XIV to free up the space and
reuse that space for other migrations. The XCLI commands shown in Example 8-65 are run
on the XIV (you can also use the XIV GUI).

Example 8-65 Unmapping and deleting the transitional volume

unmap_vol host="svcstgdemo" vol="ImageMode"
vol_delete vol="ImageMode"

8.10.7 Use transitional space as managed space

Provided that all volumes are migrated from non-XIV disk to XIV disks, we can now take the
space on the XIV that was reserved for the transitional image mode MDisks and create new
1632 GB volumes to assign to the SVC. These volumes can be put into the existing MDisk
group or a new MDisk group.

8.10.8 Remove non-XIV MDisks

The non-XIV disk controllers MDisks still exist. We can remove these MDisks and their MDisk
group. Then using the non-XIV disk interface we can un-map these LUNS from the SVC and
reuse or remove the non-XIV disk controller.

272 IBM XIV Storage System: Copy Services and Migration

8.11 Future configuration tasks
The section documents additional tasks that may be necessary after installation and
migration is complete.

8.11.1 Adding additional capacity to the XIV

If and when additional capacity is added to a partially populated XIV, take the following steps:
1. IBM adds the additional modules as a hardware upgrade (known as an MES). The
additional capacity appears as free space once the IBM Service Representative has
completed the process to equip these modules.

Note: If the XIV has the Capacity on Demand (CoD) feature, then no hardware change
or license key is necessary to use available capacity that has not yet been purchased.
The customer simply starts using additional capacity as required until all available
usable space is allocated. The billing process to purchase this capacity occurs
afterwards.

2. From the Pools section of the XIV GUI, right-click the relevant pool and resize it depending
on how the new capacity will be split between any pools. If all the space on the XIV is
dedicated to a single SVC then there must be only one pool.
3. From the Volumes by Pools section of the XIV GUI, add new volumes of 1632 GB until no
more volumes can be created. (There will be space left over, which can be used as scratch
space for testing and for non-SVC hosts.)
4. From the Host section of the XIV GUI, map these new volumes to the relevant SVC
cluster. This completes the XIV portion of the upgrade.
5. From the SVC, detect and then add the new MDisks to the existing managed disk group.
Alternatively, a new managed disk group could be created. Remember that every MDisk
uses a different XIV host port, so a new MDisk group ideally contains several MDisks to
spread the Fibre Channel traffic.
6. If new volumes are added to an existing managed disk group, it may be desirable to
rebalance the existing extents across the new space.

To explain why an extent rebalance may be desirable, the SVC uses one XIV host port as a
preferred port for each MDisk. If a VDisk is striped across eight MDisks, then I/O from that
VDisk will be potentially striped across eight separate I/O ports on the XIV. If the space on
these eight MDisks is fully allocated, then when new capacity is added to the MDisk group,
new VDisks will only be striped across the new MDisks. If additional capacity supplying only
two new MDisks is added, then I/O for VDisks striped across just those two MDisks is only
directed to two host ports on the XIV. This means that the performance characteristics of
these VDisks may be slightly different, despite the fact that all XIV volumes effectively have
the same back end disk performance. The extent rebalance script is located here:
http://www.alphaworks.ibm.com/tech/svctools

8.11.2 Using additional XIV host ports

If additional XIV host ports are zoned to an SVC, then the SVC automatically rebalances its
preferences across all available XIV host ports (provided that we do not exceed the current
SVC limit of 16 WWPNs per WWNN). Depending on the number of modules in an XIV, not all

Chapter 8. SVC migration with XIV 273

 the additional Fibre Channel ports are active. However, they are enabled as more modules
are added.

The suggested host port to capacity ratios are shown in Table 8-6.

Table 8-6 XIV host port ratio as capacity grows

Total XIV modules Capacity allocated XIV host ports
to one SVC cluster (TB) zoned to one SVC cluster

6 27 4

9 43 8

10 50 8

11 54 10

12 61 10

13 66 12

14 73 12

15 79 12

To use additional XIV host ports, run a cable from the SAN switch to the XIV and attach to the
relevant port on the XIV patch panel. Then zone the new XIV host port to the SVC cluster via
the SAN switch. No commands must be run on the XIV.

8.12 Understanding the SVC controller path values

If you display the detailed description of a controller as seen by SVC, for each controller host
port you will see a path value. Because each MDisk has a preferred XIV host port, the
path_count is the number of MDisks using that port multiplied by the number of SVC nodes
(commonly 2 or 4). In Example 8-66 the SVC cluster has two nodes and can access six XIV
volumes (MDisks), so 6 volumes times 2 nodes means 12 paths. These 12 paths will be
distributed in a round-robin fashion across all accessible XIV host ports. Because in this
example there are six XIV ports zoned to the SVC, there will be two paths per port.

We can confirm is that the SVC is utilizing all six XIV interface modules. In Example 8-66 XIV
interface modules 4 through 9 are all clearly zoned to the SVC (because the WWPN ending in
71 is from XIV module 7, the module with WWPN ending in 61 is from XIV module 6, and so
on. To decode the WWPNs use the process described in 8.3.2, “Determining XIV WWPNs”
on page 247.

Example 8-66 Path count as seen by SVC

IBM_2145:SVCSTGDEMO:admin> svcinfo lscontroller 2
id 2
controller_name XIV
WWNN 5001738000510000
mdisk_link_count 6
max_mdisk_link_count 12
degraded no
vendor_id IBM
product_id_low 2810XIV-
product_id_high LUN-0

274 IBM XIV Storage System: Copy Services and Migration

 product_revision 10.0
ctrl_s/n
allow_quorum yes
WWPN 5001738000510171
path_count 2
max_path_count 2
WWPN 5001738000510161
path_count 2
max_path_count 2
WWPN 5001738000510182
path_count 2
max_path_count 2
WWPN 5001738000510151
path_count 2
max_path_count 2
WWPN 5001738000510141
path_count 2
max_path_count 2
WWPN 5001738000510191
path_count 2
max_path_count 2

8.13 SVC with XIV implementation checklist

Table 8-7 contains a checklist that can be used when implementing XIV behind SVC. It
presumes that the XIV has already been installed by the IBM Service Representative.

Table 8-7 XIV implementation checklist

Task Completed? Where to Task
number perform

1 SVC Increase SVC virtualization license if required.

2 XIV Get XIV WWPNs.

3 SVC Get SVC WWPNs.

4 Fabric Zone XIV to SVC (one big zone).

5 XIV Define the SVC cluster as a cluster.

6 XIV Define the SVC nodes as hosts.

7 XIV Add the SVC ports to the hosts.

8 XIV Create a storage pool.

9 XIV Create 1632 GB volumes in the pool.

10 XIV Map the volumes to the SVC cluster.

11 XIV Rename the XIV.

12 SVC Detect the MDisk.

13 SVC Rename the XIV controller.

14 SVC Rename the XIV MDisks.

Chapter 8. SVC migration with XIV 275

 Task Completed? Where to Task
number perform

15 SVC Create an MDisk group.

16 SVC Relocate the quorum disks if necessary.

17 SVC Identify VDisks to migrate.

18 SVC Mirror or migrate your data to XIV.

19 SVC Monitor migration.

20 SVC Remove non-XIV MDisks.

21 SVC Remove non-XIV MDisk group.

22 Non-XIV Unmap LUNs from SVC.

Storage

23 SAN Remove zone that connects SVC to non-XIV

disk.

24 SVC Clear 1630 error that will have been generated

by task 23 (unzoning non-XIV disk from SVC).

276 IBM XIV Storage System: Copy Services and Migration

Related publications

The publications listed in this section are considered particularly suitable for a more detailed
discussion of the topics covered in this book.

IBM Redbooks publications

For information about ordering this publication, refer to “How to get IBM Redbooks
publications” on page 278. This document might be available in softcopy only:
򐂰 IBM XIV Storage System: Architecture, Implementation, and Usage, SG24-7659
򐂰 Introduction to Storage Area Networks, SG24-5470

Other publications
These publications are also relevant as further information sources:
򐂰 IBM XIV Storage System Installation and Service Manual, GA32-0590
򐂰 IBM XIV Storage System XCLI Manual, GC27-2213
򐂰 IBM XIV Storage System Introduction and Theory of Operations, GC27-2214
򐂰 IBM XIV Storage System Host System, GC27-2215
򐂰 IBM XIV Storage System Model 2810 Installation Planning Guide, GC52-1327-01
򐂰 IBM XIV Storage System Pre-Installation Network Planning Guide for Customer
Configuration, GC52-1328-01
򐂰 XCLI Reference Guide, GC27-2213-00
򐂰 Host System Attachment Guide for Windows- Installation Guide
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/index.jsp
򐂰 The iSCSI User Guide
http://download.microsoft.com/download/a/e/9/ae91dea1-66d9-417c-ade4-92d824b871
af/uguide.doc
򐂰 AIX 5L System Management Concepts: Operating System and Devices
http://publib16.boulder.ibm.com/pseries/en_US/aixbman/admnconc/hotplug_mgmt.htm
#mpioconcepts
򐂰 System Management Guide: Operating System and Devices for AIX 5L
http://publib16.boulder.ibm.com/pseries/en_US/aixbman/baseadmn/manage_mpio.htm
򐂰 Host System Attachment Guide for Linux, which can be found at the XIV Storage System
Information Center
http://publib.boulder.ibm.com/infocenter/ibmxiv/r2/index.jsp
򐂰 Sun StorEdge Traffic Manager Software 4.4 Release Notes
http://dlc.sun.com/pdf/819-5604-17/819-5604-17.pdf

© Copyright IBM Corp. 2010. All rights reserved. 277

 򐂰 Fibre Channel SAN Configuration Guide
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_san_cfg.pdf
򐂰 Basic System Administration (VMware Guide)
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_admin_guide.pdf
򐂰 Configuration of iSCSI initiators with VMware ESX 3.5 Update 2
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_iscsi_san_cfg.pdf
򐂰 ESX Server 3 Configuration Guide
http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_3_server_config.pdf

Online resources
These Web sites are also relevant as further information sources:
򐂰 IBM XIV Storage Web site
http://www.ibm.com/systems/storage/disk/xiv/index.html
򐂰 System Storage Interoperability Center (SSIC)
http://www.ibm.com/systems/support/storage/config/ssic/index.jsp
򐂰 Storage Networking Industry Association (SNIA) Web site
http://www.snia.org/
򐂰 IBM Director Software Download Matrix page
http://www.ibm.com/systems/management/director/downloads.html
򐂰 IBM Director documentation
http://www.ibm.com/systems/management/director/

How to get IBM Redbooks publications

You can search for, view, or download IBM Redbooks publications, Redpapers, Technotes,
draft publications and Additional materials, as well as order hardcopy IBM Redbooks
publications, at this Web site:
ibm.com/redbooks

Help from IBM

IBM Support and downloads
ibm.com/support

IBM Global Services

ibm.com/services

278 IBM XIV Storage System: Copy Services and Migration

Index
create 23–24
Symbols delete 32
_Define_Target_(Legacy 193 expand 25
_Pre-Data_Migration_Steps 196 given volume 172
_Toc227991967 191 maximum number 112
_Toc227991968 191 new snapshots 28
_Toc227991969 192 other volumes 156
_Toc227991970 193 remove volume 94, 104, 132, 159
_Toc227991971 195 setup 130
_Toc227991972 196 Storage Pool 24–25
_Toc227991973 196 synchronization status 157, 172
_Toc227991974 197 consistency group (CG) 7–8, 23–25, 27, 72, 76–77, 93,
_Toc227991975 200 126–127, 130–132, 149
_Toc227991976 201 consistent state 130
_Toc227991977 201 copy functions xi, 1
_Toc227991978 202 copy on write 5
copy pairs 118
A Copy-on-Write 46
activation states 81 coupling 72, 126
Active Directory 45 crash consistency 94
Active/Active Create Mirror 126
188 Create Slave 127
active/active 187–189, 194, 218, 223
active/cctive 188 D
Active/Passive 189 D3000 227
active/passive 187, 189, 191, 218, 223 data corruption protection 110
ADT 227–228 Data Migration 185–187, 193
application-consistent data 107, 109 target 187
Asynchronous mirroring 71–75, 127, 149–150, 157, 159, data migration xi, 1, 118, 185–187
169–170 back-out 200, 219–220
automatic deletion 6, 8, 21–23 checklist 220
monitoring 208
B object 197–200, 207
background copy 66, 187–188, 200, 208 steps 190, 201, 218–219
backup xi, 1, 6, 9 synchronization 201
backup script 34, 36 Data Migration (DM) volume 197
Data Protection for Exchange Server 55
deactivation 111
C define connections 119
cache 212 delete snapshot 6, 19
candidate MDisk 269 deletion priority 6, 8, 11–13
Capacity on Demand (CoD 246 dependent writes 94
cctive/passive 189 designation 77
CG 72 Destination Name 198, 207
cg_list 31 Destination Pool 198, 207
change role 77, 100, 134, 136, 142, 166 detectmdisk 257, 262, 264
clone 46 Device Manager 211
cluster 195–196, 199, 201 Disaster Recovery 133, 169
communication failure 111 Disaster Recovery (DR) 72, 75, 77, 133, 136, 149, 173,
config_get 128 179
Consistency Group 23 Disaster Recovery protection 110
consistency group 8, 24, 77, 92, 172 diskpart 215
add volume 93, 103, 132 dm_list 200, 203
configuration 155 DR test 100, 109, 134

© Copyright IBM Corp. 2010. All rights reserved. 279

DR testing Migration 271
GUI steps 179 SVC migration 267
DS4000 190, 192, 195, 227 initialization 78, 97
DS5000 227 initialization rate 90, 209–210
DS6000 189, 230 initializing state 129
DS8000 189, 230–231 initiator 113–114, 116, 186, 189, 191
DSMagent 60 Instant Restore 63
duplicate 6, 9–10 Instant Restore (IR) 62
Duplicate Snapshot 6, 8–10, 110 interval 73
duplicate snapshot 6, 9, 11 IO group 264
creation date 6, 10–11 ipinterface_list 113–114
creation time 6 iSCSI 113
iSCSI ports 91

E
environment variables 205–206 K
ESS 189, 195, 207, 229 KB 5
Ethernet 113 Keep Source Updated 188, 199, 208
Ethernet switch 176
event log 211, 227
Exchange Server 45 L
extent size 252–254 last consistent snapshot 136
new managed disk group 261 timestamp 137
last_consistent 135
last_replicated 100
F last_replicated snapshot 100–101, 157, 167–168, 172
fail-over 188–189, 218 Level 1.0 58–60
failure 72 license 258, 273, 275
fan-in 89 link
fan-out 89 failure 136
fc_port_config 117 link failure 169
fc_port_list 113, 117 link status 79
Fibre Channel 113 link utilization 79
Fibre Channel (FC) 71, 86, 91 Linux x86 192, 230
Fibre Channel ports 91, 116, 176 local site 72, 79, 88, 127–128, 133, 166, 170, 181
FLASHCOPYMANAGER 59 Master volumes 138
old Master volumes 138
lock Snapshot 35
G Logical Unit Number (LUN) 77, 92
GiB 250–252 LUN ID 187, 197–198, 200
Graphical User Interface LUN mapping 197
Remote Mirroring performance statistics 112 LUN numbering 192, 223
Graphical User Interface (GUI) 8, 11, 91, 113 LUN0 192, 195, 221, 223
GUI example 176
GUI step 178
M
MachinePool Editor 51
H managed mode 267, 271–272
Host Attachment Kit 58 map_vol host 268
Host Attachment Procedure 197 Master 72, 77
host server 186–187, 201 Master peer 78, 107–108, 133, 151, 168
I/O requests 187 master peer
host_add_port host 256 actual data 108
Deactivate XIV remote mirroring 108
I remote mirroring 108
IBM Tivoli Storage FlashCopy Manager 41 XIV remote mirroring 108
IBM Tivoli Storage Manager V6 42 Master role 74, 76–77, 96, 126, 133–134, 136, 161, 166
IBM XIV Master volume 5, 10, 15–16, 73, 78, 81, 101, 132,
data migration solution 186 134–136, 167–168
storage 187 periodic consistent copies 171
image mode 245, 252, 260, 267 master volume 5, 9, 14

280 IBM XIV Storage System: Copy Services and Migration

 actual data 98 N
additional Snapshot 110 N10116 45
Changing role 101 naming convention 9
Deactivate XIV remote mirroring 109, 111 no source updating 187–188
duplicate snapshot 9, 20 normal operation 87, 91, 133, 169
identical copy 80 data flow 91
Reactivate XIV remote mirroring 111
XIV remote mirroring 104
max_initialization_rate 209–211 O
max_resync_rate 209–210 OK button 13–14
max_syncjob_rate 209–210 ongoing operation 78
MDisk 249, 251 Open Office 223
MDisk group 252–254 operating system (OS) 46–47, 51, 68, 70
free capacity 255 original data 46
image mode VDisk 271 exact read-only copy 46
MDisks 245, 249–250 full copy 46
metadata 5, 10, 100 original snapshot 6, 9, 12
Microsoft Excel 223 duplicate snapshot points 6
Microsoft Exchange Server 46 mirror copy 21
Microsoft SQL Server 46
Microsoft Volume Shadow Copy Services (VSS) 41
migration
P
peer 72, 76
speed 189, 208
designations 77
Mirror
role 77
initialization 78
pointer 5, 15, 34
ongoing operation 78
Point-in-Time (PIT) 74
mirror
Point-in-Time (PiT) 74
activation 97
point-in-time copy 196
delete 106
port configuration 115
reactivation 102
port role 117
resynchronization 102
portperfshow 211
Mirror coupling 92, 96
ports 116, 176
deactivate 99
power loss consistency 94
mirror coupling 97
Primary 72, 77, 96
mirror_activate 145
Primary site
mirror_change_role 140
Failure 107
mirror_create_snapshot 72
primary site 78, 101, 103, 107, 131, 133, 156, 161, 166,
mirror_lis 129
170
mirror_list 145, 147
full initialization 170
mirror_list command 129–130
Master volumes 142–143
mirror_statistics_get 112
Master volumes/CG, servers 169
mirrored Cg
production activities 169
mirrored volume 105
production server 147
Mirrored consistency group
Role changeover 142–143
mirrored volume 104
primary system 114, 117
Mirroring
source volumes 169
statistics 112
Primary XIV 74, 81, 122, 155
status 79
primary XIV
mirroring
Master volume 147
activation 126
Mirror statuses 130, 145
setup 126
original direction 136
target 86
Remote Mirroring menu 147
mirroring schemes 75
Slave volumes 142–143
most_recent 100
provider 41, 46
most_recent Snapshot 159, 169–170, 172
MSCS Cluster 199
Multipathing 223–225, 230 Q
MySQL 31–32, 34 queue depth 249, 251
quiesce 47

Index 281
R secondary XIV 72–73, 81, 128, 136, 155, 161
RDAC 227–228 corresponding consistency group 155
reactivation 111 Mirror statuses 130, 145
Recovery Point Objective (RPO) 73, 149 Slave volume 147
Redbooks Web site service 45
Contact us xiv shadow copy 44
RedHat 192, 231 persistent information 45
redirect on write 5, 22, 66 single XIV
redirect-on-write 47 footprint 253
Remote Mirror 33, 71–72, 113–114 rack 253
activate 144 Storage Pool 92
remote mirror pair 198 system 82, 85, 89
Remote Mirroring 8, 23, 71–73, 80, 125–126, 129, 132, Slave 72, 77
159 Slave peer 74, 100, 132–133, 136, 151–152, 161
actions 86 slave peer
activate 155 consistent data 74
delete 165 Slave Pool 127
function 113 Slave role 73, 96, 126, 134–136, 167, 179–181
implementation 113 Slave volume 76–79, 81, 127, 129, 132–133, 150, 152
planning 111 Changing role 100
usage 82 consistent state 77
remote mirroring 116 whole group 77
consistency groups 72, 112 snap_group 31, 37
Fibre channel paths 113 snap_group_duplicate 178
first step 81 Snapshot
single unit 96 automatic deletion 6, 8, 21
synchronous 125 creation 8, 31
XIV systems 86 deletion priority 6, 11–12
remote site 71–72, 78, 127, 133, 137, 166, 170, 177, 181 details 31
secondary peers 103 duplicate 6, 9–10
Slave volumes 138 last_consistent 135
standby server 137 last_replicated 172
requestor 45, 47 lock 35
resize 113, 208, 214–215 most_recent 172
resize operation 66 restore 37–38
Resynchronization 102, 107, 136 size 31
resynchronization 136, 169 snapshot 1, 5
role 72, 76–77, 100 delete 6, 19
change 78 duplicate 6, 9–10
switching 78 last consistent 136
role reversal 166 locked 6, 9
RPO 73, 153, 162 naming convention 9
RPO_Lagging 81, 161 snapshot group 28, 30
RPO_OK 81 snapshot volume 4–5, 7, 16, 46, 56
snapshot_delete 20
Snapshot/volume copy 109
S SNMP traps 112
SAN boot 68, 186 Source LUN 198, 208, 217
SAN connectivit 113 source MDisk group
schedule 150, 162 extent size 259
schedule interval 73 Image Mode MDisks 261
Schedule Management 151 Source Target System 198, 207
schedule_create schedule 154, 163 source updating 187–188, 201, 219
SCSI initiator 191 source updating option 187
SDDPCM 237 SQL Server 45
Secondary 72, 77 SqlServerWriter 46
secondary site 131–135, 156, 166, 169–170 standby 128
mirror relation 137 state 77
remote mirroring 134 consistent 130
Role changeover 139–140 initializing 129

282 IBM XIV Storage System: Copy Services and Migration

Storage Pool 6, 10, 21 TSM Client Agent 59
additional allocations 21
Storage pool
consistency group 94 V
different CG 93 VDisk 245, 250, 252
existing volumes 94 progress percentage 265
storage pool 93, 127, 130, 132, 155, 157, 275 striping 250
storage system 133, 170, 185–186, 188 VDisk 5
storageadmin 50 copy 0 267
SuSE 192 VDisk copy 260, 266
SVC 243–245 VDisks 245, 252, 254
firmware version 244, 254, 260 virtualization license 258
MDisk group 252, 259, 261 virtualization license limit 258, 273, 275
mirror 245, 265 VMware 68–69
quorum disk 253 VMware ESX 278
zone 245 VMware File System (VMFS) 68
zoning 245 vol_copy 68
SVC cluster 244–246 vol_lock 19
svcinfo lsmdisk 251–253 volume
svctask 254–255, 257 resize 113
svctask chlicense 258 Volume Copy 65, 68
switch role 134 OS image 68
switch roles 103 volume copy 1, 66, 68, 109
switch_role 77 volume mirror
sync job 72–74, 150–151, 159, 161 coupling 97–99, 104
most_recent snapshot 172 setup 151
schedule 150 Volume Shadow Copy Services (VSS) 44
Sync Type 127 VSS 41, 44
Synchronised 187 provider 41, 46
synchronization 201, 209 requestor 45
rate 90 service 45
status 80 writer 45
Synchronized 130 VSS architecture 45
synchronized 204, 209, 218 vssadmin 51
synchronous 71
Synchronous Mirror 166 W
synchronous mirroring 73 write access 100
syncrate 266 writer 45, 47
WWPN 192–193, 195, 203, 216, 245, 247
T
target 86, 114–115, 187–189, 195 X
Target System 127 XCLI 10, 12–13, 90–91, 112–113, 116, 126, 128, 130,
target volume 66, 69, 127–128, 137, 165 150, 154, 159
target_config_sync_rates 90 XCLI command 72, 102, 117, 128, 154, 178, 245, 247,
target_config_sync_rates. 102 268
target_list 203, 210 XCLI example 122
tdpexcc 58 XCLI session 10, 12, 15, 128, 140, 143, 154
Test Data Migration 199–200, 204, 208 XIV 1, 5, 41–42, 44, 71–73, 149–151, 243–245
the 107 end disk controllers 243
thick to thin migration 212 XIV 1 104, 108
thin provisioning 212 additional Master Volume 104
Tivoli Storage FlashCopy Manager 41–43 available, deactivate mirrors 107
detailed information 63 Complete destruction 108
prerequesites 53 complete destruction 108
wizard 52 Deactivate mirror 107
XIV VSS Provider 41–42 Map Master peers 108
Tivoli Storage Manager (TSM) 43, 58 Master consistency group 104
Tivoli Storage Manager for Advanced Copy Services 43 Master peer 109
TPC (Tivoli Storage Productivity Centre) 245 production data 110
transfer rate 209

Index 283
 remote mirroring peers 110
remote targets 108
Slave peer 108
Slave volume 108
volume copy 110
XIV remote mirroring 108–109
XIV system 107, 110
XIV systems 110–111
XIV 2 104, 107
additional Slave volume 104
consistency group 104
Disaster Recovery testing 109
DR servers 108
other functions 109
production applications 107
production changes 107
production workload 107–108
Slave peer 108
Slave volume 104, 109, 111
Unmap Master peers 108
XIV asynchronous mirroring 74, 83–84, 97–98, 149
XIV Command Line Interface (XCLI) 202
XIV GUI 76, 79, 126, 150, 244, 247–248
Host section 273
Pools section 273
XIV host port 248–249, 251
XIV mirroring 84, 86, 91–92, 98, 100, 102
Advantages 112
XIV remote mirroring 82–83, 85
normal operation 107–109
Planned deactivation 100
user deactivation 111
XIV Storage
System xi, 1, 5, 71, 113, 185–186, 243
XIV Storage System 46–47, 65–67, 72, 93, 112
primary IP address 50
snapshot operations 52
XIV subsystem 5–6
XIV system 2, 73–75, 150, 152, 154
available disk drives 92
available physical disk drives 93
mirroring connections 111
planned outage 82
same number 91
single Storage Pool 96
XIV mirroring target 86
XIV Snapshot function 85
XIV volume 68–69, 78, 92, 104, 149, 250–252
XIV VSS Provider
configuration 49
XIV VSS provider 41, 47
xiv_devlist 238
XIVPCM 237

Z
zoning 190–192, 197, 216

284 IBM XIV Storage System: Copy Services and Migration

IBM XIV Storage System: Copy
Services and Migration
IBM XIV Storage System: Copy
Services and Migration
IBM XIV Storage System: Copy Services and Migration
IBM XIV Storage System: Copy Services and Migration
IBM XIV Storage System: Copy
Services and Migration
IBM XIV Storage System: Copy
Services and Migration
 Back cover ®

IBM XIV Storage System:

Copy Services and
Migration ®

Learn copy and This IBM Redbooks publication provides a practical understanding of
the XIV Storage System copy and migration functions. The XIV Storage INTERNATIONAL
migration functions
System has a rich set of copy functions suited for various data TECHNICAL
and explore practical
protection scenarios, which enables clients to enhance their business SUPPORT
scenarios continuance, data migration, and online backup solutions. These ORGANIZATION
functions allow point-in-time copies, known as snapshots and full
Integrate Snapshots volume copies, and also include remote copy capabilities in either
with Tivoli FlashCopy synchronous or asynchronous mode. These functions are included in
Manager the XIV software, and all their features are available at no additional
charge. BUILDING TECHNICAL
Understand The various copy functions are reviewed under separate chapters that INFORMATION BASED ON
include detailed information about usage as well as practical PRACTICAL EXPERIENCE
SVC-based
illustrations.
migrations IBM Redbooks are developed
This book also discusses how to integrate the snapshot function with by the IBM International
IBM Tivoli FlashCopy manager, explains the XIV built-in migration Technical Support
capability, and presents migration alternatives based on the San Organization. Experts from
Volume Controller (SVC). IBM, Customers and Partners
from around the world create
timely technical information
based on realistic scenarios.
Specific recommendations
are provided to help you
implement IT solutions more
effectively in your
environment.

For more information:

ibm.com/redbooks

SG24-7759-00 ISBN 0738434221