IBM PowerVM Best Practices

ibm.com/redbooks
IBM PowerVM
Best Practices
Adriano de Almeida
Rafael Antonioli
Urban Biel
Sylvain Delabarre
Bartłomiej Grabowski
Kristian Milos
Fray L Rodríguez
A collection of recommended practices to
enhance your use of the PowerVM features
A resource to build on knowledge
found in other PowerVM documents
A valuable reference for experienced
IT specialists and IT architects
Front cover

IBM PowerVM Best Practices
October 2012
International Technical Support Organization
SG24-8062-00

© Copyright International Business Machines Corporation 2012. All rights reserved.
Note to U.S. Government Users Restricted Rights -- Use, duplication or disclosure restricted by GSA ADP
Schedule Contract with IBM Corp.
First Edition (October 2012)
This edition applies to:
PowerVM Enterprise Edition
Virtual I/O Server Version 2.2.1.4 (product number 5765-G34)
AIX Version 7.1 (product number 5765-G99)
IBM i Version 7.1 (product number 5770-SS1)
HMC Version 7.7.4.0 SP02
POWER7 System Firmware Version AL730_87
Note: Before using this information and the product it supports, read the information in
“Notices” on page xiii.

© Copyright IBM Corp. 2012. All rights reserved. iii
Contents
Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .ix
Examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .xi
Notices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
Trademarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
The team who wrote this book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
Now you can become a published author, too! . . . . . . . . . . . . . . . . . . . . . . . xvii
Comments welcome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xvii
Stay connected to IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviii
Chapter 1. Introduction and planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Keeping track of PowerVM features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Virtual I/O Server specifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.2.1 Virtual I/O Server minimum requirements . . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Configuration considerations of the Virtual I/O Server . . . . . . . . . . . . 5
1.2.3 Logical Volume Manager limits in a Virtual I/O Server . . . . . . . . . . . . 6
1.3 Planning your Virtual I/O Server environment . . . . . . . . . . . . . . . . . . . . . . . 6
1.3.1 System Planning Tool . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.2 Document as you go . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.3 Hardware planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.4 Sizing your Virtual I/O Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.5 IBM Systems Workload Estimator. . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.6 Shared or dedicated resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3.7 Single, dual, or multiple Virtual I/O Servers . . . . . . . . . . . . . . . . . . . 11
1.3.8 Network and storage components . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.3.9 Slot numbering and naming conventions . . . . . . . . . . . . . . . . . . . . . 12
1.3.10 Operating systems on virtual client partitions . . . . . . . . . . . . . . . . . 13
Chapter 2. Installation, migration, and configuration . . . . . . . . . . . . . . . . 15
2.1 Creating a Virtual I/O Server profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.1.1 Processing mode - shared or dedicated . . . . . . . . . . . . . . . . . . . . . . 16
2.1.2 Processing settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.1.3 Memory settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.1.4 Physical I/O adapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

iv IBM PowerVM Best Practices
2.1.5 Virtual I/O adapters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.1.6 Deploying a Virtual I/O Server with the System Planning Tool . . . . . 24
2.2 Virtual I/O Server installation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.3 Updating fix packs, service packs, and interim fixes . . . . . . . . . . . . . . . . . 25
2.3.1 Virtual I/O Server service strategy . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3.2 Approved third-party applications in the Virtual I/O Server . . . . . . . . 26
2.3.3 Applying fix packs, service packs, and interim fixes . . . . . . . . . . . . . 27
2.4 Virtual I/O Server migration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.4.1 Options to migrate the Virtual I/O Server . . . . . . . . . . . . . . . . . . . . . 28
2.4.2 Virtual I/O Server migration considerations. . . . . . . . . . . . . . . . . . . . 28
2.4.3 Multipathing software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
Chapter 3. Administration and maintenance . . . . . . . . . . . . . . . . . . . . . . . 31
3.1 Backing up and restoring the Virtual I/O Server . . . . . . . . . . . . . . . . . . . . 32
3.1.1 When to back up the Virtual I/O Server. . . . . . . . . . . . . . . . . . . . . . . 32
3.1.2 Virtual I/O Server backup strategy . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1.3 Backing up user-defined virtual devices . . . . . . . . . . . . . . . . . . . . . . 35
3.1.4 Restoring the Virtual I/O Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.1.5 NIM server resilience . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.2 Dynamic logical partition operations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.2.1 Dynamically adding virtual Fibre Channel adapters . . . . . . . . . . . . . 43
3.3 Virtual media repository. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Power Systems server shutdown and startup . . . . . . . . . . . . . . . . . . . . . . 45
Chapter 4. Networking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
4.1 General networking considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.1.1 Shared Ethernet Adapter considerations . . . . . . . . . . . . . . . . . . . . . 51
4.1.2 Maximum transmission unit best practices . . . . . . . . . . . . . . . . . . . . 52
4.1.3 Network bandwidth tuning. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Single Virtual I/O Server . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 Virtual network redundancy and failover technology. . . . . . . . . . . . . . . . . 55
4.3.1 Dual Virtual I/O Server with VLAN tagging . . . . . . . . . . . . . . . . . . . . 56
4.3.2 Shared Ethernet Adapter failover with load sharing . . . . . . . . . . . . . 58
Chapter 5. Storage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.1 Storage Considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1.1 Virtual I/O Server rootvg storage. . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1.2 Multipathing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.1.3 Mixing virtual SCSI and NPIV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.1.4 Fibre Channel adapter configuration. . . . . . . . . . . . . . . . . . . . . . . . . 67
5.2 Virtual Small Computer System Interface . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.2.1 When to use a virtual Small Computer System Interface . . . . . . . . . 70
5.2.2 Configuring the Virtual I/O Server with a virtual SCSI. . . . . . . . . . . . 71
5.2.3 Exporting virtual Small Computer System Interface storage. . . . . . . 76

Contents v
5.2.4 Configuring the Virtual I/O client with virtual SCSI . . . . . . . . . . . . . . 79
5.3 Shared Storage Pools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3.1 Requirements per Shared Storage Pool node . . . . . . . . . . . . . . . . . 82
5.3.2 Shared Storage Pools specifications . . . . . . . . . . . . . . . . . . . . . . . . 83
5.3.3 When to use Shared Storage Pools . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.3.4 Creating the Shared Storage Pools . . . . . . . . . . . . . . . . . . . . . . . . . 84
5.3.5 SAN storage considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.3.6 Monitoring storage pool capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.3.7 Network considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
5.4 N-Port ID Virtualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4.1 When to use N-Port ID Virtualization . . . . . . . . . . . . . . . . . . . . . . . . 88
5.4.2 Configuring the Virtual I/O Server with N-Port ID Virtualization . . . . 89
5.4.3 Configuring the virtual I/O client with N-Port ID Virtualization . . . . . . 91
Chapter 6. Performance monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
6.1 Measuring Virtual I/O Server performance . . . . . . . . . . . . . . . . . . . . . . . . 94
6.1.1 Measuring short-term performance. . . . . . . . . . . . . . . . . . . . . . . . . . 94
6.1.2 Network and Shared Ethernet Adapter monitoring . . . . . . . . . . . . . . 97
6.1.3 Measuring long-term performance . . . . . . . . . . . . . . . . . . . . . . . . . 100
Chapter 7. Security and advanced IBM PowerVM features . . . . . . . . . . . 105
7.1 Virtual I/O Server security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.1.1 IBM PowerVM Hypervisor security . . . . . . . . . . . . . . . . . . . . . . . . . 106
7.1.2 Virtual I/O Server network services. . . . . . . . . . . . . . . . . . . . . . . . . 106
7.1.3 Viosecure command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
7.2 IBM PowerSC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.3 Live Partition Mobility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
7.3.1 General considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.3.2 Implementing Live Partition Mobility . . . . . . . . . . . . . . . . . . . . . . . . 110
7.3.3 Storage considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.3.4 Network considerations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.4 Active Memory Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.4.1 When to use Active Memory Sharing . . . . . . . . . . . . . . . . . . . . . . . 114
7.4.2 Implementing Active Memory Sharing . . . . . . . . . . . . . . . . . . . . . . 116
7.4.3 Active Memory Deduplication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Abbreviations and acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Related publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
IBM Redbooks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Other publications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Online resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Help from IBM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125

vi IBM PowerVM Best Practices
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127

© Copyright IBM Corp. 2012. All rights reserved. vii
Figures
2-1 Select processor type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2-2 Processing Settings best practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2-3 Setting the maximum number of virtual adapters . . . . . . . . . . . . . . . . . . . 22
2-4 Partition profile properties for source and target virtual adapters . . . . . . . 23
3-1 NIM resilience solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3-2 Partition context menu . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3-3 Save the running partition profile to a new profile. . . . . . . . . . . . . . . . . . . 43
3-4 Overwrite the existing partition profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3-5 The Power off system option is turned off . . . . . . . . . . . . . . . . . . . . . . . . 45
3-6 Automatically start when the managed system is powered on . . . . . . . . . 46
3-7 Partition start policy. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4-1 Dual Virtual I/O Server configuration with two virtual switches . . . . . . . . . 57
4-2 Dual Virtual I/O Server configuration with SEA and load balancing . . . . . 59
5-1 Virtual SCSI client with multipathing from dual Virtual I/O Servers . . . . . . 65
5-2 Changing the default Maximum virtual adapters number . . . . . . . . . . . . . 73
5-3 Select the appropriate virtual I/O client . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5-4 NPIV configuration with dual Virtual I/O Servers. . . . . . . . . . . . . . . . . . . . 91
7-1 Live Partition Mobility migration validation function. . . . . . . . . . . . . . . . . 111
7-2 Partitions with dedicated memory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7-3 Logical partitions with shared memory that run different regions . . . . . . 115
7-4 Partitions that support day and night workloads . . . . . . . . . . . . . . . . . . . 115
7-5 Sporadically used logical partitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
7-6 AMS memory weight on a logical partition . . . . . . . . . . . . . . . . . . . . . . . 117

viii IBM PowerVM Best Practices

© Copyright IBM Corp. 2012. All rights reserved. ix
Tables
1-1 PowerVM features. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1-2 Minimum resources that are required for a Virtual I/O Server. . . . . . . . . . . 4
1-3 Limitations for storage management. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1-4 Network and storage components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2-1 Processor weighting example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4-1 Terminology that is used in this chapter . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4-2 Typical maximum transmission units (MTUs) . . . . . . . . . . . . . . . . . . . . . . 53
5-1 Booting from a virtual SCSI with NPIV for data. . . . . . . . . . . . . . . . . . . . . 67
5-2 Recommended settings for virtual I/O client virtual SCSI disks . . . . . . . . 81
5-3 Recommended settings for virtual I/O client virtual SCSI adapters . . . . . 82
5-4 Shared Storage Pools minimum requirements per node . . . . . . . . . . . . . 82
5-5 Shared Storage Pools minimums and maximums . . . . . . . . . . . . . . . . . . 83
7-1 Default open ports on the Virtual I/O Server . . . . . . . . . . . . . . . . . . . . . . 106
7-2 IBM PowerSC security standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

© Copyright IBM Corp. 2012. All rights reserved. xi
Examples
2-1 Using the alt_root_vg command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3-1 Virtual I/O Server backup to a remote file system. . . . . . . . . . . . . . . . . . . 34
3-2 Daily viosbr schedule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3-3 List device mappings from a viosbr backup . . . . . . . . . . . . . . . . . . . . . . . 36
3-4 The installios command from the HMC . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3-5 Preparing NIM resources for Virtual I/O Server restore . . . . . . . . . . . . . . 40
3-6 HMC CLI startup commands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
5-1 Manage the DS4000 and DS5000 disk driver. . . . . . . . . . . . . . . . . . . . . . 66
5-2 The fscsi attribute modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5-3 Monitoring fcs adapter statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5-4 fcs attributes modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5-5 Check num_cmd_elems using the SDDPCM pcmpath command . . . . . . . 69
5-6 HMC virtual slots listing with a single Virtual I/O Server . . . . . . . . . . . . . . 72
5-7 Example of a virtual storage mapping convention . . . . . . . . . . . . . . . . . . 75
5-8 DS4800 mpio_get_config command . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
5-9 Change the reserve_policy disk attribute . . . . . . . . . . . . . . . . . . . . . . . . . 78
5-10 Listing the hdisk0 and vscsi0 path attributes . . . . . . . . . . . . . . . . . . . . . 80
5-11 Changing the vscsi0 priority for hdisk0 . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5-12 Creating the cluster, adding additional nodes, and assigning storage to the
virtual I/O client. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
5-13 The prepdev command to prepare the disk. . . . . . . . . . . . . . . . . . . . . . . 85
5-14 Highlighted disks which are in the storage pool . . . . . . . . . . . . . . . . . . . 86
5-15 Using the alert command to configure threshold warnings . . . . . . . . . . 87
5-16 Changing attributes on a Fibre Channel adapter on a virtual I/O client . 92
6-1 Display of nmon interactive mode commands . . . . . . . . . . . . . . . . . . . . . 95
6-2 Exporting environment variable in ksh shell to nmon . . . . . . . . . . . . . . . . 95
6-3 Extended disk output of hdisk4 and vhost1 using viostat . . . . . . . . . . . . 97
6-4 Output from the netstat command. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6-5 Output of entstat on SEA. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6-6 Enabling accounting on the Shared Ethernet Adapter . . . . . . . . . . . . . . . 98
6-7 seastat statistics that are filtered by IP address. . . . . . . . . . . . . . . . . . . . 98
6-8 Example use of the fcstat command. . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
6-9 Example of a recording in the nmon format . . . . . . . . . . . . . . . . . . . . . . . 101
6-10 Listing of the available agents. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
6-11 The lslparutil command . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
6-12 How to calculate processor utilization . . . . . . . . . . . . . . . . . . . . . . . . . . 104
7-1 Exporting viosecure high-level rules to XML . . . . . . . . . . . . . . . . . . . . . 107
7-2 Applying customized viosecure rules . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

xii IBM PowerVM Best Practices
7-3 Specifying a new name for the VTD . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

© Copyright IBM Corp. 2012. All rights reserved. xiii
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 grant 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 websites are provided for convenience only and do not in any
manner serve as an endorsement of those websites. The materials at those websites are not part of the materials
for this IBM product and use of those websites 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.
Any performance data contained herein was determined in a controlled environment. Therefore, the results
obtained in other operating environments may vary significantly. Some measurements may have been made on
development-level systems and there is no guarantee that these measurements will be the same on generally
available systems. Furthermore, some measurements may have been estimated through extrapolation. Actual
results may vary. Users of this document should verify the applicable data for their specific environment.
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. 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 IBM's application programming interfaces.

xiv IBM PowerVM Best Practices
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:
Active Memory™
AIX®
BladeCenter®
DS4000®
DS8000®
Enterprise Storage Server®
Focal Point™
GDPS®
Geographically Dispersed
Parallel Sysplex™
GPFS™
HACMP™
IBM®
iSeries®
Micro-Partitioning®
Parallel Sysplex®
PartnerWorld®
Power Systems™
POWER7®
PowerHA®
PowerVM®
POWER®
PureFlex™
PureSystems™
Redbooks®
Redpaper™
Redbooks (logo) ®
System p®
SystemMirror®
Tivoli®
The following terms are trademarks of other companies:
Linux is a trademark of
Linus Torvalds in the
United States, other
countries, or both.
Microsoft, 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.
Other company, product, or service names may be trademarks or service marks of others.

© Copyright IBM Corp. 2012. All rights reserved. xv
Preface
This IBM® Redbooks® publication provides best practices for planning,
installing, maintaining, and monitoring the IBM PowerVM® Enterprise Edition
virtualization features on IBM POWER7® processor technology-based servers.
PowerVM is a combination of hardware, PowerVM Hypervisor, and software,
which includes other virtualization features, such as the Virtual I/O Server.
This publication is intended for experienced IT specialists and IT architects who
want to learn about PowerVM best practices, and focuses on the following topics:
򐂰 Planning and general best practices
򐂰 Installation, migration, and configuration
򐂰 Administration and maintenance
򐂰 Storage and networking
򐂰 Performance monitoring
򐂰 Security
򐂰 PowerVM advanced features
This publication is written by a group of seven PowerVM experts from different
countries around the world. These experts came together to bring their broad IT
skills, depth of knowledge, and experiences from thousands of installations and
configurations in different IBM client sites.
The team who wrote this book
This book was produced by a team of specialists from around the world, while
working at the IBM International Technical Support Organization, Poughkeepsie
Center.
Adriano de Almeida is an IT Specialist with IBM Integrated Technology Services
Delivery in Brazil. He has worked at IBM for 13 years. His areas of expertise
include IBM AIX®, PowerVM, IBM PowerHA®, and IBM HACMP™. He is a
Certified Advanced Technical Expert on IBM System p®. He has worked
extensively on PowerHA, PowerVM and AIX projects, health checking on
IBM Power Systems™ environments and performing technical project
leadership. He holds a degree in Computing Technology from the Faculdade de
Tecnologia em Processamento de Dados do Litoral (FTPDL). He is also a
coauthor of the Exploiting IBM PowerHA SystemMirror® Enterprise Edition
IBM Redbooks publication.

xvi IBM PowerVM Best Practices
Rafael Antonioli is a System Analyst at Banco do Brasil in Brazil. He has 12
years of experience with Linux and five years of experience in the AIX and
PowerVM field. He holds a Master of Computer Science degree in Parallel and
Distributed Computer Systems from Pontifical Catholic University of Rio Grande
do Sul (PUCRS). His areas of expertise include implementation, support, and
performance analysis of IBM PowerVM, IBM AIX, and IBM PowerHA.
Urban Biel is an IT Specialist in IBM Slovakia. He has been with IBM for six
years. He holds a Master degree in Information Systems and Networking from
Technical University of Kosice, Slovakia. His areas of expertise include Linux,
AIX, PowerVM, PowerHA, IBM GPFS™ and also IBM enterprise disk storage
systems. He has participated in several Redbooks publications.
Sylvain Delabarre is a certified IT Specialist at the Product and Solutions
Support Center in Montpellier, France. He works as an IBM Power Systems
Benchmark Manager. He has been with IBM France since 1988. He has 20 years
of AIX System Administration and Power Systems experience working in service
delivery, AIX, Virtual I/O Server, and HMC support for EMEA.
Bartłomiej Grabowski is an IBM iSeries® Senior Technical Specialist in DHL IT
Services in the Czech Republic. He has seven years of experience with IBM i. He
holds a Bachelor degree in Computer Science from Academy of Computer
Science and Management in Bielsko-Biala. His areas of expertise include IBM i
administration, PowerHA solutions that are based on hardware and software
replication, Power Systems hardware, and IBM i virtualization that is based on
PowerVM. He is an IBM Certified System Administrator. He was a coauthor of
the IBM Active Memory™ Sharing Redbooks publication.
Kristian Milos is an IT Specialist at IBM Australia. Before working at IBM, he
spent seven years working at the largest telecommunications organization in
Australia. He has 10 years of experience working in enterprise environments,
with the past six directly involved with implementing and maintaining AIX,
PowerVM, PowerHA, and Power Systems environments.
Fray L Rodríguez is an IBM Consulting IT Specialist working in Power Systems
Competitive Sales in the United States. He has 12 years of experience in IT and
17 years of experience in customer service. He holds a Bachelor degree in
Software Engineering from the University of Texas at Dallas. Fray also holds 19
professional IBM Certifications, including IBM Expert Certified IT Specialist, IBM
Technical Sales Expert on Power Systems, and IBM Advanced Technical Expert
on Power Systems.
The project team that created this publication was managed by:
Scott Vetter, PMP
IBM Austin

Preface xvii
Thanks to the following people for their contributions to this project:
Aaron Bolding, Thomas Bosworth, Ben Castillo, Shaival J Chokshi,
Gareth Coates, Pedro Alves Coelho, Julie Craft, Rosa Davidson, Ingo Dimmer,
Michael Felt, Rafael Camarda Silva Folco, Chris Gibson, Chris Angel Gonzalez,
Paranthaman Gopikaramanan, Randy Greenberg, Hemantha Gunasinghe,
Margarita Hammond, Jimi Inge, Narutsugu Itoh, Chandrakant Jadhav,
Robert C. Jennings, Anil Kalavakolanu, Bob Kovac, Kiet H Lam,
Dominic Lancaster, Luciano Martins, Augie Mena, Walter Montes,
Guérin Nicolas, Anderson Ferreira Nobre, Rajendra Patel, Viraf Patel,
Thomas Prokop, Xiaohan Qin, Vani Ramagiri, Paisarn Ritthikidjaroenchai,
Björn Rodén, Humberto Roque, Morgan J Rosas, Stephane Saleur,
Susan Schreitmueller, Luiz Eduardo Simeone, Renato Stoffalette,
Naoya Takizawa, Humberto Tadashi Tsubamoto, Morten Vaagmo,
Tomaž Vincek, Richard Wale, Tom Watts, Evelyn Yeung, and Ken Yu.
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 client 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
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

xviii IBM PowerVM Best Practices
򐂰 Send your comments in an email to:
redbooks@us.ibm.com
򐂰 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/IBMRedbooks
򐂰 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

© Copyright IBM Corp. 2012. All rights reserved. 1
Chapter 1. Introduction and planning
IBM PowerVM Best Practices, SG24-8062, provides a summary of best practices
that are implemented by IBM PowerVM experts from around the world. PowerVM
is available for use on all IBM Power Systems, including IBM POWER®
technology-based IBM BladeCenter® servers and IBM PureSystems™.
PowerVM provides industrial-strength virtualization for IBM AIX, IBM i, and Linux
operating systems. If there are any differences in best practices that pertain to
these different operating systems, we point that out, as appropriate.
We suggest that you become familiar with the IBM Redbooks publications:
IBM PowerVM Virtualization Introduction and Configuration, SG24-7940, and
IBM PowerVM Virtualization Managing and Monitoring, SG24-7590. We also
suggest that you have some practical experience with virtualization before you
use the material that is contained in this publication.
The IBM PowerVM Virtualization features are designed to meet all your
enterprise production workload virtualization requirements. The development
and test process for our Virtual I/O Server and virtual client partitions is as
rigorous as it is for dedicated partitions. We also run our benchmarks in virtual
environments. We suggest PowerVM for production, development, testing, or any
other environment in which you would commonly use stand-alone servers or
dedicated partitions.
1

2 IBM PowerVM Best Practices
This paper provides recommendations for the architecture, implementation, and
configuration of PowerVM virtual environments. In areas where there are multiple
configuration options, this publication provides recommendations that are based
on the environment and the experiences of the authors and contributors.
It is possible that new best practices emerge as new hardware technologies
become available.
Although this publication can be read from start to finish, it is written to enable the
selection of individual topics of interest, and to go directly to them. The content is
organized into seven major headings:
򐂰 Chapter 1, “Introduction and planning” on page 1, provides an introduction to
PowerVM and best practices to plan and size your virtual environment.
򐂰 Chapter 2, “Installation, migration, and configuration” on page 15, describes
best practices for the creation and installation of a Virtual I/O Server, and
migration to new versions.
򐂰 Chapter 3, “Administration and maintenance” on page 31, covers best
practices for daily maintenance tasks, backup, recovery, and troubleshooting
of the Virtual I/O Server.
򐂰 Chapter 4, “Networking” on page 49, describes best practices for network
architecture and configuration within the virtual environment.
򐂰 Chapter 5, “Storage” on page 61, covers best practices for storage
architecture and configuration.
򐂰 Chapter 6, “Performance monitoring” on page 93, describes best practices for
monitoring the Virtual I/O Server performance.
򐂰 Chapter 7, “Security and advanced IBM PowerVM features” on page 105,
covers best practices for some of the more advanced PowerVM features.

Chapter 1. Introduction and planning 3
1.1 Keeping track of PowerVM features
IBM is committed to the future of PowerVM, and is constantly enhancing current
features and adding new ones to meet the needs of its clients.
For this reason, our first best practice is to become familiar with these PowerVM
enhancements because they can enrich your environment. Table 1-1 provides a
summary of these features at the time of writing, but you can continue tracking
new features through the following website, along with other useful information
about PowerVM:
http://ibm.com/systems/power/software/virtualization/features.html
Table 1-1 PowerVM features
PowerVM
features
Description
PowerVM
Hypervisor
Supports multiple operating environments on a single system.
Micro-Partitioning Enables up to 10 virtual machines (VMs) per processor core.
Dynamic Logical
Partitioning
Processor, memory, and I/O resources can be dynamically moved
between VMs.
Shared
Processor Pools
Processor resources for a group of VMs can be capped, reducing
software license costs. VMs can use shared (capped or
uncapped) processor resources. Processor resources can
automatically move between VMs based on workload demands.
Shared Storage
Pools
Storage resources for Power Systems servers and a Virtual I/O
Server can be centralized in pools to optimize resource utilization.
Integrated
Virtualization
Manager
Simplifies VM creation and management for entry Power Systems
servers and blades.
Live Partition
Mobility
Live AIX, IBM i, and Linux VMs can be moved between frames,
eliminating planned downtime.
Active Memory
Sharing
Intelligently flows memory from one VM to another for increased
memory utilization.
Active Memory
Deduplication
Reduces memory consumption for Active Memory Sharing (AMS)
configurations by detecting and eliminating duplicate memory
pages.
NPIV Simplifies the management and improves performance of Fibre
Channel SAN environments.

1.2 Virtual I/O Server specifications
The following section defines the range of configuration possibilities, which
includes the minimum number of resources that are needed and the maximum
number of resources allowed.
1.2.1 Virtual I/O Server minimum requirements
Table 1-2 provides the minimum resources that are needed to create a
Virtual I/O Server. Consider these values when you are planning and designing
your Virtual I/O Server.
Table 1-2 Minimum resources that are required for a Virtual I/O Server
Note: Table 1-1 on page 3 shows all the PowerVM features, but these features
are grouped in three edition packages: Express, Standard, and Enterprise, to
best meet your virtualization needs. The following website shows the features
that ship with each PowerVM edition:
http://www.ibm.com/systems/power/software/virtualization/editions.html
Resource Requirement
Hardware Management Console (HMC)
or Integrated Virtualization Manager
(IVM)
The HMC or IVM is required to create the
logical partition and assign resources.
Storage adapter The server logical partition needs at least
one storage adapter.
Physical disk The disk must be at least 30 GB.
Ethernet adapter If you want to route network traffic from
virtual Ethernet adapters to a Shared
Ethernet Adapter (SEA), you need an
Ethernet adapter.
Memory For POWER7 processor-based systems,
at least 768 MB of memory is required.
Processor At least 0.1 processor is required.
PowerVM Editions One of the three PowerVM Editions:
Express, Standard, or Enterprise.

1.2.2 Configuration considerations of the Virtual I/O Server
Consider the following general configuration design points:
򐂰 Virtual Small Computer System Interface (SCSI) supports the following
connection standards for backing devices: Fibre Channel, SCSI,
SCSI Redundant Array of Independent Disks (RAID), IP-based SCSI (iSCSI),
serial-attached SCSI (SAS), Serial Advanced Technology Attachment (SATA),
Universal Serial Bus (USB), and integrated development environment (IDE).
򐂰 The SCSI protocol defines mandatory and optional commands. While
the virtual SCSI supports all of the mandatory commands, not all of the
optional commands are supported.
򐂰 There might be utilization implications when you use virtual SCSI devices.
Because the client/server model is made up of layers of function, the use of
virtual SCSI can use more processor cycles when it is processing I/O
requests.
򐂰 The Virtual I/O Server is a dedicated logical partition, to be used only for
Virtual I/O Server operations. Third-party applications are supported by the
vendors that produced them.
򐂰 If there is a resource shortage, performance degradation might occur. If a
Virtual I/O Server is serving many resources to other logical partitions, ensure
that enough processor power is available. If there is a high workload across
virtual Ethernet adapters and virtual disks, logical partitions might experience
delays in accessing resources.
򐂰 Logical volumes and files that are exported as virtual SCSI disks are always
configured as single path devices on the client logical partition.
򐂰 Logical volumes or files that are exported as virtual SCSI disks that are part of
the root volume group (rootvg) are not persistent if you reinstall the
Virtual I/O Server. However, they are persistent if you update the
Virtual I/O Server to a new service pack. Therefore, before you reinstall the
Virtual I/O Server, ensure that you back up the corresponding virtual disks of
those virtual clients. When you export logical volumes, it is best to export
logical volumes from a volume group other than the root volume group. When
you export files, it is best to create the file storage pools and a virtual media
repository in a parent storage pool other than the root volume group.
򐂰 Only Ethernet adapters can be shared. Other types of network adapters
cannot be shared (at the time of writing).
򐂰 IP forwarding is not supported on the Virtual I/O Server.
򐂰 The maximum number of virtual adapters can be any value in the range
2 - 65536. However, if you set the maximum number of virtual adapters to a
value higher than 1024, the logical partition might fail to activate.

Furthermore, server firmware might require more system memory to manage
the virtual adapters.
򐂰 The Virtual I/O Server supports client logical partitions that run IBM AIX 5.3 or
later, IBM i 6.1 or later, and Linux. The minimum version of Linux that is
supported might be different depending on the hardware model. Check the
hardware manual for the model you need to confirm the minimum version you
can use.
1.2.3 Logical Volume Manager limits in a Virtual I/O Server
If you plan to use logical volumes as backing devices, consider the following
limitations for the Logical Volume Manager (LVM) components that are used in a
Virtual I/O Server. Table 1-3 shows these limitations.
Table 1-3 Limitations for storage management
1.3 Planning your Virtual I/O Server environment
PowerVM offers great flexibility when you configure a Virtual I/O Server, and with
that flexibility, there are a number of choices you can make. This flexibility means
that there is more than one correct way to design and implement your virtual
environment.
Based on this flexibility, it is important for you to illustrate a picture of the
environment by listing all the possible correct options for implementing each
scenario. Also include the consequences of each choice. You can make an
informed decision by using this practice. Reaching best practices for your
environment are a matter of choosing the option that best fits the workloads you
plan to run. What is considered a best practice in a lab, might not be a best
practice for an enterprise banking server.
Section 1.3 describes the choices that you encounter to help you plan your
environment.
Category Limit
Volume groups 4096 per system
Physical volumes 1024 per volume group
Physical partitions 1024 per volume group
Logical volumes 1024 per volume group
Logical partitions No limit

1.3.1 System Planning Tool
The System Planning Tool (SPT) is a browser-based application that helps you
plan, design, and validate system configurations. It is a best practice to use this
tool to create your initial setup and to distribute the hardware resources among
the dedicated and virtual partitions. This tool has many functions and many
built-in checks to assist you with your system planning.
It is also a good practice to use this tool to plan your partitions. This practice is
beneficial because you can deploy the system plans that are generated with the
SPT. You can use these plans to create the partition profiles on your Hardware
Management Console (HMC) or the Integrated Virtualization Manager (IVM).
Another benefit of using the SPT is that it can help you calculate how much of the
system memory that the Power Hypervisor uses depending on your
configuration. This way, you can plan according to how much remains available
for use in your partitions. Finally, it is also a best practice to print the SPT
validation report to ensure that you are meeting all the system requirements. The
SPT is at this website:
http://ibm.com/systems/support/tools/systemplanningtool
1.3.2 Document as you go
Do not wait until the end of the project to document your environment. Whether
you prefer to use a spreadsheet, a database, or any other type of software, write
down all the requirements you have for your environment, your objectives, your
choices, and your solutions. Then update it as you plan, design, and build your
virtual environment.
1.3.3 Hardware planning
If you are involved in the decision making process of purchasing the
IBM Power Systems server that is used to build your virtual environment, this is a
good time to start filling out your documentation spreadsheet. List all the
functional and nonfunctional requirements that you have in your environment to
ensure that the system is configured with all the hardware components, software,
and PowerVM features that you need.
Think of accessibility, availability, scalability, performance, security, compliance,
disaster recovery, and so on, and everything you need to do to meet the
requirements of these categories. For example, you might decide that you want
to use N-Port ID Virtualization (NPIV). Then you need to ensure that you
purchase 8-Gb adapters, which are a hardware requirement. Another

requirement might be that you are able to move your virtual clients from one
physical server to another without any downtime. This feature is called
Live Partition Mobility, and is included in the IBM PowerVM Enterprise Edition.
Therefore, you need to include this requirement to ensure that Enterprise Edition
is included in the frame, as opposed to the Standard or Express Edition.
Power Systems servers offer outstanding flexibility as to the number of adapters,
cores, memory, and other components and features that you can configure in
each server. Therefore, discuss your requirements with your IBM technical sales
team, and they can configure a server that meets your needs.
Other hardware features
When you are planning your hardware configuration, remember that there are
other features that are available in Power Systems servers that are not part of
PowerVM. However, these other features can be useful in your virtual
environment.
For example, Active Memory Expansion (AME) is a feature that can be activated
with any Power Systems server and can be used to increase memory capacity
up to 100%. This feature can be used with the PowerVM feature AMS, or
independently.
There are other features, such as Active Memory Mirroring (AMM), that are only
available on Enterprise class servers. AMM can provide mirroring of the
Power Hypervisor code among different dual inline memory modules (DIMMs).
This feature enhances the availability of a server in case a DIMM failure occurs in
one of the DIMMs that hold the Hypervisor code.
As a best practice, get a list of all the optional Reliability, Availability, and
Serviceability (RAS) features that are available in the Power Systems server that
you purchase. Plan to implement the features.
1.3.4 Sizing your Virtual I/O Server
Sizing a Virtual I/O Server can be a challenging task for many IT Architects and
IT Specialists. In fact, it can be argued that the only way to truly size any server is
to run it with a real workload, monitor it, and tune it accordingly.
The following guidelines are intended to help you get started with the sizing of a
Virtual I/O Server environment. The Virtual I/O Servers offer great flexibility, such
as the option to run different virtual clients with different operating systems and
thousands of independent software vendor (ISV) applications. Therefore, it is
important that you monitor the performance of the Virtual I/O Server and adjust
its resources, if needed.

As a best practice, have at least two Ethernet adapters and two Fibre Channel
adapters in your Virtual I/O Server.
For small Power Systems servers, with eight cores or less, you can start with 0.5
of entitled processor capacity and 2 GB of memory.
For more powerful Power Systems servers, we use the example of a
Virtual I/O Server that uses two 1 Gb Ethernet adapters and two 4 Gb
Fibre Channel adapters, supporting around 20 virtual clients. For this scenario,
use one core for the entitled processor capacity and 4 GB of entitled memory
when you run AIX or Linux. If the virtual clients are running IBM i and are using
virtual SCSI, use 1.5 for entitled processor capacity and 4 GB of entitled memory.
If the IBM i virtual client is using NPIV, one core of entitled capacity is a good
starting point.
High speed adapters, such as 10 Gb Ethernet and 8-Gb Fibre Channel, require
more memory for buffering. For those environments, use 6 GB of memory in
each scenario, regardless of the operating system that is running on the virtual
I/O client. This is also especially important to consider if you plan to use NPIV. It
is important because the Virtual I/O Server conducts work that is similar to a
virtual storage area network (SAN) switch that is passing packets back and forth
between your SAN and the virtual I/O clients. Therefore, the Virtual I/O Server
requires more memory for each virtual Fibre Channel adapter that is created in a
virtual client.
Estimating an exact amount of memory that is needed per adapter is difficult
without using tools to look at the specific workloads. It is difficult because the
requirements of the adapters vary depending on their technologies, Peripheral
Component Interconnect-X (PCI-X) and PCI Express (PCIe), and their
configurations. A rule of thumb for estimating the amount of memory that is
needed by each adapter, is to add 512 MB for each physical high speed Ethernet
adapter in the Virtual I/O Server. This number also includes the memory that is
needed for virtual Ethernet adapters in the virtual clients. Also, add 140 MB for
each virtual Fibre Channel adapter in the client.
Based on this example, for a Virtual I/O Server that has two 10 Gb Ethernet
adapters and two 8 Gb Fibre Channel adapters, supports 20 virtual clients, and
assumes that each client has one Virtual Fibre Channel adapter, we have the
following scenario:
2 GB for base workload + 1 GB (512 MG x 2) for Ethernet adapters + 2.8 GB
(20 x 140 MB) for Virtual Fibre Channel adapters = 5.8 GB. Therefore, our
general recommendation is to have 6 GB of memory.

1.3.5 IBM Systems Workload Estimator
If you want a more precise method than what is described in the guidelines in
section 1.3.4, “Sizing your Virtual I/O Server” on page 8, use the IBM Systems
Workload Estimator (WLE) tool. WLE is a web-based sizing tool, which allows
you to size Power Systems servers, IBM System X, and IBM PureFlex™
Systems. For information about this tool and to obtain a download version, see
this website:
http://ibm.com/systems/support/tools/estimator
For the Virtual I/O Server, the WLE sizing is based on metrics that drive activity
through it. When sizing, consider metrics such as the number and size of the disk
I/Os, and the number and size of the network I/Os.
1.3.6 Shared or dedicated resources
During the planning session, decide whether your Virtual I/O Server and virtual
client partitions are to use dedicated resources, processors and memory, or
shared resources.
If you use shared processors and shared memory, you get better utilization, and
you can often add new workloads or increase existing ones without more
resources. This sharing translates into lower core-based software licensing
costs, therefore, saving money.
Some people prefer to have a dedicated amount of processing units and memory
for a production partition. If you choose to use dedicated resources in each
partition, your sizing of the environment needs to consider peak times during the
day. Peak times include backup time or peak seasons, such as back to school
and holidays, which might require you to have unused resources during the rest
of the year.
Core speeds vary: Remember that core speeds vary from system to system,
and IBM is constantly improving the speed and efficiency of the IBM POWER
processor cores and memory. Therefore, the referenced guidelines are, again,
just a starting point. When you create your Virtual I/O Server environment,
including all the virtual clients, be sure to test and monitor it to ensure that the
assigned resources are appropriate to handle the load on the Virtual I/O
Server.

For planning purposes, whichever option you choose, it is a good practice to
understand how these options affect your environment. Also consider the amount
of resources that you need to satisfy your workloads.
1.3.7 Single, dual, or multiple Virtual I/O Servers
IBM Power Systems servers are resilient. Therefore, a single Virtual I/O Server is
sufficient to support many small environments, but redundancy is always a plus.
Therefore, having dual or multiple Virtual I/O Servers is a best practice for
production environments.
The key here is to understand what your availability requirements are to decide
what you need. Do you need redundancy at the hardware level or are you
implementing redundancy at the application level across multiple logical
partitions or physical frames? Do you have any Service Level Agreements
(SLAs) that require your servers to be up 99.999% of the time? Are you able to
schedule maintenance once or twice a year to do updates? Are there any
penalties for planned or unplanned downtime?
These are the type of questions you need to answer to plan whether you get a
single or dual Virtual I/O Server, or even multiple Virtual I/O Servers. For
instance, some environments have heavy amounts of network and storage I/O
traffic. Therefore, some administrators prefer to have four Virtual I/O Servers:
Two to handle the network traffic and two to handle the I/O from the SAN.
Besides availability, the idea behind this type of configuration is to improve
performance.
Mission critical workloads have different requirements to development or to test
partitions. Therefore, as a best practice, plan for the right number of resources
that are needed by one or multiple Virtual I/O Servers, depending on the
workloads you plan to run.
1.3.8 Network and storage components
Now you can also plan on the number and type of network and storage
components you need in your Virtual I/O Server. For instance, if you plan to have
dual Virtual I/O Servers, you might ask yourself how many Ethernet adapters you
need and at what speed? Will Etherchannel be used? Will you use shared
Ethernet adapter failover from one Virtual I/O Server to the other? Are multiple
Virtual LANs (VLANs) needed to separate the network traffic for security
purposes?

You need to think the same way about your storage component needs. Will NPIV
or virtual SCSI be used? If you use virtual SCSI, will Shared Storage Pools be
used? Will the boot occur from internal disk drives or SAN?
Think about all these components in advance to determine the best way to
configure and implement them. Remember to document your choices and the
reasons behind them. Table 1-4 contains some of the network and storage
components you might consider.
Table 1-4 Network and storage components
1.3.9 Slot numbering and naming conventions
The use of a slot numbering system and naming convention, and sticking to it,
can save a great deal of time when troubleshooting issues. This system is also
helpful when new employees are learning how everything is mapped from the
Virtual I/O Server to the virtual clients. Therefore, plan to create a numbering
system that suits your needs, and follow it.
With any of the slot numbering options, it is best if you map the same slot number
on the Virtual I/O Server with the same slot number in the virtual client.
Assuming you use that approach, you can use the following ideas to help create
a numbering method.
One easy option for a dual Virtual I/O Server environment is to use even
numbers for one server and odd numbers for the other. In this case, one
Virtual I/O Server has 11, 13, 15, and so on, while the second Virtual I/O Server
has 12, 14, and so on.
Network components Storage components
Physical Ethernet adapters Physical Fibre Channel adapters
Virtual Ethernet adapters Virtual Fibre Channel adapters
Shared Ethernet adapters Virtual SCSI adapters
Backup virtual Ethernet adapters SAN volumes
Virtual LANs (VLANs) Multi-path I/O
Etherchannel or Link Aggregation devices Mirroring internal drives
Storage pools (volume groups)
Shared Storage Pools

If you also want to identify the virtual client through the slot number without
having to look at the mapping, you can add its partition ID to the odd or even
number. The following example shows this process:
<virtual_client_ID><(virtual_server_odd or Virtual_server_even)>
For example, in a system with dual Virtual I/O Servers with IDs 1 and 2, and a
virtual client with ID 3, we use odd numbers for Virtual I/O Server 1, as in 31, 33,
35, and so on. The second Virtual I/O Server uses even numbers, as in 32, 34,
36, and so on.
This system might not be appropriate though if you have many clients, because
the slot number is too high. Also, consider that when you use odd and even
numbers, it is difficult to maintain a system if you are frequently using Live
Partition Mobility (LPM) to move virtual clients from one physical system to
another.
A different method for slot numbering is to use adapter ID = XY, where the
following values apply:
X= adapter type, using 1 for Ethernet, 2 for Fibre Channel, and 3 for SCSI
Y= next available slot number
With this system, your virtual Ethernet adapters are 11 - 19. Your Virtual Fibre
Channel is 20 - 29, and your virtual SCSI is 30 - 39.
For naming convention methods for the virtual adapters in a Virtual I/O Server,
use a name that is descriptive of the virtual client that uses this virtual adapter.
1.3.10 Operating systems on virtual client partitions
A Virtual I/O Server can host client partitions that run AIX, IBM i, or Linux.
Therefore, it is important to consider the operating system that is running in the
Use low slot numbers: It is a best practice to use low slot numbers and
define only the number of virtual adapters that you use. This practice is
preferred because the Power Hypervisor uses a small amount of memory for
each virtual device that you define.
LPM: When you use LPM, if a target system does not have the slot available
that you want to use, it asks if it can use a new slot number. If slot numbers
change, it is important that you keep current documentation of your
environment.

virtual client when you plan your Virtual I/O Server configuration. For instance,
some tuning options might be more beneficial on one virtual client operating
system than another. Throughout the book, we mention such differences as they
arise.
You can also use the Fix Level Recommendation Tool (FLRT) to ensure that the
version of the Virtual I/O Server is compatible with other components. The
version needs to be compatible with the firmware on the server, the version of the
operating system on the virtual client, the HMC or IVM, and other IBM software
that might be in your environment. To use the FLRT, see this website:
http://www14.software.ibm.com/webapp/set2/flrt/home

Chapter 2. Installation, migration, and
configuration
This chapter describes the best practices of installing, migrating, and configuring
your Virtual I/O Server. We also address the creation of the Virtual I/O Server
profile.
2

2.1 Creating a Virtual I/O Server profile
Section 2.1 describes the best practices to create a Virtual I/O Server profile. To
create a profile, you have the options of using the Hardware Management
Console (HMC), the Integrated Virtualization Management (IVM), and the
System Planning Tool (SPT). For ease of deployment of multiple virtual
partitions, we suggest you use the SPT to create and deploy your partitions.
The following subsections are best practices to create the Virtual I/O Server and
virtual client profiles, regardless of the management tool that is used to create
them.
The complete set of instructions for creating a Virtual I/O Server is described in
the Redbooks publication, PowerVM Virtualization Introduction and Configuration
SG24-7940.
2.1.1 Processing mode - shared or dedicated
The Power Hypervisor manages the allocation of processor resources to
partitions within a system. Each partition is allocated processor resources
according to its profile. The processor allocation can be dedicated to that
partition or shared with other partitions within a processor pool.
It is a best practice to use the shared processing mode in a Virtual I/O Server
because of the changes in the virtual client workloads. Figure 2-1 on page 17
shows the selection of the shared processing mode.

Chapter 2. Installation, migration, and configuration 17
Figure 2-1 Select processor type
2.1.2 Processing settings
When you are defining the values for physical and virtual processor units, you
need to assign a minimum, a wanted, and a maximum value.
Physical processing units
For minimum processing units, the best practice is to set this value to 0.1. This
number is the lowest configured unit value that can be assigned to a partition,
and which allows the Virtual I/O Server to boot.

For wanted processing units, follow the sizing considerations that are described
in 1.3.4, “Sizing your Virtual I/O Server” on page 8. Set this value to a number
that meets the needs of the estimated workload.
For maximum processing units, round up the wanted processing units value to
the next whole number, and add 50%. For example, if the wanted value is 1.2, the
maximum value is 3.
It is important to allow room between the wanted and the maximum processing
units. This suggestion is because you can only increase the wanted value
dynamically, through a dynamic LPAR operation, up to the maximum value. At
the same time, it is important not to set the maximum value too high because the
Power Hypervisor uses more memory the higher it is.
Virtual processing units
It is important to remember that there is Hypervisor CPU usage that is
associated with the number of online virtual processors. Therefore, carefully
consider the capacity requirements of the Virtual I/O Server before you make a
decision.
Uncapped partitions require virtual processors so they can exceed their
entitlement. However, do not make these values too high because of the
Hypervisor CPU usage and because if there is not enough processing capacity, it
might cause excessive context switching. This switching might, in turn, cause
performance degradation. Below are some general best practices to follow:
򐂰 For minimum virtual processors, round up the minimum processing units
value to the next whole number. For example, if the minimum processing units
are 0.1, the minimum virtual processors value is 1.
򐂰 For wanted virtual processors, add 50% to the wanted processing units value,
and round it up to a whole number. For example, if the wanted number of
processing units is set to 1, the wanted number of virtual processors is 2.
򐂰 For maximum virtual processors, add 50% to the wanted virtual processors.
For example, if the number of wanted virtual processors is 2, the maximum
value is 3.
Capped or uncapped
The sharing mode of processor cores can be set to capped or uncapped.
Capped partitions have a preset amount of maximum processing unit
entitlement. However, partitions that are configured with uncapped processor
resources are able to use all of their allocation, plus any unused processing units
in a shared processor pool.

The load on the Virtual I/O Server varies depending on the demands of the
virtual clients. Therefore, you might see spikes in processor or memory usage
throughout the day. To address these changes in workload, and to achieve better
utilization, use shared and uncapped processors. Therefore, uncapping can
provide a significant benefit to partitions that have spikes in utilization.
Weight
When you choose the uncapped partition option, you also need to choose a
weight value. For the Virtual I/O Server, best practice is to configure a weight
value higher than the virtual clients. The maximum configured value is 255. The
Virtual I/O Server must have priority to the processor resources in the frame.
Table 2-1 shows an example of distributing weight among different types of
environments, depending on their importance.
Table 2-1 Processor weighting example
Figure 2-2 on page 20 shows an example of a best practice configuration of
Processing Settings.
Value Usage
255 Virtual I/O Server
200 Production
100 Pre-production
50 Development
25 Test

Figure 2-2 Processing Settings best practices
2.1.3 Memory settings
When you define the values for physical memory, you need to assign a minimum,
a wanted, and a maximum value.
For wanted memory, follow the sizing suggestions that are described in 1.3.4,
“Sizing your Virtual I/O Server” on page 8. Use these sizes to meet the workload
demands of your Virtual I/O Server. From this value, we can determine the
minimum and maximum memory values.
For minimum memory, use 50% of the wanted memory value. For example, if the
wanted value is 4 GB, the minimum value is 2 GB.

For maximum memory, add 50% to the wanted memory value. For example, if
the wanted value is 4 GB, the maximum value is 6 GB.
Similarly to processing units, it is important to allow room between the wanted
and the maximum memory values. This consideration is because you can only
increase the wanted value dynamically, through a dynamic LPAR operation, up to
the maximum value.
The maximum memory value is also the number that is used when you calculate
the amount of memory that is needed for the page tables to support this partition.
For this reason, it is not advantageous to set this maximum setting to an
unreasonably high amount. This recommendation is because it would waste
memory by setting memory aside for page tables that the partition does not
need.
2.1.4 Physical I/O adapters
It is a best practice to configure all the physical I/O adapters to the wanted
memory value. Having them set to the required value, prevents dynamic logical
partition (DLPAR) operations from working. The aim is to allow adapter changes
to happen dynamically with no downtime required on the partition. Also, an
adapter which is set to required that is in a failed or removed state, prevents the
LPAR from activating.
2.1.5 Virtual I/O adapters
It is a good practice to have a good planning naming process in the
Virtual I/O Server because managing a large environment can be complex.
Maximum virtual adapters
Before you set the maximum virtual adapter limit, you need to consider the
number of slots that your virtualized environment uses. The larger you set the
maximum virtual adapter value, the more memory the Hypervisor needs to
reserve to manage them. Figure 2-3 on page 22 shows the process for setting
the maximum number of virtual adapters.
Amount of memory: The SPT can help you estimate the amount of memory
that is required by the Power Hypervisor depending on your IBM Power
Systems server and its configuration. Also, the HMC can show you the amount
of memory that is available for partition use under the frame properties, and
the amount that is reserved for the Hypervisor.

Figure 2-3 Setting the maximum number of virtual adapters
Virtual adapter slot numbers
When you configure the virtual SCSI and virtual Fibre Channel adapters, it is a
best practice to keep the virtual adapter IDs on the client and the server, the
same. This practice can be useful when you manage and trace your storage
configuration.
In Figure 2-4 on page 23, we show an example of the virtual adapter relationship
between two Virtual I/O Servers and one virtual I/O client.

Figure 2-4 Partition profile properties for source and target virtual adapters

2.1.6 Deploying a Virtual I/O Server with the System Planning Tool
After you configure your logical partitions in the System Planning Tool, you can
save your system plan file (.sysplan) and import it into an HMC or IVM. You can
then deploy the system plan to one or more frames. System plan deployment
delivers the following benefits to you:
򐂰 You can use a system plan to partition a frame and deploy partitions without
having to re-create the partition profiles. This plan saves time and reduces the
possibility of errors.
򐂰 You can easily review the partition configurations within the system plan, as
necessary.
򐂰 You can deploy multiple identical systems, almost as easily as a single
system.
򐂰 You can archive the system plan as a permanent electronic record of the
systems that you create.
2.2 Virtual I/O Server installation
Described here are the best practices to install the Virtual I/O Server. In general,
we suggest that you use the Network Installation Manager (NIM) in your
environment for installations, updates, and fixes. This suggestion applies to the
Virtual I/O Server, too. If you are going to install more than one Virtual I/O Server,
the NIM helps you significantly. Following is a basic installation procedure:
򐂰 Install the Virtual I/O Server by using the NIM. For details on how to create
NIM resources for the Virtual I/O Server installation, see this website:
http://www.ibm.com/support/docview.wss?uid=isg3T1011386#4
򐂰 Do not perform any virtualization configurations, and do not mirror the root
volume group (rootvg).
򐂰 Install all necessary device drivers, apply the updates and fixes, and then
reboot.
򐂰 Perform a backup of the Virtual I/O Server to create a golden image.
򐂰 Use the golden image to deploy all the other Virtual I/O Servers.
Live Partition Mobility: If you are planning to implement Live Partition
Mobility, set all virtual adapters to wanted. Required virtual I/O adapters
prevent Live Partition Mobility operations.

If you do not have a NIM in your environment, install it from a DVD, and then
apply the basic installation procedures. You can use the alt_root_vg command
to deploy other Virtual I/O Servers.
If you boot the Virtual I/O Server from the local Small Computer System Interface
(SCSI) disks, remember to mirror the rootvg by using the mirrorios command.
And, set the boot list with both the SCSI disks by using the bootlist command.
2.3 Updating fix packs, service packs, and interim fixes
In addition to fix packs, the Virtual I/O Server might also have service packs,
depending on the number of changes that are needed. Virtual I/O Server Service
Packs consist of critical changes that are found between fix pack releases.
2.3.1 Virtual I/O Server service strategy
To ensure the reliability, availability, and serviceability of the Virtual I/O Server,
we suggest that you update it to the latest available release, fix pack, or service
pack. The latest level contains all of the cumulative fixes for the Virtual I/O
Server. Interim fixes are only provided at the latest level of the Virtual I/O Server.
You need to upgrade to the latest service level or fix pack to install an interim fix.
Updating strategy
The following suggestions assist you in determining the best update strategy for
your enterprise:
򐂰 Set a date, every six months, to review your current firmware and software
patch levels.
򐂰 Verify the suggested code levels by using the Fix Level Recommendation Tool
(FLRT) on the IBM Support Site:
http://www.software.ibm.com/webapp/set2/flrt/home
alt_root_vg command: If you boot from a cloned disk that is made by the
alt_root_vg command, we suggest you remove obsolete devices that are in
the defined state. Also, you might need to reconfigure the Reliable Scalable
Cluster Technology (RSCT) subsystem by using the recfgct command. For
more information about the recfgct command, see this website:
http://www.ibm.com/support/docview.wss?uid=swg21295662

򐂰 Check for the Virtual I/O Server release lifecycles to plan your next upgrade.
See this website:
http://www.ibm.com/software/support/lifecycleapp/PLCSearch.wss?q=%28
Virtual+I%2FO+Server%29+or++%28PowerVM%29&scope=&ibm-view-btn.x=3&ib
m-view-btn.y=9&sort=S
򐂰 There are several options for downloading and installing a Virtual I/O Server
update, such as downloading ISO images, packages, or installing from optical
media. To check the latest release and instructions for Virtual I/O Server fix
updates, see IBM Fix Central at this website:
http://www.ibm.com/support/fixcentral/
򐂰 Ensure that a regular maintenance window is available to conduct firmware
updates and patching. Once a year is the suggested time frame to conduct
the updates.
򐂰 When you do system firmware updates from one major release to another,
always update the HMC to the latest available version first, along with any
mandatory HMC patches. Then, do the firmware updates. If the operating
system is being updated as well, update the operating system first, then the
HMC code, and lastly the system firmware.
򐂰 In a dual HMC configuration always update both HMCs in a single
maintenance window. Or, disconnect one HMC until it is updated to the same
level as the other HMC.
2.3.2 Approved third-party applications in the Virtual I/O Server
It is a best practice to check third-party application lists that are allowed to be
installed in the Virtual I/O Server. Installing applications that are not listed as a
valid application on the IBM PartnerWorld® website, invalidate the
IBM Virtual I/O Server support. See this website:
http://dbluedst.pok.ibm.com/DCF/isg/isgintra.nsf/all/T1010620
IBM support teams do not support these applications. For problems with these
applications, contact your appropriate vendor support.
ISO images: Do not use utilities to extract ISO images of Virtual I/O fix packs
or service packs to a local directory or NFS mount point. Burn these images to
media. If you need fix packs or service packs on a local directory, download
them as a package.

2.3.3 Applying fix packs, service packs, and interim fixes
Before you begin installing your applications, confirm that the following
statements are true:
򐂰 The managed frame of the HMC is attached to the system and both the
managed frame and HMC are working properly.
򐂰 When you update your Virtual I/O Server, do it in a console window (instead
of a telnet) or run Secure Shell (SSH) to the Virtual I/O Server. For example,
on AIX, run the script file name to keep a log, run SSH to connect to the HMC,
then run vtmenu.
򐂰 Always check the readme instructions before you apply your updates.
򐂰 If you do not have a recent backup, run the backupios command and keep the
output file in a safe location.
You can use the alt_root_vg command to clone your rootvg to an alternate
disk and update the Virtual I/O Server to the next fix pack level; see
Example 2-1. This command can be done without taking the system down for
an extended period and mitigating an outage risk.
This cloning can be done by creating a copy of the current rootvg on an
alternate disk and simultaneously applying the fix pack updates. If needed,
the bootlist command can be run after the new disk is booted. The bootlist
can be changed to boot back to the older level of the operating system in the
event of an issue. Example 2-1 shows the alt_root_vg command.
Example 2-1 Using the alt_root_vg command
$ alt_root_vg -target hdisk1 -bundle update_all -location /mnt
Before you shut down the Virtual I/O Server, follow these steps:
򐂰 On a single Virtual I/O Server environment, shut down the virtual I/O clients
that are connected to the Virtual I/O Server. Or disable any virtual resource
that is in use.
򐂰 In a dual Virtual I/O Server environment, check that the alternate Virtual I/O
Server is up and running and is serving I/O to the client. (Shared Ethernet
Adapter (SEA) failover, virtual SCSI mapping, virtual Fibre Channel mapping,
and so on).
– If there is Logical Volume Manager mirroring on clients, check that both
disks of any mirrored volume group are available in the system and the
mirroring is properly synchronized.

– If there is a SEA failover, check the configuration of the priority, and that
the backup is active on the second Virtual I/O Server.
– If there is a Network Interface Backup (NIB), check that the Etherchannel
is configured properly on the virtual I/O clients.
2.4 Virtual I/O Server migration
The purpose of migration is to change the Virtual I/O Server software to a new
version while you preserve your configuration files. For example, the migration
process saves configuration files, installs the new software on your system, and
then restores your configuration.
2.4.1 Options to migrate the Virtual I/O Server
You can migrate the Virtual I/O Server in the following ways:
򐂰 Migrate the Virtual I/O Server using the NIM.
򐂰 Migrate the Virtual I/O Server from the HMC.
򐂰 Migrate the Virtual I/O Server from optical media.
The best practice is to migrate the Virtual I/O Server using the NIM.
Detailed information about each of the migration options is on the
IBM Power Systems Hardware Information Center. See this website:
http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp
2.4.2 Virtual I/O Server migration considerations
Before you start, confirm that the following statements are true:
򐂰 The managed system of the HMC is attached to the system and both the
managed system and the HMC are working properly.
򐂰 If you do not have a recent backup, run the backupios command and keep the
output file in a safe location.
Cloning: You can use the alt_root_vg command to clone your Virtual I/O
Server. Cloning the running rootvg allows for the creation of a backup copy
of the root volume group. This copy can be used as a backup in case the
rootvg fails.

򐂰 Note the location of the rootvg disks.
򐂰 Confirm that you have the Migration DVD for the Virtual I/O Server instead of
the Installation DVD for the Virtual I/O Server.
Before you shut down the Virtual I/O Server that you want to migrate, we suggest
you check the following scenarios:
򐂰 In a single Virtual I/O Server configuration, during the migration, the client
partition needs to be shut down. When the migration is complete, and the
Virtual I/O Server is restarted, the client partition can be brought up without
any further configuration.
򐂰 In a dual Virtual I/O Server environment, you can migrate one Virtual I/O
Server at a time to avoid any interruption of service to the clients:
– If there is Logical Volume Manager mirroring on clients, check that both
disks of any mirrored volume group are available in the system and the
mirroring is properly synchronized.
– If there is Shared Ethernet Adapter (SEA) failover, check the configuration
of the priority, and that the backup is active on the second Virtual I/O
Server.
– If there is a Network Interface Backup (NIB), check that the Etherchannel
is configured properly on the virtual I/O clients.
Installation media: The media for the migration and for the installation are
different. Using the installation media overwrites your current Virtual I/O
Server configuration.
Folding: Processor folding currently is not supported for Virtual I/O Server
partitions. If folding is enabled on your Virtual I/O Server and migration media
is used to move from Virtual I/O Server 1.5 to 2.1.0.13 FP 23, or later,
processor folding remains enabled. Upgrading by using migration media does
not change the processor folding state. If you installed Virtual I/O Server
2.1.3.0, or later, and did not change the folding policy, then folding is disabled.

2.4.3 Multipathing software
When you plan to migrate your Virtual I/O Server, you must check that the
multipathing driver software is supported on the new version.
For example, when you migrate the Virtual I/O Server from version 1.x to 2.x, you
must migrate the Subsystem Device Driver (SDD) or the SSD Path Control
Module (SDDPCM) software. Detailed information is on the following website:
http://www.ibm.com/support/docview.wss?uid=ssg1S7002686
For third-party multipathing software, consult the vendor documentation to
ensure compatibility with the Virtual I/O Server version.

Chapter 3. Administration and
maintenance
This chapter describes best practices for general administration topics of the
virtual I/O environment, with a focus on backing up and restoring the
Virtual I/O Server. We also address dynamic logical partition operations, network
installation manager resilience, and the virtual media repository.
Virtual I/O clients depend on the Virtual I/O Server for services. Therefore, it is
critical that the entire virtual I/O environment is backed up, and that system
restore and startup procedures include the Virtual I/O Server.
3

3.1 Backing up and restoring the Virtual I/O Server
The Virtual I/O Server, as is the case of any other critical server within an IT
environment, needs to be backed up as part of a data recovery program for an
enterprise. Section 3.1 sets out a strategy to coordinate the backup of the Virtual
I/O Server into your current backup strategy.
This section also describes a complete solution that can be used to restore the
Virtual I/O Server to another system, independent of the system type or model
number. If you want to conduct a backup of just the Virtual I/O Server, only a
subset of Section 3.1 is required.
3.1.1 When to back up the Virtual I/O Server
A complete disaster recovery strategy for the Virtual I/O Server includes backing
up the following components that make up a virtualized environment. A change
to any one of the following devices requires a new backup:
򐂰 External device configuration, for example, storage area network (SAN) and
storage subsystem configuration.
򐂰 Memory, processor, virtual, and physical devices.
򐂰 The Virtual I/O Server operating system.
򐂰 User-defined virtual devices that couple the virtual and physical
environments. This configuration can be considered virtual device mappings,
or metadata.
You might notice that there is no mention of the operating system, installed
applications, or application data of the virtual clients that are listed. There are no
references because the Virtual I/O Server manages only the devices and the
linking of these devices along with the Virtual I/O Server operating system itself.
The virtual clients that are running IBM AIX, IBM i, or Linux need to have a
backup strategy that is independently defined as part of your existing server
backup strategy.
For example, if you have an AIX virtual client that is made up of virtual disks and
a virtual network, you would still have a mksysb and savevg. Or you have an
equivalent strategy in place to back up the system. This backup strategy can rely
on the virtual infrastructure. For example, this strategy includes backing up to an
IBM Tivoli® Storage Manager (TSM) server over a virtual network interface
through a physical Shared Ethernet Adapter (SEA).

Chapter 3. Administration and maintenance 33
3.1.2 Virtual I/O Server backup strategy
The following section defines best practices to perform backup operations.
External device configuration
If a natural or man-made disaster destroys a complete site, planning for that
occurrence can be included into the end-to-end backup strategy. This planning is
part of your current disaster recovery (DR) strategy. The backup strategy
depends on the hardware specifics of the storage, networking equipment, and
SAN devices, to name a few. Examples of the type of information that is needed
to record, include the virtual local area network (VLAN) or logical unit number
(LUN) information from a storage subsystem.
This recording information is beyond the scope of this document. However, we
mention it here to make you aware that a complete DR solution for a physical or
virtual server environment has a dependency on this information. The method to
collect and record the information depends not only on the vendor and model of
the infrastructure systems at the primary site, but also what is present at the DR
site.
Resources that are defined on the HMC and IVM
The definition of the Virtual I/O Server logical partition on the Hardware
Management Console (HMC) and Integrated Virtualization Manager (IVM)
includes, for example, how much processor and memory allocation is needed.
Also consider what physical adapters are used. In addition to this consideration,
you have the virtual device configuration (for example, virtual Ethernet adapters
and what VLAN IDs to which they belong) that needs to be captured. The backup
and restore of this data is beyond the scope of this document. For more
information, see these IBM Redbooks publications:
򐂰 Hardware Management Console V7 Handbook, SG24-7491
http://www.redbooks.ibm.com/abstracts/sg247491.html?Open
򐂰 Integrated Virtualization Manager on IBM System p5, REDP-4061
http://www.redbooks.ibm.com/abstracts/redp4061.html?Open
Consider that you might need to rebuild selected HMC and IVM profiles from
scratch on new hardware, especially if you are planning for disaster recovery. In
this case, it is important to have detailed documentation of the configuration,
such as how many Ethernet cards and Fibre adapters are needed. Using the
System Planning Tool (SPT) to create system plans can help you record such
information.

Backing up the Virtual I/O Server
The Virtual I/O Server operating system consists of the base code, fix packs,
custom device drivers to support disk subsystems, and user-defined
customization. An example of user-defined customization can be as simple as
the changing of the Message of the Day or the security settings.
These settings, after an initial setup, will probably not change aside from the
application of fix packs. Therefore, a sensible backup strategy for the
Virtual I/O Server is needed before you apply the fix packs or make configuration
changes. Section 3.1.3, “Backing up user-defined virtual devices” on page 35
covers this strategy. However, having a daily backup of the virtual and logical
configuration can save time in restoring from configuration errors.
There are three ways to back up the Virtual I/O Server:
򐂰 Back up to tape.
򐂰 Back up to DVD-RAM.
򐂰 Back up to a remote file.
Backing up the Virtual I/O Server to a remote file is the most common and best
practices method.
backupios command
The backupios command performs a backup of the Virtual I/O Server to a tape
device, an optical device, or a file system (local file system or a remotely
mounted Network File System (NFS)).
Backing up to a remote file system satisfies having the Virtual I/O Server backup
at a remote location. It also allows the backup to be restored from either a
Network Installation Management (NIM) server or the HMC. In Example 3-1, a
Virtual I/O Server backup is done to an NFS mount which resides on a NIM
server.
Example 3-1 Virtual I/O Server backup to a remote file system
$ mount nim:/export/vios_backup /mnt
$ backupios -file /mnt -nomedialib
Backup in progress. This command can take a considerable amount of
time to complete, please be patient...
-nomedialib flag: The -nomedialib flag excludes the contents of the virtual
media repository from the backup. Unless explicitly required, excluding the
repository significantly reduces the size of the backup.

The backupios command that is used in Example 3-1 on page 34, creates a full
backup tar file package named nim_resources.tar. This package includes all of
the resources that are needed to restore a Virtual I/O Server (mksysb image,
bosinst.data, network boot image, and the Shared Product Object Tree (SPOT))
from a NIM or HMC by using the installios command. Section 3.1.4, “Restoring
the Virtual I/O Server” on page 38, describes the restoration methods.
Best practice dictates that a full backup is taken before you make any
configuration changes to the Virtual I/O Server. In addition to the full backup, a
scheduled weekly backup is also a good practice. You can schedule this job by
using the crontab command.
viosbr command
The viosbr command performs a backup of the Virtual I/O Server virtual and
logical configuration. In Example 3-2, a scheduled backup of the virtual and
logical configuration is set up, and existing backups are listed. The backup
frequency is daily, and the number of backup files to keep is seven.
Example 3-2 Daily viosbr schedule
$ viosbr -backup -file vios22viosbr -frequency daily -numfiles 7
Backup of this node (vios22) successful
$ viosbr -view -list
vios22viosbr.01.tar.gz
At a minimum, backing up the virtual and logical configuration data before you
make changes to the Virtual I/O Server, can help in recovering from configuration
errors.
3.1.3 Backing up user-defined virtual devices
The backupios command backs up the Virtual I/O Server operating system, but
more than that is needed to rebuild a server:
򐂰 If you are restoring to the same server, some information might be available
such as data structures (storage pools, volume groups, and logical volumes)
that are held on non-root volume group (rootvg) disks.
򐂰 If you are restoring to new hardware, these devices cannot be automatically
recovered because the disk structures do not exist.
򐂰 If the physical devices exist in the same location and structures such that the
logical volumes are intact, the virtual devices such as the virtual
Small Computer System Interface (SCSI), virtual Fibre Channel, and SEAs
are recovered during the restoration.

In a DR situation where these disk structures do not exist and network cards are
at different location codes, you need to ensure that you back up the following
devices:
򐂰 Any user-defined disk structures such as storage pools or volume groups and
logical volumes.
򐂰 The linking of the virtual device through to the physical devices.
These devices are mostly created at the Virtual I/O Server build and deploy time,
but change depending on when new clients are added or changes are made.
Backing up disk structures
Use the savevgstruct command to back up user-defined disk structures. This
command writes a backup of the structure of a named volume group (and
therefore, storage pool) to the /home/ios/vgbackups directory.
The savevgstruct command is automatically called before the backup
commences for all active non-rootvg volume groups or storage pools on a Virtual
I/O Server when the backupios command is run. Because this command is called
before the backup commences, the volume group structures are included in the
system backup. For this reason, you can use the backupios command to back up
the disk structure as well.
Backing up device mappings
The physical and virtual device mappings are contained in the viosbr backup
that is described in “viosbr command” on page 35. You can use the viosbr
command to view the device mappings, as shown in Example 3-3.
Example 3-3 List device mappings from a viosbr backup
$ viosbr -view -file vios22viosbr.01.tar.gz -mapping
Details in: vios22viosbr.01
SVSA Physloc Client Partition ID
------------------- ---------------------------------- --------------------
vhost0 U8233.E8B.061AB2P-V2-C30 0x00000003
VTD rootvg_lpar01
Status Available
Activate volume groups: The volume group needs to be activated for the
backup to succeed. Only the active volume groups or storage pools are
automatically backed up by the backupios command. Use the lsvg or lssp
commands to list and activatevg to activate the volume groups or storage
pools if necessary before you start the backup.

LUN 0x8100000000000000
Backing Device hdisk3
Physloc U78A0.001.DNWHZS4-P2-D6
Mirrored false
SVEA Physloc
------- ---------------------------------------
ent6 U8233.E8B.061AB2P-V2-C111-T1
VTD ent11
Status Available
Backing Device ent10
Physloc U78A0.001.DNWHZS4-P1-C6-T2
In addition to the information that is stored in the viosbr backup file, a full system
configuration backup can be captured by using the snap command. This
information enables the Virtual I/O Server to be rebuilt from the installation media
if necessary. The crucial information in the snap is the output from the following
commands:
򐂰 Network settings
– netstat -state
– netstat -routinfo
– netstat -routtable
– lsdev -dev entX -attr
– cfgnamesrv -ls
– hostmap -ls
– optimizenet -list
– entstat -all entX
Slot numbers: It is also vitally important to use the slot numbers as a
reference for the virtual SCSI, virtual Fibre Channel, and virtual Ethernet
devices. Do not use the vhost/vfchost number or ent number as a reference.
The vhost/vfchost and ent devices are assigned by the Virtual I/O Server as
they are found at boot time or when the cfgdev command is run. If you add in
more devices after subsequent boots or with the cfgdev command, these
devices are sequentially numbered.
The important information in Example 3-3 on page 36, is not vhost0, but that
the virtual SCSI server in slot 30 (the C30 value in the location code) is
mapped to physical volume hdisk3.

򐂰 Physical and logical volume devices
– lspv
– lsvg
– lsvg -lv VolumeGroup
򐂰 Physical and logical adapters
– lsdev -type adapter
򐂰 Code levels, users, and security
– ioslevel
– motd
– loginmsg
– lsuser
– viosecure -firewall view
– viosecure -view -nonint
We suggest that you gather this information in the same time frame as the
previous information.
The /home/padmin directory (which contains the snap output data) is backed up
using the backupios command. Therefore, it is a good location to collect
configuration information before a backup.
3.1.4 Restoring the Virtual I/O Server
Similarly to backups, there are a few different methods in restoring the Virtual I/O
Server. The most common and best practice is to restore from either the HMC or
NIM.
Restoring the Virtual I/O Server from the HMC
If you made a full backup to the file, as shown in “backupios command” on
page 34 (not a mksysb backup, but a full backup that creates the
nim_resources.tar file), you can use the HMC to restore the Virtual I/O Server
by using the installios command.
The tar file needs to be available on either the HMC, a DVD, or an NFS share. In
this scenario, we use the preferred NFS method. Assuming that the directory that
holds the nim_resources.tar file is exported from an NFS server, you now log on
to the HMC with a suitable user ID. Then, run the installios command.
Example 3-4 on page 39 shows this command.
snap output: Keep in mind that if a system memory dump exists on the Virtual
I/O Server, it is also captured in the snap output.

Example 3-4 The installios command from the HMC
hscroot@hmc9:~> installios -p vios22 -i 172.16.22.33 -S 255.255.252.0
-g 172.16.20.1 -d 172.16.20.41:/export/vios_backup/ -s
POWER7_2-SN061AB2P -m 00:21:5E:AA:81:21 -r default -n -P auto -D auto
...
...Output truncated
...
# Connecting to vios22
# Connected
# Checking for power off.
# Power off complete.
# Power on vios22 to Open Firmware.
# Power on complete.
# Client IP address is 172.16.22.33.
# Server IP address is 172.16.20.111.
# Gateway IP address is 172.16.20.1.
# Subnetmask IP address is 255.255.252.0.
# Getting adapter location codes.
# /lhea@200000000000000/ethernet@200000000000002 ping successful.
# Network booting install adapter.
# bootp sent over network.
# Network boot proceeding, lpar_netboot is exiting.
# Finished.
Now open a terminal console on the server to which you are restoring, in case
user input is required.
installios command: The trailing slash in the NFS location
nim:/export/vios_backup/ must be included in the command as shown.
The configure client network interface setting must be disabled, as shown by
the -n option. This step is necessary because the physical adapter in which
we are installing the backup, might already be used by a SEA. If so, the IP
configuration fails. Log in and configure the IP if necessary after the
installation by using a console session.
A definition of each command option is available in the installios man page.
Tip: If the installios command seems to be taking a long time to restore, this
lag is most commonly caused by a speed or duplex misconfiguration in the
network.

Restoring the Virtual I/O Server using a NIM server
The installios command is also available on the NIM server, but now, it only
supports installations from the base media of the Virtual I/O Server. The method
that we used from the NIM server is to restore the mksysb image. This image can
be the mksysb image that is generated with the -mksysb flag in the backupios
command. Or, you can extract the mksysb image from the nim_resources.tar file.
Whichever method you use to obtain the mksysb, you need to register the mksysb
as a NIM resource and generate a SPOT from the mksysb. Do this registration
after you have the backup image on the NIM server. With those prerequisites
satisfied, you can now prepare the system for the restore. Example 3-5 shows
this process.
Example 3-5 Preparing NIM resources for Virtual I/O Server restore
# nim -o define -t mksysb
-a server=master
-a location=/export/vios_backup/vios22.mksysb vios22_mksysb
# nim -o define -t spot
-a server=master
-a location=/export/vios_backup/spot
-a source=vios22_mksysb vios22_spot
# nim -o bos_inst
-a source=mksysb
-a mksysb=vios22_mksysb
-a spot=vios22_spot
-a installp_flags=-agX
-a no_nim_client=yes
-a boot_client=no
-a accept_licenses=yes vios22
With all the NIM resources ready for installation, you can now proceed to starting
the Virtual I/O Server logical partition in SMS mode and perform a network boot.
Further information about this process can be found in IBM PowerVM
Virtualization Managing and Monitoring, SG24-7590.
Important: The no_nim_client=yes option instructs the NIM server to not
register the Virtual I/O Server as a NIM client. When this option is set to no,
the NIM server configures an IP address on the physical adapter which was
used during installation. If this adapter is apart from the SEA, it causes errors
during boot.

Restoring disk structure and device mappings
If you restore a Virtual I/O Server to the same logical partition, all devices and
links to those devices are restored. However, if you are restoring to another
logical partition or system, these device mappings fail. This failure occurs
because the system serial number makes up part of the virtual device location
code; therefore, it is not present during the linking process.
If the Virtual I/O Server was presenting file-backed devices to the virtual I/O
clients, these volume group structures need to be restored by using the
restorevgstruct command. These files are in the /home/ios/vgbackups
directory if you performed a full backup by using the backupios command.
After you restore all of the logical volume structures, the only remaining step is to
restore the virtual devices that link the physical backing device to the virtual. To
restore these devices, use the viosbr outputs recorded from the backup steps in
“Backing up device mappings” on page 36. It is important to remember the slot
numbers and backing devices when you restore these mappings.
3.1.5 NIM server resilience
It is instrumental to have a working NIM server during disaster recovery
procedures. In an ideal setup, the NIM server that is responsible for restoring the
Virtual I/O Server, sits independently of the virtualized environment. But, there
are other solutions which can help NIM availability. Figure 3-1 details the best
practices in providing a reliable NIM solution.
Figure 3-1 NIM resilience solutions
VIOS VIOS
NIM
Dedicated I/O
AIX
VIOS VIOS
NIM
AIX
IBM i
Linux
VIOS VIOS
NIM
AIX
IBM i
Linux
VIOS VIOS
AIX
IBM i
Linux
NIM
Frame A
1
Frame B Frame A
3
Frame A
2
IBM i
Linux

The following guide explains the numbers that are shown at the top of Figure 3-1
on page 41.
1. The NIM is a stand-alone server, independent of the virtualized environment.
2. The NIM that is responsible for managing the logical partitions on Frame A, is
a logical partition on Frame B, and vice versa.
3. The NIM uses dedicated resources (that is, the network and storage are not
managed by the Virtual I/O Server).
In all three scenarios, the NIM continues to function if one or all of the
Virtual I/O Servers on a single frame are unavailable.
3.2 Dynamic logical partition operations
When you remove, change, or add: I/O adapters, memory, processor, or virtual
adapters by using a dynamic logical partition (LPAR), they are not automatically
reflected in the permanent profile of the partition. Therefore, whenever you
reconfigure, we suggest that you update the profile. A good practice is to first put
the changes into the partition profile. Then, make the changes dynamically to
prevent them from being lost in the case of a deactivation or shut down. The
changes that are made to the partition in the partition profile are not reflected in
the LPAR if you perform a reboot. Changes are only reflected when you start the
LPAR from a Not Activated state.
An alternative way to save the active partition assignments is by using the option
Configuration  Save Current Configuration, which is selected from the
partition context menu, as shown in Figure 3-2 on page 43. When you save the
new profile, select a meaningful name such as normal_new_<today_date>, as
shown in Figure 3-3 on page 43. After you test the new partition profile, rename
the old partition profile to, for example, normal_old_<today_date> and rename or
copy the new partition profile normal_new_<today_date> to normal. Remember to
document what you changed in the new partition profile. In addition, you might
want to clean up any old unused partition profiles from the last save or rename
them now.
An alternative method is to use the Overwrite existing profile option, as shown in
Figure 3-4 on page 44. Use caution when you use this option so you do not
overwrite a profile which is still required.
Figure 3-2 on page 43 shows the partition context menu. Example 3-3 on
page 36 shows how to save the running partition profile to a new profile.

Figure 3-2 Partition context menu
Figure 3-3 Save the running partition profile to a new profile
If you have multiple workloads that require multiple partition profiles, do not clean
up or rename the partition profiles. Activate the profiles and do not use a date
naming scheme, but one that is meaningful to this workload, such as DB2_high.
3.2.1 Dynamically adding virtual Fibre Channel adapters
One method of adding a virtual Fibre Channel adapter to a virtual I/O client, is by
using a dynamic LPAR operation, and then modifying the permanent partition
profile. This method results in a different pair of worldwide port names (WWPNs)
between the active and saved partition profiles.

When a virtual Fibre Channel adapter is created for a virtual I/O client, a pair of
unique WWPNs is assigned to this adapter by the Power Hypervisor. An attempt
to add the same adapter at a later stage results in the creation of another pair of
unique WWPNs.
When you add virtual Fibre Channel adapters into a virtual I/O client with a
dynamic LPAR operation, use the Overwrite existing profile option to save the
permanent partition profile. This option is shown in Figure 3-4. This choice
results in the same pair of WWPNs in both the active and saved partition profiles.
Figure 3-4 Overwrite the existing partition profile
3.3 Virtual media repository
The virtual media repository, also known as the virtual media library, allows the
loading of disc images in the ISO format onto the Virtual I/O Server to share with
virtual I/O clients. Configuration of the virtual media repository is simple, and is
covered in IBM PowerVM Virtualization Managing and Monitoring, SG24-7590.
One of the key benefits of the virtual media repository is the ability to install
operating systems onto virtual I/O clients. This feature is essentially a similar
functionality to what the NIM provides, but can be much quicker. A good practice
is to keep copies of the operating system gold images that can be deployed to
multiple logical partitions dynamically if the NIM is unavailable.
AIX tip: On AIX, the mkdvd and mkcd commands allow for the conversion of a
mksysb image to an ISO format, which can be loaded into the virtual media
repository. Images in the virtual media repository, which have read-only
permissions, can be shared between multiple logical partitions.

3.4 Power Systems server shutdown and startup
The following section describes best practices to shut down and start a whole
IBM Power Systems server by using the HMC.
There are two types of a Power Systems server shutdown, planned and
unplanned. In either case, with an emphasis on unplanned shutdowns, we
suggest that you have partition startup under manual control. Power Systems
server shutdowns are a rare operation, and occur mostly as a result of a disaster.
Before a planned server shutdown, save your current profiles, check if you have
the correct profiles that are set as the default, and perform any other
administration tasks. However, be prepared for unplanned server shutdowns and
do backups regularly. For more information about backups, see 3.1, “Backing up
and restoring the Virtual I/O Server” on page 32.
A best practice is to disable Power off the system after all the logical partitions
are powered off. Figure 3-5 shows how to enable and disable this option.
Figure 3-5 The Power off system option is turned off
IBM i tip: On IBM i, you can use the virtual media repository as an alternate
restart device for Licensed Internal Code installation.

If you use Live Partition Mobility (LPM) to optimize your power consumption, set
partition auto startup for the Virtual I/O Servers. This setting ensures that after
the server is started, it will be ready for LPM as soon as possible. For more
information, see 7.3, “Live Partition Mobility” on page 109.
Select Automatically start when the managed system is powered on for the
Virtual I/O Servers, and for any logical partitions that do not depend on the
Virtual I/O Servers. We also suggest enabling this function on logical partitions
that you want to achieve the best memory and processor affinity.
To set this option in the HMC, select the logical partition and the profile that you
want to start automatically. Choose Settings and select the check box for an
automatic start, as shown in Figure 3-6.
Figure 3-6 Automatically start when the managed system is powered on
You can specify an order in which LPARs are started automatically. The
automatic startup of all the logical partitions might work in your environment;
however, there are some things to consider:
򐂰 After the Power Systems server starts up, for every LPAR that is being
activated, new memory and processor affinity will be defined.
Schedule startup and shutdown: You can schedule the Power Systems
server startup and shutdown via the HMC:
Servers -> ServerName -> Operations -> Schedule Operations.

򐂰 If you start all the partitions at the same time, the client logical partitions with
the boot devices that are dependent on the Virtual I/O Server (virtual SCSI or
NPIV), wait and try again to boot. The partitions try again to boot until at least
one Virtual I/O Server finishes its activation. In dual Virtual I/O Server setups,
the following steps occur:
– Your logical partition will boot after the first Virtual I/O Server comes up.
This action might result in stale physical partitions (PPs) in volume groups
that are mirrored through both Virtual I/O Servers.
– Some storage paths might be in an inactive state.
– All logical partitions might be using only a single Virtual I/O Server for
network bridging.
򐂰 There might be application dependencies, for example, a
Domain Name System (DNS), a Lightweight Directory Access Protocol
(LDAP), a Network File System (NFS), or a database that is running on the
same Power Systems server. If there are dependencies, check that these
partitions activate correctly.
Consider the following options when you are defining your startup order list:
򐂰 Start the Virtual I/O Servers first or set them to start automatically.
򐂰 Set Partition start policy to Auto-Start Always or Auto-Start for
Auto-Recovery, as shown in Figure 3-7. In the HMC, choose Server ->
ServerName -> Properties, Power-On Parameters.
Figure 3-7 Partition start policy
򐂰 The LPARs with the highest memory and the highest number of processors
are started first to achieve the best processor and memory affinity. For critical
production systems, you might want to activate them before the Virtual I/O
Servers.

򐂰 Start the LPARs that provide services for other partitions. For example, first
start with the DNS and LDAP servers. Also, start the NFS server first,
especially if there are NFS clients with NFS hard mounts on startup.
򐂰 Prepare the HMC commands to simplify startup. A spreadsheet might help
you create those commands. To start an LPAR and open its virtual terminal
on a specific system, use the commands shown in Example 3-6.
Example 3-6 HMC CLI startup commands
chsysstate -m <servername> -o on -r lpar -n vios1a -f <profilename>
mkvterm -m <servername> -p vios1a
The startup sequence does not need to be serial; many LPARs can be started at
the same time. The following scenario provides a server startup example:
򐂰 Activate the critical systems with a high amount of memory and processors,
first.
򐂰 Start all Virtual I/O Servers at the same time.
򐂰 After the Virtual I/O Servers are up and running, activate all your DNS, LDAP,
or NFS servers at the same time.
򐂰 Activate all database LPARs at the same time. There might be a cluster that
needs to be started.
򐂰 After the database LPARs are up and running, activate all the application
LPARs at the same time.
Grouping partitions: You can use system profiles to group the partitions. For
more information, see this website:
http://www.ibm.com/developerworks/aix/library/au-systemprofiles/

Chapter 4. Networking
Virtual network infrastructure that is used in IBM Power Systems is extremely
flexible and allows for a number of configuration options. This chapter describes
the most common networking setups that use a dual Virtual I/O Server
configuration as a best practice.
4

4.1 General networking considerations
Chapter 4 describes several best practice scenarios. The basic difference is
determining how many Virtual I/O Servers that you use and whether you are
going to use virtual local area networks (VLANs). Table 4-1 shows the most
important terminology that is used in Chapter 4.
Table 4-1 Terminology that is used in this chapter
The following contains a list of best practices for your virtual environment:
򐂰 Before you start planning your virtual network infrastructure, speak with your
network administrators and specialists to synchronize your terminology.
򐂰 Keep things simple. Document them as you go.
򐂰 Use VLAN tagging. Use the same VLAN IDs in the virtual environment as
they exist in the physical networking environment.
򐂰 For virtual adapters on client logical partitions (LPARs), use Physical Volume
Identifiers (PVIDs). For simplification of the installation, do not configure
multiple VLANs on one adapter and do not use AIX VLAN tagged interfaces
on it.
򐂰 Use hot pluggable network adapters for the Virtual I/O Server instead of the
built-in integrated network adapters. They are easier to service.
Terminology Definition in this chapter
Link
Aggregation
(LA)
LA aggregates more physical connections that operate as one link. It
increases network throughput and high availability. It has two basic
modes of operation: Etherchannel and 802.3ad, also known as port
trunking.
Network
Interface
Backup (NIB)
NIB provides a highly available connection by having a backup
adapter that, by default, is not active. It is usually used on different
LAN cards that are connected to different LAN switches.
VLAN tagging Standard 802.1Q allows you to have more VLANs defined on one
physical connection or a Link Aggregation. Also known as VLAN
trunking or simply, trunking.
Virtual
network
infrastructure
Virtual network infrastructure implementation that is used on
IBM Power Systems. The Power Hypervisor supports 802.1Q
standard.
External
network
infrastructure
A physical network infrastructure outside of the IBM Power Systems.
Although it is referred to as a physical network infrastructure,
virtualization techniques do exist there, too.

Chapter 4. Networking 51
򐂰 Use two Virtual I/O Servers to allow concurrent online software updates to the
Virtual I/O Server.
򐂰 Spread physical Virtual I/O Server resources across multiple enclosures and
I/O drawers.
򐂰 Configure an IP address on a Shared Ethernet Adapter (SEA) to allow the
ping feature of a SEA failover.
򐂰 Use separate virtual Ethernet adapters for Virtual I/O Server management
rather than putting it on the SEA. You can lose network connectivity if you are
changing SEA settings.
򐂰 Use a vterm when you configure the network on a Virtual I/O Server.
򐂰 Find a balance between the number of virtual Ethernet adapters and the
number of virtual local area networks (VLANs) for each virtual Ethernet
adapter in a SEA. Try to group common VLANs on the same virtual Ethernet
adapter. For example, group production VLANs on one adapter and
non-production VLANs on another.
򐂰 Use multiple virtual switches when you work in multi-tenant environments.
Remember the following technical constraints during your configuration:
򐂰 The maximum number of VLANs per virtual adapter is 21 (20 VLAN IDs
(VIDs) and 1 Physical Volume Identifier (PVID)).
򐂰 The maximum number of virtual adapters for each SEA is 16.
򐂰 The maximum number of physical Ethernet adapters in a link aggregated
adapter is eight for primary and one for backup.
򐂰 The maximum virtual Ethernet frame size on the Power Systems server is
65,408 bytes.
4.1.1 Shared Ethernet Adapter considerations
The following considerations pertain to Shared Ethernet Adapters (SEAs).
VLAN bridging
If you use the suggested VLAN tagging option for the virtual network
infrastructure, there are three ways to define VLANs for one SEA:
򐂰 All VLANs are on one virtual adapter, one adapter under a SEA.
VLAN ID 3358: Avoid the use of VLAN ID 3358 in your infrastructure. It is
reserved to enable the tme attribute for the virtual Fibre Channel and SAS
adapters.

򐂰 One VLAN per one virtual adapter, many adapters under a SEA.
򐂰 A combination of the two configurations.
Changing a list of virtual adapters that are bridged by a SEA can be done
dynamically. One SEA can have a maximum of 16 virtual Ethernet adapters. If
you need to bridge more than 16 VLANs for each SEA, define more VLANs for
each virtual Ethernet adapter.
To change a set of VLANs on a virtual Ethernet adapter, check whether your
environment supports dynamic change. For more information, see the following
website:
http://pic.dhe.ibm.com/infocenter/powersys/v3r1m5/index.jsp?topic=/p7hb
1/iphb1_vios_managing_vlans.htm
If your environment does not support dynamic changes, remove the adapter from
the SEA and then dynamically remove the adapter from the LPAR. Alter the
VLAN definitions on the virtual adapter, add it dynamically back to the system,
and alter the SEA configuration. During this operation, all access to the VLANs
defined on the virtual adapter is unavailable by that Virtual I/O Server. However,
all client LPARs automatically fail over and use the other Virtual I/O Server.
Remember to apply the changes to the partition profile.
Shared Ethernet Adapter threading
There are two Shared Ethernet Adapter (SEA) threading modes of operation:
threaded and non-threaded. At the time of this writing, threaded is the default
option. If your Virtual I/O Server is virtualizing both the SCSI and the network and
you experience disk performance issues, consider switching the threading off.
For more information, see this website:
http://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/perf
.html
4.1.2 Maximum transmission unit best practices
Table 4-2 on page 53 shows the maximum transmission unit (MTU) values in
different environments. We suggest checking with your network administrator
which MTU value is appropriate for your network. Usually you can use a higher
MTU for communication within a network. When the communication must pass
through a router, the MTU is 1500, by default.

Table 4-2 Typical maximum transmission units (MTUs)
When data over 1500 bytes per packet are sent over a network, consider
switching to a larger MTU value. For example, the size of a DNS query is small,
and a large MTU value has no effect here. However, backups over the network
are very demanding on the bandwidth, and might benefit from a larger MTU
value.
Ensure that path MTU discovery is on. Use the pmtu command on AIX, the
CFGTCP command on IBM i, and the tracepath command on Linux.
To change the MTU on a specific adapter, follow these steps:
򐂰 On AIX, enter chdev -a mtu=<new mtu> -l enX
򐂰 On Linux, see the distribution documentation. The generic rule is to use the
ifconfig command. Also, use the appropriate parameters under
/proc/sys/net/
򐂰 On IBM i, use the following procedure:
a. In the command line, type CFGTCP and press Enter.
b. Select option 3 (Change TCP/IP attributes).
c. Type the command 2 (Change) next to the route you want to change.
d. Change the MTU size in the Maximum transmission unit field. Press Enter
to apply the change.
If you send data through a network and the data is larger than the MTU of the
network, it becomes fragmented, which has a negative affect on performance. If
you experience network issues after an MTU change, check the MTU for a
specific remote host or network by using ping -s <size> on AIX or Linux. Or on
IBM i, use PING RMTSYS(‘<remote address>’) PKTLEN(<size>).
Network type MTU (bytes)
Standard maximum MTU 65,535
IBM Power Systems maximum MTU 65,390
Ethernet (gigabit) 9000
Ethernet (10 or 100 MBps) 1500
Standard minimum MTU 68

4.1.3 Network bandwidth tuning
To achieve the maximum out of high speed adapters, consider the following best
practices:
򐂰 Use the largest possible MTU that is available in your network environment.
򐂰 IBM Power Systems servers support an MTU up to 65,390. This size is
sensible for huge network transfers that do not leave the Power Systems
server. An example is a backup within a Power Systems server.
򐂰 Use path MTU discovery on client partitions.
򐂰 It is important to set flow control on both the switch and the Virtual I/O Server
to avoid packet resending in case the receiver is busy.
򐂰 Unless you have IBM i or Linux in your environment, enable the large_send
and large_recieve attributes.
For optimum performance, especially on a 10-Gb network, tuning needs to be
performed. The most important settings are provided in the following list:
򐂰 Physical adapters on the Virtual I/O Server
jumbo_frames=yes, large_send=yes, large_recieve=yes, flow_ctrl=yes
򐂰 Shared Ethernet adapters on the Virtual I/O Server
jumbo_frames=yes, large_send=yes, large_recieve=yes
򐂰 Client partitions
– Running AIX
• NIB on client, if applicable
jumbo_frames=yes
• Network interface (enX)
mtu to the largest possible value for your network, mtu_bypass=yes
• Network options
tcp_pmtu_discovery=1, udp_pmtu_discovery=1
– Running IBM i, set the MTU as described in 4.1.2, “Maximum transmission
unit best practices” on page 52
– Running Linux, see the documentation for your distribution. The generic
rule is to use the ifconfig command. Also, set the appropriate
parameters under /proc/sys/net/

4.2 Single Virtual I/O Server
The Virtual I/O Server is reliable and stable. If you are using a single
Virtual I/O Server, the best practices for networking are simple: Keep every
connection highly available. However, for concurrent updates and maintenance,
you must either have two Virtual I/O Servers or use the Live Partition Mobility
(LPM):
򐂰 Use LA wherever possible. You achieve high availability and an increase in
network performance.
򐂰 If you cannot use LA, use NIB.
򐂰 Single Virtual I/O Server setups are common on hardware that have a limited
number of ports. If possible, use VLAN tagging.
򐂰 Use ports from different adapters that are in different enclosures and I/O
drawers.
4.3 Virtual network redundancy and failover technology
There are multiple ways to provide virtual network redundancy and failover
capability. Broadly, they can be classified into two types that depend on which
entity makes the failover decision:
򐂰 A server-side failover solution that is based on the SEA failover capability and
configuration (in the Virtual I/O Server).
򐂰 A client-side failover solution that is based on the EtherChannel NIB failover
capability and configuration (in the client LPAR).
The main advantage of a server-side failover solution is that it simplifies the client
configuration because you only need to create a single virtual Ethernet adapter
in the client, which is served by two SEAs that are configured in the high
availability mode (failover configuration) in two Virtual I/O Servers. While the
client-side failover solution requires two virtual Ethernet adapters and an
additional NIB configuration for each client LPAR.
Furthermore, for the virtual Ethernet, since the link status is always UP, the NIB
failover must rely on the ping address feature. Also, there are limitations in the
case of a failback. See 4.3.1, “Dual Virtual I/O Server with VLAN tagging” on
page 56.

One clear advantage that the client-side failover solution had over the server-side
failover solution was that the client side option allows load sharing. With the new
load sharing feature that is added to the SEA high availability configuration, this
advantage is reduced.
4.3.1 Dual Virtual I/O Server with VLAN tagging
It is essential that you balance the network workload through as many adapters
as possible. Especially in a 10-Gb network environment, where adapters can be
costly, it is important to use your entire network infrastructure. The following
configurations are two of the best options to use your network adapters in a dual
Virtual I/O Server environment:
򐂰 The use of two virtual switches, one virtual switch for each Virtual I/O Server.
With this option, you can balance network bridging between Virtual I/O
Servers for every client and every VLAN, individually.
򐂰 The use of SEA failover with load sharing. With this option, you can balance
network bridging between Virtual I/O Servers for every VLAN, individually. All
LPARs with a specific VLAN are bridged by the same Virtual I/O Server.
VLAN distribution between Virtual I/O Servers is automatic.
Dual virtual switch configuration
Figure 4-1 on page 57 shows two Virtual I/O Servers and two LPARs with a dual
virtual switch configuration. Each LPAR has a different set of VLANs.
Each Virtual I/O Server has a Link Aggregation of two physical 1-Gb ports. All
connections operate in trunking mode and are used to trunk VLAN IDs 1,2, and
3. Those VLAN IDs are used in the external network infrastructure (physical
switches), as well. Virtual Ethernet adapters that are in Virtual I/O Servers
operate in the trunking mode. Therefore, the VLAN tag is not removed from the
Ethernet packet when a packet is received from the external network
infrastructure.
In the example that is shown in Figure 4-1 on page 57, the shared Ethernet
adapter connects to two virtual Ethernet adapters. There is one adapter for VLAN
IDs 1,2, and one adapter for VLAN IDs 3,4,5. You can define up to 20 VLANs for
each virtual Ethernet adapter in 802.1Q mode.
IBM i: For IBM i, we suggest that you use a SEA failover with load sharing. If
you use a virtual Interface in a dual virtual switch configuration and there is a
network path failure, IBM i does not detect the loss of connectivity through the
virtual Ethernet adapter.

All the virtual Ethernet adapters of each Virtual I/O Server connect to one virtual
switch. Try to avoid defining virtual Ethernet adapters from different Virtual I/O
Servers in one virtual switch. This way, you eliminate the chance of networking
issues because of a network misconfiguration.
Figure 4-1 Dual Virtual I/O Server configuration with two virtual switches
Each client partition has a NIB that is made up of two virtual Ethernet adapters
for every VLAN to which it connects. For details about a NIB setup that uses
virtual Ethernet adapters, see 4.1.1, “Shared Ethernet Adapter considerations”
on page 51. For more information and to obtain a setup guide, see this website:
http://www.ibm.com/support/techdocs/atsmastr.nsf/WebIndex/WP101752
PVID: The PVID that is used on the virtual Ethernet adapter, which makes up
the SEA, can be a PVID which is not used in your network.
NIB NIB
viosA
Link aggregation
802.3ad
External network infrastructure
PVID=99
VIDs=1,2
PVID=98
VIDs=3,4,5,..
SEA
ent
(1 Gb)
VLANs
1,2,3,..
VLANs
1,2,3,..
viosB
Link aggregation
802.3ad
PVID=99
VIDs=1,2
PVID=98
VIDs=3,4,5,..
SEA
Link aggregation
802.3ad
1 Gb 1 Gb
Link aggregation
802.3ad
1 Gb 1 Gb
NIB
Client Partition 1
POWER
Hypervisor
Client Partition 2
ent
(1 Gb)
ent
(1 Gb)
VLANs
1,2,3,..
VLANs
1,2,3,..
ent
(1 Gb)
vswitchB
vswitchA
NIB
PVID=1
PVID=1
NIB
PVID=2 PVID=2
NIB
PVID=3 PVID=3

Network Interface Backup on client LPARs
If you use the Network Interface Backup (NIB) with virtual Ethernet adapters on
AIX, the NIB is unable to obtain a link status because the virtual Ethernet adapter
is always up. This configuration has a few impacts:
򐂰 The NIB must ping an IP address that is outside of the Power Systems server.
To avoid the error log from filling up with NIB entries, apply an error log filter
for error ID 5561971C by using the errupdate command.
򐂰 The error log contains entries about the NIB being unable to perform certain
operations. Best practice is to ping a gateway of a specific network, because
this IP address is always up. If this IP is not reachable, something is
misconfigured on either the Virtual I/O Server or the external network
infrastructure.
򐂰 In a failover scenario, the NIB does not fail back to the primary adapter when
the network path through the primary adapter recovers. It fails back only if the
network path through the backup adapter fails.
If the address that is configured to ping in a NIB does not respond, the error log
fills up rapidly with errors about the NIB failover and that the destination is
unreachable. Message IDs 6169289D and DC32143C are logged to the error log.
We suggest to not filter these errors, but check and clean them periodically until
the problem is resolved. Also, consider expanding the AIX error log size by using
the /usr/lib/errdemon -s <size> command.
4.3.2 Shared Ethernet Adapter failover with load sharing
Another approach is to use the traditional Shared Ethernet Adapter (SEA)
failover setup, enhanced with load sharing, as shown in Figure 4-2 on page 59.
Load balancing between SEAs is on a per virtual Ethernet adapter basis. This
means that all VLANs defined in one virtual Ethernet adapter that are used by a
SEA, are bridged by one Virtual I/O Server.
Linux and IBM i tips: For Linux systems, check the distribution
documentation about bonding. For IBM i, we do not suggest that you use this
functionality on virtual Ethernet adapters.

Figure 4-2 Dual Virtual I/O Server configuration with SEA and load balancing
The SEA is connected to two virtual Ethernet adapters; each adapter has a
different set of VLAN IDs. Virtual Ethernet adapters on different Virtual I/O
Servers must have the same set of VLANs and different trunk priorities.
In addition, the SEA is connected to another virtual Ethernet adapter with a
PVID=100 that is used as a SEA control channel. The control channel is used for
SEA heartbeating and exchanging information between the two SEA adapters on
the set of VLAN IDs that each SEA bridges.
For more information and to obtain a setup guide, see this website:
http://www.ibm.com/support/docview.wss?uid=isg3T7000527
NIB NIB
viosA
Link aggregation
802.3ad
External network infrastructure
PVID=99
VIDs=1,2
PVID=98
VIDs=3,4,..
ent
(1 Gb)
VLANs
1,2,3,..
VLANs
1,2,3,..
viosB
Link aggregation
802.3ad
PVID=99
VIDs=1,2
PVID=98
VIDs=3,4,..
Link aggregation
802.3ad
1 Gb 1 Gb
Link aggregation
802.3ad
1 Gb 1 Gb
NIB
Client Partition 1
POWER
Hypervisor
Client Partition 2
PVID=2 PVID=3
ent
(1 Gb)
ent
(1 Gb)
VLANs
1,2,3,..
VLANs
1,2,3,..
ent
(1 Gb)
PVID=1
PVID=100 PVID=100
ETHERNET0
SEA SEA

Chapter 5. Storage
The Virtual I/O Server has different means to provide storage access to virtual
I/O clients. This chapter addresses the best practices on the different methods
for presenting and managing storage on the Virtual I/O Server and
virtual I/O clients.
5

5.1 Storage Considerations
This section describes the various storage requirements from the Virtual I/O
Server perspective.
We also describe the different methods of providing storage access to a virtual
I/O client:
Virtual Small Computer System Interface
The virtual Small Computer System Interface (SCSI) was the first method for
IBM PowerVM to provide virtual storage to virtual I/O clients. With the SCSI,
storage resources that are presented to the virtual clients, need to be mapped to
physical storage by the Virtual I/O Server. The most current versions of the
Virtual I/O Server introduced new features, such as file-backed devices, virtual
optical devices, and virtual tape devices.
Virtual Fibre Channel
Virtual Fibre Channel is the PowerVM implementation of the Fibre Channel
industry standard N_Port ID Virtualization (NPIV).
NPIV can share a physical Fibre Channel adapter across multiple virtual I/O
clients. It has the advantage to provide direct Fibre Channel access to the virtual
clients from the external storage devices, such as disks arrays and tape libraries.
It does not require an extensive storage mapping effort from the Virtual I/O
Server.
For more information about setting up and managing virtual storage, see IBM
PowerVM Virtualization Managing and Monitoring, SG24-7590.
5.1.1 Virtual I/O Server rootvg storage
There are many aspects to consider regarding the storage management at the
Virtual I/O Server. One of most common questions is whether to install the
root volume group (rootvg) of the Virtual I/O Server on the internal or external
storage. There are some advantages to both options. The choice for a booting
device depends on your infrastructure and on the number of internal disks, disk
adapters, and Fibre Channel adapters that are available.
Newer server technology, such as IBM PureSystems, also adds to the decision
process because their high density design favors the use of SAN booting
because of the number of internal drives and adapters that are available.

Chapter 5. Storage 63
The best practice for booting a Virtual I/O Server is to use internal disks rather
than external SAN storage, if the server architecture and available hardware
allows. The following list provides reasons for booting from internal disks:
򐂰 The Virtual I/O Server does not require specific multipathing software to
support the internal booting disks. This configuration helps when you perform
maintenance, migration, and update tasks.
򐂰 The Virtual I/O Server does not need to share Fibre Channel adapters with
virtual I/O clients, which helps if a Fibre Channel adapter replacement is
required.
򐂰 The virtual I/O clients might have issues with the virtual SCSI disks presented
by the Virtual I/O Server that is backed by SAN storage. If so, the
troubleshooting can be performed from the Virtual I/O Server.
򐂰 We do not recommend the allocation of logical volume backing devices for
virtual I/O clients in the rootvg of the Virtual I/O Server.
򐂰 The SAN design and management is less critical.
򐂰 Access to the dump device is simplified.
A SAN boot provides the following advantages:
򐂰 SAN hardware can accelerate booting through their cache subsystems.
򐂰 Redundant Array of Independent Disks (RAID) and mirroring options might be
improved.
򐂰 A SAN boot is able to take advantage of advanced features that are available
with SAN storage.
򐂰 Easy to increase capacity.
򐂰 Provides smaller server hardware footprints.
򐂰 Zones and cables can be set up and tested before client deployment.
In general, avoid single points of failure. Following is a list of best practices for
maintaining the Virtual I/O Server availability:
򐂰 Mirror rootvg by using the mirrorios -defer hdiskX command, and then
reboot the Virtual I/O Server at your earliest convenience.
򐂰 If rootvg is mirrored, use disks from two different disk controllers. Ideally,
those disk controllers are on different Peripheral Component Interconnect
(PCI) busses.
Booting from external storage: If you are booting from external storage, use
multipathing software, with at least two Fibre Channel adapters that are
connected to different switches.

For more information about installing the Virtual I/O Server on a SAN, see:
򐂰 IBM Flex System p260 and p460 Planning and Implementation Guide,
SG24-7989
http://www.redbooks.ibm.com/redbooks/pdfs/sg247989.pdf
򐂰 IBM System Storage DS8000 Host Attachment and Interoperability,
SG24-8887
http://www.redbooks.ibm.com/redbooks/pdfs/sg248887.pdf
5.1.2 Multipathing
Multipathing is a best practice in terms of performance and redundancy; it can be
implemented from both the virtual I/O client and the Virtual I/O Server.
Multipathing provides load balancing and failover capabilities if an adapter
becomes unavailable. This process also helps if there is a SAN problem on the
path to the external storage.
A virtual SCSI configuration requires more management from the Virtual I/O
Server than NPIV. Figure 5-1 on page 65 shows an example of dual Virtual I/O
Servers that provide a multipath SAN disk to a virtual I/O client. In this case,
multipathing is managed from both the virtual I/O client and Virtual I/O Server.
Multipathing for clients with dual Virtual I/O Servers
Whether you implement multipathing at the Virtual I/O Server, the virtual I/O
client, or both, it is a best practice to have dual Virtual I/O Servers that are
backed by external storage. The following list provides reasons to back the
servers by external storage:
򐂰 Unscheduled outages, such as physical device failures and human errors.
򐂰 Scheduled operations that are required for Virtual I/O Server maintenance.
򐂰 High availability and load balancing of storage I/Os.
򐂰 Automatic path failover protection.
򐂰 Prevention of single points of failure on the path to the external storage.
With NPIV, multipathing is only managed from the virtual I/O client. For more
information, see section 5.4, “N-Port ID Virtualization” on page 88.
Figure 5-1 on page 65 shows a virtual SCSI client with multipathing from dual
Virtual I/O Servers.

Figure 5-1 Virtual SCSI client with multipathing from dual Virtual I/O Servers
For more information about virtual I/O client configuration, see section 5.2.2,
“Configuring the Virtual I/O Server with a virtual SCSI” on page 71.
Physical Resources
Virtual Resources
VIOS 1
Server SCSI Adapter
Physical
FC Adapter
Physical
FC Adapter
SAN
Switch
Client Partition
Client SCSI Adapter Client SCSI Adapter
VIOS2
Server SCSI Adapter
Physical
FC Adapter
Physical
FC Adapter
Multipath Software Multipath Software
Default MPIO
SAN
Disk
SAN
Switch
SAN
Switch
Primary Path
Secondary Path

Supported storage on the Virtual I/O Server
The Virtual I/O Server supports a number of storage devices. Contact your
storage vendor for more information about Virtual I/O Server support. A list of the
current supported storage solutions is available at this website:
http://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/data
sheet.html#solutions
Confirm that the Virtual I/O Server supports the multipathing device driver that
you plan to use. For example, the Subsystem Device Driver Path Control Module
(SDDPCM), is the path control module that is designed to support a multipathing
configuration with most of the IBM storage servers. At the time of this writing, the
SDDPCM does not support the Virtual I/O Server with the IBM DS4000®,
DS5000, or DS3950 storage subsystems. If you need SDDPCM support
for the IBM DS8000® or SVC, you must use the manage_disk_drivers command
to ensure that the driver for your DS4000 and DS5000 is set to AIX_APPCM, not to
AIX_SDDAPPCM. Example 5-1 shows this command. This setting is valid for both
the virtual SCSI clients and Virtual I/O Server usage.
For more information about the SDDPCM, see this website:
Example 5-1 Manage the DS4000 and DS5000 disk driver
$ oem_setup_env
# manage_disk_drivers -l
Device Present Driver Driver Options
2810XIV AIX_AAPCM AIX_AAPCM,AIX_non_MPIO
DS4100 AIX_APPCM AIX_APPCM,AIX_fcparray,AIX_SDDAPPCM
DS3950 AIX_APPCM AIX_APPCM,AIX_SDDAPPCM
DS5020 AIX_APPCM AIX_APPCM,AIX_SDDAPPCM
DCS3700 AIX_APPCM AIX_APPCM
DS5100/DS5300 AIX_APPCM AIX_APPCM,AIX_SDDAPPCM
DS3500 AIX_APPCM AIX_APPCM
XIVCTRL MPIO_XIVCTRL MPIO_XIVCTRL,nonMPIO_XIVCTRL

For more information, see the SDDPCM for AIX and Virtual I/O Server support
matrix at this website:
http://www.ibm.com/support/docview.wss?rs=540&uid=ssg1S7001350#AIXSDDPCM
5.1.3 Mixing virtual SCSI and NPIV
It is possible to mix a virtual Small Computer System Interface (SCSI) and
N-Port ID Virtualization (NPIV) within the same virtual I/O client. You can have
rootvg or booting devices that are mapped via virtual SCSI adapters, and data
volumes that are mapped via NPIV.
Mixing NPIV and a virtual SCSI has advantages and disadvantages, as shown in
Table 5-1.
Table 5-1 Booting from a virtual SCSI with NPIV for data
5.1.4 Fibre Channel adapter configuration
The following recommendations apply to both a virtual SCSI and NPIV. There are
two levels of configuration for the Fibre Channel adapters at the Virtual I/O
Server:
fscsiX Fibre Channel protocol driver.
fcsX Fibre Channel adapter driver.
Fibre Channel protocol driver attributes
The fscsi devices include specific attributes that must be changed on
Virtual I/O Servers. It is a best practice to change the fc_err_recov and the
dyntrk attributes of the fscsi device. Both attributes can be changed by using the
chdev command, as shown in Example 5-2 on page 68.
SDDPCM: If you need to have SDDPCM installed or upgraded, it toggles the
DS4000 and DS5000 device driver management to SDDPCM
(AIX_SDDAPPCM). Check and change the parameter before you reboot the
Virtual I/O Server.
Advantages It makes multipathing software updates easier for data disks.
You can use the Virtual I/O Server to perform problem
determination when virtual I/O clients have booting issues.
Disadvantages Requires extra management at the Virtual I/O Server level.
Live Partition Mobility (LPM) is easier with NPIV.

Example 5-2 The fscsi attribute modification
$ chdev -dev fscsi0 -attr fc_err_recov=fast_fail dyntrk=yes
fscsi0 changed
Changing the fc_err_recov attribute to fast_fail fails any I/Os immediately if
the adapter detects a link event, such as a lost link between a storage device and
a switch. The fast_fail setting is only recommended for dual Virtual I/O Server
configurations. Setting the dyntrk attribute to yes allows the Virtual I/O Server to
tolerate cable changes in the SAN.
It is a best practice to change the fscsi attributes before you map the external
storage so that you do not need to reboot the Virtual I/O Server.
Fibre Channel device driver attributes
The fcs devices include specific attributes that can be changed on Virtual I/O
Servers. These attributes are num_cmd_elems and max_xfer_size. As a best
practice, complete a performance analysis on the adapters and change these
values, if needed. This analysis can be done with tools such as the fcstat
command as shown in Example 5-3 on page 69.
num_cmd_elems Modifies the number of commands that can be queued to the
adapter. Increasing num_cmd_elems decreases the No
Command Resource Count. Increasing num_cmd_elems also
makes it more likely to see No Adapter Elements Count, or
No DMA Resource Count increasing.
max_xfer_size Has an effect on the direct memory access (DMA) region
size that is used by the adapter. Default max_xfer_size
(0x100000) gives a small DMA region size. Tuning up
max_xfer_size to 0x200000 provides a medium or large
DMA region size, depending on the adapter.
Example 5-3 on page 69 shows that there is no need to change the
max_xfer_size, since the No DMA Resource Count did not increase. In the same
example, consider increasing the num_cmd_elem since the No Command Resource
Count increased. These values are measured since the last boot or last reset of
the adapter statistics.
Fscsi and fcs changes: Any change to the fscsi and fcs attributes need to be
first checked with your storage vendor.

Example 5-3 Monitoring fcs adapter statistics
$ fcstat fcs0|grep -p 'Driver Information'
IP over FC Adapter Driver Information
No DMA Resource Count: 0
No Adapter Elements Count: 395
FC SCSI Adapter Driver Information
No Command Resource Count: 2415
Both attributes can be changed by using the chdev command, as shown in
Example 5-4.
Example 5-4 fcs attributes modification
$ lsdev -dev fcs0 -attr max_xfer_size,num_cmd_elems
value
0x100000
500
$ chdev -dev fcs0 -attr num_cmd_elems=1024 -perm
fcs0 changed
$ chdev -dev fcs0 -attr max_xfer_size=0x200000 -perm
fcs0 changed
In summary, consider the following best practices for the Fibre Channel device
driver values:
򐂰 Do not increment these values without an appropriate analysis.
򐂰 Increment these values gradually.
򐂰 Reboot the Virtual I/O Servers for these changes to take effect.
Example 5-5 demonstrates how to check whether there is a need to increase the
num_cmd_elem by using the SDDPCM commands. We can verify that the value of
869 did not reach the num_cmd_elems limit of 1024.
Example 5-5 Check num_cmd_elems using the SDDPCM pcmpath command
$ oem_setup_env
# pcmpath query adapter
Total Dual Active and Active/Asymmetric Adapters : 6
Adpt# Name State Mode Select Errors Paths Active
0 fscsi16 NORMAL ACTIVE 3032 0 16 16

> 5 fscsi12 NORMAL ACTIVE 111487891 0 150 150
{kfw:root}/ # pcmpath query adaptstats 5 aa
Adapter #: 5
=============
Total Read Total Write Active Read Active Write Maximum
I/O: 106925273 33596805 0 0 869
SECTOR: 3771066843 1107693569 0 0 27784
{kfw:root}/ # lsattr -El fcs12 -a num_cmd_elems
num_cmd_elems 1024 Maximum number of COMMANDS to queue to the adapter True
5.2 Virtual Small Computer System Interface
With a virtual Small Computer System Interface (SCSI), a Virtual I/O Server
maps storage resources to virtual I/O clients. The Virtual I/O Server provides an
abstraction layer to the virtual I/O clients. Section 5.2 covers the following virtual
SCSI mapping options:
򐂰 Physical volumes
򐂰 Logical volumes
򐂰 File backed devices
򐂰 Logical units that are mapped from Shared Storage Pools (SSPs)
򐂰 Virtual optical devices
򐂰 Virtual tape devices
5.2.1 When to use a virtual Small Computer System Interface
A virtual SCSI is the only choice when there is no SAN infrastructure available.
This interface is also the choice for sharing internal optical devices and
presenting Shared Storage Pool logical units.
IBM i functionality: IBM i can host a virtual SCSI server adapter similarly to a
Virtual I/O Server. It can also provide virtual disk units, virtual optical drives,
and virtual Ethernet adapters to the clients. This functionality is not described
in this book because it is not a PowerVM feature.

The best practice for selecting between these virtual SCSI options, is that the
disk devices that are backed by SAN storage, are exported as physical volumes.
With internal devices, if you have enough for all the virtual I/O clients, assign the
entire disk as a backing device. However, if you have a limited number of disks,
create storage pools so that storage can be managed from the Virtual I/O Server.
IBM i
A virtual SCSI allows for the connection to external storage, which natively does
not support an IBM i block size. IBM i, for historical reasons, works with
520 bytes per sector of formatted storage. However, the Virtual I/O Server uses
industry-standard 512 bytes per sector. In this case, the Hypervisor manages the
block size conversion for the IBM i virtual I/O clients.
5.2.2 Configuring the Virtual I/O Server with a virtual SCSI
Managing, documenting, and tracking the virtual to physical resources
configuration is important, especially when it comes to large infrastructures. It is
also a best practice for problem determination because it is usually not the time
to determine how your virtual to physical relationship is built.
As stated in 1.3.9, “Slot numbering and naming conventions” on page 12, we
want to emphasize that you need to implement a naming convention for your
virtual devices. Although it must be managed for NPIV clients, it is sensitive for
virtual SCSI mapped resources.
Virtual adapters slot considerations
There are many different options to establish a virtual adapter slot numbering
convention. Slot numbers are shared between virtual storage and virtual network
devices. In complex systems, there tend to be far more storage devices than
network devices because each virtual SCSI device can communicate with only
one server or client. We suggest reserving the slot numbers through 20 for
network devices on all logical partitions to keep the network and storage devices
grouped.
Note: With virtual SCSI, mirroring of client volume groups must be
implemented at the level of the client. The logical volume backing device
cannot be mirrored on the Virtual I/O Server. Ensure that the logical volume
backing devices sit on different physical volumes and are served from dual

In simple environments, the management can be simplified by keeping virtual
adapter slot numbers consistent between the client and the server. Start the
virtual SCSI slots at 20, then add the lpar_id of the client to this base.
Example 5-6 shows the HMC command output.
Example 5-6 HMC virtual slots listing with a single Virtual I/O Server
hscroot@hmc9:~>lshwres -r virtualio --rsubtype scsi -m
POWER7_1-SN061AA6P --level lpar -F
lpar_name,lpar_id,slot_num,adapter_type,remote_lpar_name,remote_lpar_id
,remote_slot_num | grep AIX_02
vios1a,1,25,server,AIX_02,5,25
AIX_02,5,25,client,vios1a,1,25
In Example 5-6, vios1a has slot 25, which maps to AIX_02 slot 25.
With dual Virtual I/O Servers, the adapter slot numbers can be alternated from
one Virtual I/O Server to the other. The first Virtual I/O Server can use
odd-numbered slots, and the second can use even-numbered slots. In a dual
servers scenario, allocate slots in pairs, with each client using two adjacent slots
such as 21 and 22, or 33 and 34.
Increase the Maximum virtual adapters value above the default value of 10 when
you create a logical partition (LPAR) profile. The appropriate number for your
environment depends on the virtual adapters slot scheme that you adopt. The
allocation needs to be balanced because each unused virtual adapter slot uses a
small amount of memory. Figure 5-2 on page 73 is an example of the Maximum
virtual adapters value in a logical partition profile.
Slot numbering scheme: The slot numbering scheme can be altered when a
Live Partition Mobility (LPM) migrates a logical partition.

Figure 5-2 Changing the default Maximum virtual adapters number
It is a best practice to use separate virtual adapter pairs for different types of
backing devices:
򐂰 For UNIX client logical partitions, do not map boot and data disks on the same
virtual adapter. As an example, for AIX, have rootvg and data volume groups
on separate virtual SCSI server adapters.
򐂰 Do not map shared storage pool logical units, logical volumes, and physical
volumes on the same virtual SCSI server adapter.
Partition profile: The maximum number of virtual adapter slots that are
available on a partition is set by the profile of the partition. It is required to
reactivate the LPAR to have the profile taken into account.

򐂰 For optimum performance and availability, do not share a vhost to map
different types of physical volumes. The max_transfer_size for storage
devices might be different. The max_transfer_size is negotiated when the
SCSI adapter of the virtual client is first configured. If a new disk is mapped to
the related vhost with a higher max_transfer_size, it cannot be configured at
the client side. It might require a reboot.
򐂰 In a SAN environment, a LUN is assigned to the Virtual I/O Server Fibre
Channel adapter. The Virtual I/O Server maps the LUN to the vhost that is
associated to a virtual SCSI client adapter. From a security perspective,
ensure that the specific client is connected to the virtual Fibre Channel server
adapter, by explicitly selecting a partition, as shown in Figure 5-3.
Figure 5-3 Select the appropriate virtual I/O client
Naming conventions
The choice of a naming convention is essential for tracking virtual to physical
relationship, especially in virtual SCSI configurations.
If you want to uniquely identify the virtual client target devices, use a convention
similar to the following definitions:
<client_lpar_name><bd_type><client_vg><hdisk number> :
client_lpar_name This field needs to be a representative subset, not too
long.
bd_type You want to include the type of backing device used:
<L> for Logical Volume

<D> for physical disk or LUN mapping
<S> for Shared Storage Pool mapping
<V> for Virtual optical devices
<T> for Tape devices
client_vg For an AIX client, you would put a subset of the VG name.
hdisk_number You can start with hd0, then, increment. The disk number
can change at the client partition. Choosing the same
number is not the best idea. It is probably better to
increment it.
Example 5-7 shows how you can use this naming convention.
Example 5-7 Example of a virtual storage mapping convention
$ mkvdev -vdev hdisk10 -vadapter vhost35 -dev Lpar1DrvgHd0
Lpar1DrvgHd0 Available
$ mkvdev -vdev hdisk11 -vadapter vhost36 -dev Lpar1DsapHd1
Lpar1DsapHd1 Available
Naming convention for external physical volumes
It is a best practice to share and plan a naming convention with the SAN
administration team. For example, commands, such as the AIX mpio_get_config
command on a DS4000 and DS5000, or svcinfo on a SAN Volume Controller
(SVC) cluster, help to identify a LUN with logical names.
Most of the storage arrays have similar commands that are meant to provide
information that is relevant and useful to have an appropriate naming convention.
Appropriate naming facilitates the administration and problem determination.
In Example 5-8, we take the benefit of the user label from a DS4800 as the
virtual target device name.
Example 5-8 DS4800 mpio_get_config command
$ oem_setup_env
# mpio_get_config -Av
Frame id 0:
Storage Subsystem worldwide name: 60ab800114b1c00004fad119f
Controller count: 2
Partition count: 1
Note: The mkvdev command currently cannot exceed 15 characters in length.

Partition 0:
Storage Subsystem Name = 'DS4800POK-3'
hdisk# LUN # Ownership User Label
hdisk6 3 A (preferred) aix02DrvgHd0
hdisk7 4 B (preferred) aix02DdvgHd1
# exit
$ mkvdev -vdev hdisk6 -vadapter vhost33 -dev aix02DrvgHd0
aix02DrvgHd0 Available
$ lsmap -vadapter vhost33
SVSA Physloc Client Partition ID
--------------- -------------------------------------------- ------------------
vhost33 U8233.E8B.061AA6P-V1-C51 0x00000005
VTD aix02rvgHd0
Status Available
LUN 0x8100000000000000
Backing device hdisk6
Physloc U5802.001.0086848-P1-C2-T1-W201600A0B829AC12-L3000000000000
Mirrored false
5.2.3 Exporting virtual Small Computer System Interface storage
This section provides best practices for exporting storage by using a virtual Small
Computer System Interface (SCSI) to virtual I/O clients.
Logical volumes
The Virtual I/O Server can export logical volumes to virtual I/O clients. This
method does have some advantages over physical volumes:
򐂰 Logical volumes can subdivide physical disk devices between different clients.
򐂰 System administrators with AIX experience are already familiar with the
Logical Volume Manager (LVM).
In this section, the term volume group refers to both volume groups and storage
pools. Also, the term logical volume refers to both logical volumes and storage
pool backing devices.
oem_setup_env: Using oem_setup_env is not a best practice unless directed
by IBM support, but is permitted for some storage configuration.

Logical volumes cannot be accessed by multiple Virtual I/O Servers concurrently.
Therefore, they cannot be used with Microsoft Multipath I/O (MPIO) on the virtual
I/O client. Multipathing needs to be managed at the Virtual I/O Server, as
explained in 5.1.2, “Multipathing” on page 64.
The following list provides best practices for logical volume mappings:
򐂰 Avoid the use of a rootvg on the Virtual I/O Server to host exported logical
volumes. Certain types of software upgrades and system restores might alter
the logical volume to target device mapping for logical volumes within rootvg,
requiring manual intervention. Also, it is easier to manage Virtual I/O Server
rootvg disk replacement when it does not have virtual I/O clients that use
logical volumes as backing devices.
򐂰 The default storage pool in the Integrated Virtualization Manager (IVM) is the
root volume group of the Virtual I/O Server. Ensure that you create and
choose different storage pools to host client backing devices.
򐂰 Although logical volumes that span multiple physical volumes are supported,
it is best if a logical volume fully resides on a single physical volume for
optimum performance and for maintenance.
򐂰 Mirror the disks in the virtual I/O client. Ensure that the logical volume backing
devices are on different physical disks. This configuration also helps with
physical disk replacements at the Virtual I/O Server.
򐂰 In dual Virtual I/O Server configurations, if one server is rebooted, ensure that
the mirroring is synced on the virtual I/O clients.
Exporting physical volumes
The Virtual I/O Server can export entire physical volumes to virtual I/O clients.
Exporting physical volumes has several advantages when compared to using
logical volumes as backing devices:
򐂰 Physical volumes can be exported from two or more Virtual I/O Servers,
providing multipathing to virtual I/O clients.
򐂰 Physical volumes might be exported with concurrent access to more than one
virtual I/O client with an appropriate concurrency management solution, such
as Concurrent LVM (CLVM) and PowerHA SystemMirror.
򐂰 The code path for exporting physical volumes is shorter, which leads to better
performance.
򐂰 Virtual I/O clients that are backed by external SAN devices can be moved
from one Virtual I/O Server to another using LPM.
򐂰 In certain cases, existing LUNs from physical servers can be migrated into the
virtual environment with the data intact. For more information, see
PowerVM Migration from Physical to Virtual Storage, SG24-7825.

Exporting physical volumes makes administration easier, since it does not
require that you manage the size by the Virtual I/O Server. The Virtual I/O Server
does not allow partitioning of a single internal physical disk among multiple
clients.
In the Virtual I/O Server, there is no need to subdivide SAN-attached disks,
because storage allocation can be managed at the storage server. In the SAN
environment, provision and allocate LUNs to Virtual I/O Server. Then, export
them to the virtual I/O clients as physical volumes.
reserve_policy
Whenever you need to map the same external volume to a virtual I/O client from
a Virtual I/O Servers configuration, change the reserve_policy, as shown in
Example 5-9. The attribute name reserve_policy might be called different by
your storage vendor.
Example 5-9 Change the reserve_policy disk attribute
$ lsdev -dev hdisk12 -attr reserve_policy
value
single_path
$ chdev -dev hdisk12 -attr reserve_policy=no_reserve
hdisk12 changed
$ lsdev -dev hdisk12 -attr reserve_policy
value
no_reserve
Small Computer System Interface queue_depth
Increasing the value of the queue_depth attribute improves the throughput of the
disk in some configurations. However, there are several other factors that must
be considered. These factors include the value of queue_depth for all of the
physical storage devices on the Virtual I/O Server being used as a virtual target
device by the disk instance on the client partition. It also includes the maximum
transfer size for the virtual Small Computer System Interface (SCSI) client
adapter instance that is the parent device for the disk instance.
Rootvg disk: Moving a rootvg disk from one physical managed system to
another is only supported by using LPM.
Cloning services: Usage of cloning services of the rootvg disk is only
supported for offline backup, restore, and recovery.

For more information about tuning the SCSI queue_depth parameter, see IBM
PowerVM Virtualization Managing and Monitoring, SG24-7590.
Exporting virtual tape and optical devices
Virtual tape devices are assigned and operated similarly to virtual optical
devices. Only one virtual I/O client can have access at a time. It is a best practice
to have such devices attached to a Virtual I/O Server, instead of moving the
physical parent adapter to a single client partition.
When internal tapes and optical devices are physically on the same controller as
the boot disks of the Virtual I/O Server, it is a best practice to map them to a
virtual host adapter. Then, use dynamic logical partitioning to assign this virtual
host adapter to a client partition.
5.2.4 Configuring the Virtual I/O client with virtual SCSI
This section covers best practices to configure the virtual I/O clients. The best
practices include disk and adapter tuning parameters and implementing disk load
balancing between dual Virtual I/O Servers to obtain the best performance.
Assign path priorities on virtual I/O clients
In a dual Virtual I/O Server configuration, with virtual I/O clients that run AIX, the
system administrator must manually configure the load balancing of the SAN
traffic.
If you have a single LUN attached to two Virtual I/O Servers, use the chpath
command to raise the priority of one virtual SCSI path over the other. This
command is required because the MPIO algorithm only supports the failover
method on the virtual I/O clients.
Prioritizing the communication paths is important to optimize and maximize the
utilization of all available Virtual I/O Servers and SAN fabric bandwidth. This
prioritization is also important to distribute the SAN traffic across all available
Virtual I/O Server paths.
Example 5-10 on page 80 illustrates how to use the lspath command to verify
the priority attribute for hdisk0 on path vscsi0. Repeat the same command for
vscsi1. If vscsi0 and vscsi1 have the same priority, the virtual I/O client always
uses the first path that is listed in the output of lspath.

Example 5-10 Listing the hdisk0 and vscsi0 path attributes
# lspath -l hdisk0
Enabled hdisk0 vscsi0
Enabled hdisk0 vscsi1
# lspath -AHE -l hdisk0 -p vscsi0
attribute value description user_settable
priority 1 Priority True
Example 5-11 shows how to set the path priority. This example sets vscsi0 to the
lowest priority path. When the setting of the path priority is completed, all new I/O
requests use vscsi1, which in this case is represented by Virtual I/O Server 2.
The chpath command does not require a reboot and the changes take effect
immediately.
Example 5-11 Changing the vscsi0 priority for hdisk0
# chpath -l hdisk0 -p vscsi0 -a priority=2
IBM i
For IBM i, best practice is to use dual or multiple Virtual I/O Servers. Multipathing
capabilities are available because of IBM i V6.1.1, which provides redundancy
across multiple Virtual I/O Servers. With IBM i V7.1TR 2 or later, the IBM i
multipathing algorithm is enhanced from round-robin to dynamic load balancing.
With this enhancement, IBM i is able to use paths to optimize resource utilization
and performance.
Virtual Small Computer System Interface tuning for AIX
When you use virtual SCSI adapters, there are some tuning parameters to
maximize the performance of the virtual adapters in a virtualized environment.
The Virtual I/O Server and virtual I/O clients contain general tuning parameters
that might need to be adjusted to obtain better performance. Before you make
changes to any attribute, consult your storage vendor documentation.
Table 5-2 on page 81 shows the recommended settings for virtual I/O clients
when you use virtual SCSI disks.
Recommended values: Remember that default values of the tuning
parameters are the recommended values in most cases.

Table 5-2 Recommended settings for virtual I/O client virtual SCSI disks
Parameter Considerations and recommendations
algorithm Recommended value: fail_over
Sets how the I/Os are balanced over multiple paths to the SAN
storage. A virtual SCSI disk only supports fail_over mode.
hcheck_cmd Recommended value: test_unit_rdy or inquiry.
Used to determine if a device is ready to transfer data. Change to
inquiry only if you have reservation locks on your disks. In other
cases, use the default value test_unit_rdy.
hcheck_interval Recommended value: 60
Interval in seconds between health check polls to the disk. By
default, the hcheck_interval is disabled. The minimum
recommended value is 60 and must be configured on both the
Virtual I/O Server and the virtual I/O client. The hcheck_interval
value should not be lower than the rw_timeout that is configured
on the Virtual I/O Server physical volume. If there was a problem
with communicating to the storage, the disk driver on the Virtual
I/O Server would not notice until the I/O requests times out
(rw_timeout value). Therefore, you would not want to send a path
health check before this time frame.
hcheck_mode Recommended value: nonactive
Determines which paths are checked when the health check
capability is used. The recommendation is to only check paths
with no active I/O.
max_transfer Recommended value: Same value as in the Virtual I/O Server.
The maximum amount of data that can be transferred to the disk
in a single I/O operation. See the documentation from your
storage vendor before you set this parameter.
queue_depth Recommended value: Same value as in the Virtual I/O Server.
The number of concurrent outstanding I/O requests that can be
queued on the disk. See the documentation from your storage
vendor before you set this parameter.
reserve_policy Recommended value: no_reserve
Provides support for applications that are able to use the SCSI-2
reserve functions.

Table 5-3 shows the recommended settings for virtual SCSI adapters in a virtual
I/O client.
Table 5-3 Recommended settings for virtual I/O client virtual SCSI adapters
5.3 Shared Storage Pools
A Shared Storage Pool (SSP) allows up to four Virtual I/O Servers. The servers
can be from the same frame or across different frames and can have a common
clustered pool of Logical Unit Numbers (LUNs). The pool of LUNs can be used to
provide storage to their virtual clients.
5.3.1 Requirements per Shared Storage Pool node
The minimum system requirements per SSP node are shown in Table 5-4.
Table 5-4 Shared Storage Pools minimum requirements per node
Parameter Considerations and recommendations
vscsi_err_recov Recommended value: delayed_fail or fast_fail
Fast I/O failure might be desirable in situations where
multipathing software is being used. The suggested value for this
attribute is fast_fail when you are using a dual or multiple Virtual
I/O Server configuration. Using fast fail, the virtual I/O client
adapter might decrease the I/O fail times because of link loss
between the storage device and switch. This value allows for a
faster failover to alternate paths. In a single path configuration,
especially configurations with a single path to a paging device,
the default delayed_fail setting is suggested.
vscsi_path_to Recommended value: 30
The virtual SCSI client adapter path timeout feature allows the
client adapter to detect whether the Virtual I/O Server is not
responding to I/O requests.
Category Requirement per SSP node
Processor 1 virtual and 1 physical CPU of entitlement
Memory 4 GB
Repository storage 1 Fibre Channel attached disk of 10 GBs in size. All
storage needs to be allocated on hardware RAIDed
storage for redundancy.

5.3.2 Shared Storage Pools specifications
The minimum and maximum options for SSPs are listed in Table 5-5.
Table 5-5 Shared Storage Pools minimums and maximums
5.3.3 When to use Shared Storage Pools
Shared Storage Pools (SSPs) provide increased flexibility for companies where
system administrators constantly need to assign new storage LUNs to virtual
clients. In this case, the SAN administrator provides a large portion of storage
that the system administrator assigns to the SSP and distributes it to the virtual
clients, as needed. This flexibility eliminates the need to contact the SAN
administrator each time a new LUN is needed.
SSPs also offer various features, such as linked clones, snapshots and roll-back,
and rapid provisioning. You can also easily perform an LPM operation within the
frames that belong to the SSP cluster.
Data storage 1 Fibre Channel attached disk of 10 GBs in size. All
storage needs to be allocated on hardware RAIDed
storage for redundancy.
Category Requirement per SSP node
Feature Minimum Maximum
Number of Virtual I/O Server Nodes in Cluster 1 4
Number of Physical Disks in Pool 1 256
Number of Virtual Disks (LUNs) Mappings in Pool 1 1024
Number of Client LPARs per Virtual I/O Server node 1 40
Capacity of Physical Disks in Pool 5 GB 4 TB
Storage Capacity of Storage Pool 10 GB 128 TB
Capacity of a Virtual Disk (LU) in Pool 1 GB 4 TB
Number of Repository Disks 1 1

It is important to remember the limitations of SSPs, though. For example, the
Virtual I/O Server software updates must be done while all clients that use
storage in the SSP are shut down. If you have service level agreements (SLAs)
where you cannot schedule downtime for your clients, consider using NPIV or
regular virtual SCSI to provide storage to your virtual clients.
5.3.4 Creating the Shared Storage Pools
Creating the Shared Storage Pools (SSPs) is done by using the cluster
command, and is covered in IBM PowerVM Virtualization Managing and
Monitoring, SG24-7590.
This is a simple process that requires only a few commands. Consider the
following best practices when you create SSPs:
򐂰 Create the cluster by using a single Virtual I/O Server, first.
򐂰 Add more Virtual I/O Servers as nodes to the cluster, one at a time.
򐂰 Assign storage to the virtual I/O clients.
Example 5-12 shows the process of creating the cluster, adding additional
nodes, and assigning storage to the virtual I/O client.
Example 5-12 Creating the cluster, adding additional nodes, and assigning storage to the virtual I/O client
$ cluster -create -clustername ClusterA -repopvs hdisk1 -spname > StorageA -sppvs
hdisk2 hdisk3 hdisk4 -hostname vios1.itso.ibm.com
Cluster ClusterA has been created successfully.
$ cluster -addnode -clustername ClusterA -hostname vios2.itso.ibm.com
Partition vios2.itso.ibm.com has been added to the ClusterA cluster.
$ mkbdsp -clustername ClusterA -sp StorageA 10G
> -bd aix01_datavg -vadapter vhost0
Lu Name:aix01_datavg
Lu Udid:84061064c7c3b25e2b4404568c2fcbf0
Assigning file "aix01_datavg" as a backing device.
VTD:vtscsi0

5.3.5 SAN storage considerations
First, plan which disks are used for the cluster repository, and which are used for
the virtual I/O clients. Remember that you need a minimum of one for data and
one for the repository. Then, verify that the SAN disks are present across each
Virtual I/O Server that participates in the cluster.
Before a disk can be used for either a repository or storage pool disk, it is a good
idea to run the prepdev command to determine whether the disk is already
assigned or in use. Example 5-13 shows the prepdev command that is being run
against a disk that already belongs to an SSP cluster, and the output of the
recommendations before proceeding.
Example 5-13 The prepdev command to prepare the disk
$ prepdev -dev hdisk10
WARNING!
The VIOS has detected that this physical volume is
currently in use. Data will be lost and cannot be
undone when destructive actions are taken.
These actions should only be done after confirming
that the current physical volume usage and data are
no longer needed.
The VIOS detected that this device is a cluster disk.
Destructive action: Remove the physical volume from
the cluster by running the following command:
cleandisk -s hdisk#
Verifying a disk belongs to a Shared Storage Pool
Disks which make up the Shared Storage Pool (SSP) repository are placed into a
volume group named caavg_private. However, disks which are assigned to the
storage pool for virtual I/O client use, when viewed by using the lspv command,
look to be free.
Viewing the cluster configuration with the lscluster -c command shows the
disks that are assigned to the storage pool. Example 5-14 on page 86 shows the
disks which are in the storage pool.

Example 5-14 Highlighted disks which are in the storage pool
$ lspv
NAME PVID VG STATUS
hdisk0 00f61ab26fed4d32 rootvg active
hdisk1 00f61aa6c31760cc caavg_private active
hdisk2 00f61aa6c31771da None
hdisk3 00f61aa6c3178248 None
hdisk4 00f61aa6c31794bf None
$ lscluster -c | grep disk
Number of disks in cluster = 3
for disk hdisk4 UUID = 4aeefba9-22bc-cc8a-c821-fe11d01a5db1 cluster_major =
0 cluster_minor = 3
for disk hdisk2 UUID = c1d55698-b7c4-b7b6-6489-c6f5c203fdad cluster_major =
0 cluster_minor = 2
for disk hdisk3 UUID = 3b9e4678-45a2-a615-b8eb-853fd0edd715 cluster_major =
0 cluster_minor = 1
Thin or thick provisioning
Thin provisioning is a method of allocating more disk space than is physically
available. Unlike thick provisioning, which allocates all the blocks up front, thin
provisioning blocks are only allocated as they are written.
Best practice is to allocate storage to virtual I/O clients by using thick
provisioning. However, this allocation is dependent on having sufficient capacity
in the storage pool. If thin provisioning is being used, ensure that you have a
reliable method of monitoring the storage utilization. Best practice is to configure
threshold warnings by using the alert command, as shown in 5.3.6, “Monitoring
storage pool capacity” on page 87.
Other storage considerations
Be aware of the following other storage considerations:
򐂰 Uninterrupted access to the repository disk is required for operation.
򐂰 Physical disks in the SAN storage subsystem that are assigned to the SSD,
cannot be resized.
򐂰 Virtual SCSI devices that are provisioned from the SSP might drive higher
processor utilization than the classic virtual SCSI devices.
The lspv commands: Using the lspv -free or lspv -avail commands does
not display the storage pool disks.

򐂰 Using SCSI reservations (SCSI Reserve/Release and SCSI-3 Reserve) for
fencing physical disks in the SSP, is not supported.
򐂰 High availability SAN solutions can be used to mitigate outages.
5.3.6 Monitoring storage pool capacity
Over committing the storage pool is a risk. You can monitor the free percentage
threshold by using the alert command. A threshold of 35% is configured by
default, and when triggered, will log in an entry to the error log. The default
threshold is sufficient, but evaluate and configure as appropriate for your
environment.
Example 5-15 shows how to list and change the threshold and the error log entry
when the threshold is triggered.
Example 5-15 Using the alert command to configure threshold warnings
$ alert -list -clustername ClusterA -spname StorageA
PoolName: StorageA
PoolID: FFFFFFFFAC10161E000000004FCFA454
ThresholdPercent: 35
$ alert -set -clustername ClusterA -spname StorageA -type threshold
-value 25
$ alert -list -clustername ClusterA -spname StorageA
PoolName: StorageA
PoolID: FFFFFFFFAC10161E000000004FCFA454
ThresholdPercent: 25
$ errlog
IDENTIFIER TIMESTAMP T C RESOURCE_NAME DESCRIPTION
0FD4CF1A 0606145212 I O VIOD_POOL Informational Message
5.3.7 Network considerations
The most important thing to remember about the network is that the hostname/IP
address that is used to create the storage pool resolves to the fully qualified
domain name. As an example, if you run the hostname command, it reports the
system hostname as in mydivision.mycompany.com. If the hostname resolves only
to the short name, then the CLMD command will fail.
Available space: The ThresholdPercent value is the percentage of free
space that is available in the storage pool.

Other considerations
Remember the following additional networking considerations:
򐂰 Uninterrupted network connectivity is required for operation. That is, the
network interface that is used for SSP configuration must be on a highly
reliable network, which is not congested.
򐂰 Changing the hostname/IP address for a system is not supported when it is
configured in an SSP.
򐂰 Only compliant with Internet Protocol Version 4 (IPv4).
򐂰 When a Virtual I/O Server is configured for an SSP environment, VLAN
tagging is not supported. For workaround, see section 2.6.6, “Managing
VLAN tagging” on page 57, in the IBM Redbooks publication, IBM PowerVM
Virtualization Managing and Monitoring, SG24-7590.
򐂰 An SSP configuration configures the TCP/IP resolver routine for name
resolution to resolve host names locally first, and then to use the Domain
Name System (DNS). For step-by-step instructions, see the TCP/IP name
resolution documentation in the AIX Information Center:
http://publib.boulder.ibm.com/infocenter/pseries/v5r3/index.jsp
򐂰 When you restore the Virtual I/O Server LPAR configuration from a viosbr
backup, all network devices and configurations must be restored before SSP
configurations are restored.
򐂰 The forward and reverse lookup resolves to the IP address/hostname that is
used for the SSP configuration.
򐂰 It is suggested that the Virtual I/O Servers that are part of the SSP
configuration, keep their clocks synchronized.
5.4 N-Port ID Virtualization
This section covers the benefits and best practices for choosing
N-Port ID Virtualization (NPIV) as the method for virtual I/O clients to access
SAN storage.
For more information about NPIV, see IBM PowerVM Virtualization Managing
and Monitoring, SG24-7590.
5.4.1 When to use N-Port ID Virtualization
N-Port ID Virtualization (NPIV) is now the preferred method of providing virtual
storage to virtual I/O clients, whenever a SAN infrastructure is available. The

main advantage for selecting NPIV, compared to a virtual SCSI, is that the
Virtual I/O Server is only used as a pass through to the virtual I/O client virtual
Fibre Channel adapters. Therefore, the storage is mapped directly to the virtual
I/O client, with the storage allocation managed in the SAN. This strategy
simplifies storage mapping at the Virtual I/O Server.
Consider the following additional benefits of NPIV:
򐂰 Provides storage resources to the virtual I/O client, just as they would be with
actual physical Fibre Channel adapters.
򐂰 Multipathing takes place at the virtual I/O client.
򐂰 Allows virtual I/O clients to handle persistent reservations, which are useful in
high availability cluster solutions, such as PowerHA SystemMirror.
򐂰 Makes Virtual I/O Server maintenance easier because there is no need for
multipathing software.
򐂰 Is a preferred method for LPM operations.
򐂰 Using NPIV on virtual clients that run IBM i is a best practice, primarily
because of performance reasons. The amount of work that the Virtual I/O
Server needs to do with NPIV is less involved than with virtual SCSI.
5.4.2 Configuring the Virtual I/O Server with N-Port ID Virtualization
There are many aspects to consider when you configure the Virtual I/O Server
with N-Port ID Virtualization (NPIV). Aspects to consider include LUN naming
conventions, SAN infrastructure, redundancy at the Virtual I/O Server and virtual
I/O client, and Fibre Channel adapter attributes.
Although the naming convention for NPIV is not as important as it is for virtual
SCSI, you need to pay attention to the same naming considerations for LUNs. In
this case, the naming considerations are likely to be the same as for regular
non-virtualized logical partitions. An example of LUN naming in provided in,
“Naming convention for external physical volumes” on page 75.
The most important thing for NPIV is to track the relationship between the virtual
and physical Fibre Channel adapters. Document and maintain this relationship in
the spreadsheet that you created during the planning session, described in 1.3,
“Planning your Virtual I/O Server environment” on page 6.
Worldwide port names: If you plan to use LPM, remember that both the
active and inactive worldwide port names (WWPNs) of a virtual Fibre Channel
adapter need to be zoned from the switches. The names also need to be
mapped from the storage array.

Follow these best practices for better availability, performance, and redundancy:
򐂰 Use NPIV in dual or multiple Virtual I/O Server configurations.
򐂰 Have Virtual I/O Server physical Fibre Channel adapters that split across
separate system PCI busses, when possible.
򐂰 Operate at the highest possible speed on the SAN switches.
򐂰 Have the physical Fibre Channel adapter ports connected to different
switches in the fabric, even in dual Virtual I/O Server configurations.
򐂰 Although it is supported to connect storage and tape libraries through NPIV
by using the same Fibre Channel adapter, it is a best practice to separate
them among different adapters. Separation is suggested because the disk
and tape traffic have different performance characteristics and error recovery
scenarios.
Configuring redundancy with N-Port ID Virtualization
As in any SAN configuration, redundancy is crucial. Therefore, avoid single
points of failure. Multipathing with a dual Virtual I/O Server configuration provides
the level of redundancy that is required, better performance, and flexibility for the
maintenance of the Virtual I/O Servers.
Figure 5-4 on page 91 shows a dual Virtual I/O Server configuration. In this
setup, virtual I/O clients avoid the loss of access to SAN storage in the event of a
SAN switch, Virtual I/O Server, or physical link failure.
Each virtual I/O client has a virtual Fibre Channel adapter that is mapped to each
Virtual I/O Server. Each Virtual I/O Server has a redundant link to each SAN
switch.
Figure 5-4 on page 91 shows an NPIV configuration with dual

Figure 5-4 NPIV configuration with dual Virtual I/O Servers
5.4.3 Configuring the virtual I/O client with N-Port ID Virtualization
The level of configuration that is required for N-Port ID Virtualization (NPIV) in the
virtual I/O client can be compared to any non-virtual logical partition. The current
versions of AIX set the values of the fc_err_recov to fast_fail, by default.
Confirm these values are set as shown in Example 5-16 on page 92. Change this
value by using the chdev command, if needed. Dynamic tracking, dyntrk, is
always enabled for virtual Fibre Channel client adapters and cannot be disabled.
Dynamic tracking is enabled for the express purpose of recovering from
Client 2
VIO B
SAN Switch SAN Switch
Client 1
Storage
VIO A
NPIV Server adapter
NPIV Client adapter
Switch port
Physical Fibre adapter

reconfiguration events in the SAN. These events are likely to occur during LPM
operations, and it is a good idea for dyntrk to always be set to yes.
Example 5-16 shows how to change values on a Fibre Channel adapter.
Example 5-16 Changing attributes on a Fibre Channel adapter on a virtual I/O client
# lsattr -El fcs0
intr_priority 3 Interrupt priority False
lg_term_dma 0x800000 Long term DMA True
max_xfer_size 0x100000 Maximum Transfer Size True
num_cmd_elems 200 Maximum Number of COMMAND Elements True
sw_fc_class 2 FC Class for Fabric True
# lsattr -El fscsi0
attach none How this adapter is CONNECTED False
dyntrk yes Dynamic Tracking of FC Devices True
fc_err_recov fast_fail FC Fabric Event Error RECOVERY Policy True
scsi_id Adapter SCSI ID False
sw_fc_class 3 FC Class for Fabric True
Adopt the following best practices for virtual I/O clients that use NPIV:
򐂰 Do not change the NPIV num_cmd_elem and max_xfer_size values to be
higher than the Virtual I/O Server physical adapter values. The virtual I/O
client might not be able to configure new devices and might fail to boot.
򐂰 Configure a reasonable number of paths, such as 4 - 8. An excessive number
of paths can increase error recovery, boot time, and the time it takes to run
cfgmgr. This configuration can be done by zoning on SAN switches and by
using LUN masking at the storage arrays.
򐂰 Balance virtual I/O client workloads across multiple physical adapters in the
Virtual I/O Server.
򐂰 To avoid boot issues, balance the paths to the booting devices with bosboot,
as documented at the following website:
http://www.ibm.com/support/docview.wss?uid=isg3T1012688
NPIV support: Most of the storage vendors support NPIV. Check what their
requirements are to support the operating system of your virtual I/O client.
Adding a virtual Fibre Channel adapter: If you need to add a virtual
Fibre Channel (FC) adapter with a dynamic logical partition operation, save
the current logical partition configuration afterward. For more information, see
section 3.2, “Dynamic logical partition operations” on page 42.

Chapter 6. Performance monitoring
Monitoring the virtual I/O environment is essential. This process includes
monitoring the critical parts of the operating system that are running on the
virtual I/O clients, such as memory, processors, network, and storage. This
chapter highlights best practices for monitoring the virtual environment.
6

6.1 Measuring Virtual I/O Server performance
Measuring the performance of a Virtual I/O Server can be generally divided into
categories:
򐂰 Short-term measuring is used in testing, sizing, or troubleshooting scenarios.
򐂰 Long-term measuring is used as capacity management input because it
includes performance trends or changes in workload.
6.1.1 Measuring short-term performance
On the Virtual I/O Server, use the following tools for short-term performance
measurements:
viostat Reports processor statistics, and I/O statistics for the
entire system, adapters, tty devices, disks, and optical
devices.
netstat Reports network statistics.
vmstat Reports statistics about kernel threads, virtual memory,
disks, traps, and processor activity.
topas Reports selected local system statistics, such as
processor, network, processes, and memory.
seastat Reports Shared Ethernet Adapter (SEA) statistics, and
generates a report for each client.
svmon Reports a snapshot of virtual memory.
fcstat Reports the statistics that are gathered by the specific
Fibre Channel device driver.
If you start measuring short-term performance on the Virtual I/O Server and you
do not have a specific target, such as network degradation, start with the topas
command. The topas command displays local system statistics, such as system
resources and Virtual I/O Server SEA statistics.
The viostat, netstat, vmstat, svmon, tseastat, and fcstat commands, provide
more detailed output information than topas. Document the output of these
commands and the time that they were run because it is valuable information if
you need to research a performance bottleneck.
In Virtual I/O Server 2.1, the nmon functionality is integrated within the topas
command. You can start the topas command and switch between the two modes
by typing ~ (tilde).

Chapter 6. Performance monitoring 95
Example 6-1 shows the main functions that are used in the topas_nmon
command. You can export the environment variable nmon to start topas_nmon the
same way every time.
Example 6-1 Display of nmon interactive mode commands
•h = Help information q = Quit nmon 0 = reset peak counts •
•+ = double refresh time - = half refresh r = ResourcesCPU/HW/MHz/AIX•
•c = CPU by processor C=upto 128 CPUs p = LPAR Stats (if LPAR) •
•l = CPU avg longer term k = Kernel Internal # = PhysicalCPU if SPLPAR •
•m = Memory & Paging M = Multiple Page Sizes P = Paging Space •
•d = DiskI/O Graphs D = DiskIO +Service times o = Disks %Busy Map •
•a = Disk Adapter e = ESS vpath stats V = Volume Group stats •
•^ = FC Adapter (fcstat) O = VIOS SEA (entstat) v = Verbose=OK/Warn/Danger •
•n = Network stats N=NFS stats (NN for v4) j = JFS Usage stats •
•A = Async I/O Servers w = see AIX wait procs "="= Net/Disk KB<-->MB •
•b = black&white mode g = User-Defined-Disk-Groups (see cmdline -g) •
•t = Top-Process ---> 1=basic 2=CPU-Use 3=CPU(default) 4=Size 5=Disk-I/O •
•u = Top+cmd arguments U = Top+WLM Classes . = only busy disks & procs•
•W = WLM Section S = WLM SubClasses •
•[ = Start ODR ] = Stop ODR •
•~ = Switch to topas screen
Example 6-2 shows the nmon variable that is defined to start topas_nmon with
options to monitor processor, memory, paging, disks, and service times.
Example 6-2 Exporting environment variable in ksh shell to nmon
$ export NMON=cmD
$ topas_nmon
Monitoring processor and multiple Shared Processor Pools
To monitor the utilization of multiple Shared Processor Pools (SPPs) in a Virtual
I/O Server, enable the partition profile to collect performance information. To
enable the collection of cross partition processor performance data, use the
Hardware Management Console (HMC) to open the partition profile properties.
Then choose the hardware tab and select the check box for Allow performance
information collection.

After the activation of the performance information collection, you can use the
topas -fullscreen lpar command in the Virtual I/O Server to determine
whether the processor resources are optimally used. Two important fields in the
command output, Psize and app, are described here:
Psize Shows the number of online physical processors in the
shared pool.
app Shows the available physical processors in the shared
pool.
Using the same command, you can see statistics for all logical processors
(LCPU) in the system, such as Power Hypervisor calls, context switching, and
processes waiting to run in the queue.
On an AIX virtual I/O client, you can use the lparstat command to report logical
partition-related information, statistics, and Power Hypervisor information.
On an IBM i virtual I/O client, data is collected by the collection services in the
QAPMLPARH table. The data can be displayed either as text through SQL, or
graphically through IBM System Director Navigator for i.
For more information, see IBM PowerVM Virtualization Managing and
Monitoring, SG24-7590.
Disk monitoring
A good start to monitor the disk activity in the Virtual I/O Server is to use the
viostat command.
You can get more specialized output from the viostat command. Example 6-3
on page 97 shows monitoring of vhost1 and hdisk4. It is not recommended to
use this output for long-term measuring because it requires a significant amount
of disk space to store the data.
Example 6-3 on page 97 shows the extended disk output by using the
viostat command.

Example 6-3 Extended disk output of hdisk4 and vhost1 using viostat
$ viostat -adapter 1 1 | grep -p vhost1 | head -2 ; viostat -extdisk hdisk4 1 1
Vadapter: Kbps tps bkread bkwrtn
vhost1 88064.0 688.0 344.0 344.0
System configuration: lcpu=16 drives=12 paths=17 vdisks=41
hdisk4 xfer: %tm_act bps tps bread bwrtn
100.0 90.2M 344.0 0.0 90.2M
read: rps avgserv minserv maxserv timeouts fails
0.0 0.0 0.0 0.0 0 0
write: wps avgserv minserv maxserv timeouts fails
344.0 7.1 5.3 13.4 0 0
queue: avgtime mintime maxtime avgwqsz avgsqsz sqfull
0.0 0.0 0.0 0.0 2.0 0.0
6.1.2 Network and Shared Ethernet Adapter monitoring
To monitor network and Shared Ethernet Adapters (SEAs), use the entstat,
netstat, and seastat commands.
netstat command
The Virtual I/O Server netstat command provides performance data and also
provides network information, such as routing table and network data. To show
performance-related data, use the command with an interval, as shown in
Example 6-4. Note, that you must stop the command with a Ctrl+C. This setting
makes it more difficult to use the netstat command in a long-term measuring
script because you specify only an interval, and not a count.
Example 6-4 Output from the netstat command
$ netstat 1
input (en11) output input (Total) output
packets errs packets errs colls packets errs packets errs colls
43424006 0 66342 0 0 43594413 0 236749 0 0
142 0 3 0 0 142 0 3 0 0
131 0 1 0 0 131 0 1 0 0
145 0 1 0 0 145 0 1 0 0
143 0 1 0 0 143 0 1 0 0
139 0 1 0 0 139 0 1 0 0
137 0 1 0 0 137 0 1 0 0

entstat command
The entstat command displays the statistics that are gathered by the specified
Ethernet device driver. Use the -all flag to display all the statistics, including the
device-specific statistics.
On the Virtual I/O Server, use the entstat command to check the status and
priority of the SEA. Example 6-5 shows which SEA has the highest priority. This
example uses a dual Virtual I/O Server configuration with SEAs that use the
failover mode.
Example 6-5 Output of entstat on SEA
$ entstat -all ent9 | grep -i priority
Priority: 1
Priority: 1 Active: True
Priority: 2 Active: True
seastat command
By using the seastat command, you can generate a report for each client to view
the SEA statistics. Before you use the seastat command, enable accounting on
the SEA. Example 6-6 demonstrates how to enable this option.
Example 6-6 Enabling accounting on the Shared Ethernet Adapter
$ lsdev -dev ent11 -attr | grep accounting
accounting disabled Enable per-client accounting of network statistics True
$ chdev -dev ent11 -attr accounting=enabled
ent11 changed
With accounting enabled, the SEA tracks the MAC address of all the packets it
receives from the virtual I/O clients, increment packets, and byte counts for each
virtual I/O client, independently. Example 6-7 shows an output of the packets
received and transmitted in a SEA, filtered by IP address.
Example 6-7 seastat statistics that are filtered by IP address
$ seastat -d ent9 -s ip=172.16.22.34
=======================================================================
Advanced Statistics for SEA
Device Name: ent9
=======================================================================
MAC: 22:5C:2B:95:67:04
----------------------
VLAN: None
VLAN Priority: None

IP: 172.16.22.34
Transmit Statistics: Receive Statistics:
-------------------- -------------------
Packets: 29125 Packets: 2946870
Bytes: 3745941 Bytes: 184772445
=======================================================================
Fibre Channel monitoring
The fcstat command provides information about the adapter queue and
resource use. This command is useful to determine whether modifying
max_xfer_size or num_cmd_elems can increase performance. Example 6-8 shows
some typical values of a Fibre Channel adapter.
Example 6-8 Example use of the fcstat command
$ fcstat fcs0 | grep -E "Count|Information"
LIP Count: 0
NOS Count: 0
Link Failure Count: 3
Loss of Sync Count: 261
Primitive Seq Protocol Error Count: 0
Invalid Tx Word Count: 190
Invalid CRC Count: 0
IP over FC Adapter Driver Information
FC SCSI Adapter Driver Information
No Command Resource Count: 0
Example 6-8 shows an example of an adapter that has sufficient values for
max_xfer_size and num_cmd_elems. Non zero values indicate that I/Os are being
queued at the adapter because of a lack of resources. “Fibre Channel device
driver attributes” on page 68 has some considerations and tuning
recommendations about the max_xfer_size and num_cmd_elems parameters.
Memory monitoring
A best practice to monitor the memory consumption in a Virtual I/O Server is to
use the vmstat and svmon commands.

The vmstat command
The vmstat command displays statistics about kernel threads, virtual memory,
disks, traps, and processor activity. Regarding memory, at a high level, track the
active virtual memory (AVM) column under memory, and page in (PI), and page
out (PO) under the page column.
AVM is a metric that reflects virtual memory usage and is represented in 4K
pages. You can convert from 4K pages to megabytes dividing the AVM value by
256. If AVM is higher than the amount of real memory, this metric can cause your
Virtual I/O Server to page, which can be confirmed under the PI and PO columns.
This usage is considered suboptimal for performance. We suggest a 95%
threshold for AVM of the total amount of physical memory in the Virtual I/O Server.
The svmon command
The svmon command provides more in-depth memory information than the
vmstat command. The displayed information does not constitute a true snapshot
of memory because the padmin user does not have sufficient privileges.
However, if role-based access control (RBAC) is activated, and the
vios.system.stat.memory role is assigned to the padmin user, the user can see
the same view that the root user sees.
Enable connection monitoring
In the partition profile for the Virtual I/O Servers and each virtual I/O client on the
HMC, it is suggested to check the option for Enable Connection Monitoring.
When connection monitoring is enabled, the service IBM Focal Point™ (FP)
periodically tests the communication channel between the logical partitions and
the HMC. If the channel fails, the service FP generates a serviceable event in the
service FP log on the HMC. This step ensures that the communication channel
can carry service requests from the logical partition to the HMC when needed.
If this option is not selected, the service FP still collects service request
information when there are issues on the managed frame. This option only
controls whether the service FP automatically tests the connection and
generates a serviceable event if the channel fails.
For more information, see the Hardware Management Console V7 Handbook,
SG24-7491.
6.1.3 Measuring long-term performance
In previous Virtual I/O Server versions, the Workload Manager-based commands
were commonly used to measure long-term performance, but topasrec and
topas_nmon commands are now a best practice for recordings.

The topasrec command generates binary reports of local recordings, Central
Electronic Complex (CEC) recordings, and cluster recordings. In the Virtual I/O
Server, persistent local recordings are stored in the /home/ios/perf/topas
directory by default.
You can verify if the topasrec command is running by using the ps command.
The output data is collected in a binary file, in the format hostname.yymmdd, for
example, localhost_120524.topas. On an AIX server, this file can be converted
to a nmon analyzer report by using the topasout command.
Another way to generate records is by using the nmon spreadsheet output
format. You can start collecting performance data in the nmon format from the
cfgassist menu by selecting Performance  Topas  Start New
Recording  Start Persistent local recording  nmon. The reports can be
customized according to your needs. By default, the files are stored in the
directory /home/ios/perf/topas with file extension nmon. For an example of a
recording in nmon format, see Example 6-9.
You must decide which approach is better in your environment. Best practice is to
use only one format, binary or nmon, and to not increase the processor
utilization. You can use the cfgassist menu by selecting Performance 
Topas  Stop Persistent Recording to stop an unwanted recording, and to
remove entries in the /etc/inittab file.
Example 6-9 shows an example of a recording in the nmon format.
Example 6-9 Example of a recording in the nmon format
.
. (Lines omitted for clarity)
.
ZZZZ,T0002,00:12:39,23-May-2012,,
CPU_ALL,T0002,0.12,0.43,0.00,99.45,4.00,
CPU03,T0002,0.00,7.56,0.00,92.44,
CPU02,T0002,0.00,7.14,0.00,92.86,
CPU01,T0002,0.16,10.17,0.00,89.67,
CPU00,T0002,19.81,63.11,0.01,17.08,
DISKBUSY,T0002,0.00,0.00,0.00,0.00,
DISKREAD,T0002,0.00,0.00,0.00,0.00,
DISKWRITE,T0002,0.00,0.00,4.93,0.00,
DISKXFER,T0002,0.00,0.00,0.59,0.00,
DISKSERV,T0002,0.00,0.00,0.00,0.00,
DISKWAIT,T0002,0.00,0.00,0.00,0.00,
MEMREAL,T0002,524288.00,226570.59,26.08,46.18,10.60,10.60,
MEMVIRT,T0002,393216.00,99.48,0.52,0.00,0.06,0.00,0.00,0.06,209.38,0.00,
PROC,T0002,0.00,1.00,454.05,375.93,37.79,3.49,1.14,1.14,0.40,0.00,99.00,

LAN,T0002,0.00,0.00,0.00,0.00,0.00,,0.00,0.00,0.00,0.00,0.00,0.00,,,0.00,0.00,0.00,0.
00,0.00,,0.00,0.00,0.00,0.00,0.00,0.00,,
,0.00,0.00,0.00,0.00,0.00,,0.00,0.00,0.00,0.00,0.00,0.00,,,0.00,0.00,0.00,0.00,0.00,,
0.00,0.00,0.00,0.00,0.00,0.00,,,0.00,0.0
0,0.00,0.00,0.00,,0.00,0.00,0.00,0.00,0.00,0.00,,,
IP,T0002,0.54,0.54,0.03,0.03,
TCPUDP,T0002,0.50,0.50,0.09,0.17,
FILE,T0002,0.00,23.94,0.00,112817.25,1125.65,
JFSFILE,T0002,36.61,0.08,81.23,39.24,0.27,80.93,
JFSINODE,T0002,11.46,0.00,26.58,6.43,0.01,21.06,
LPAR,T0002,0.01,1.00,4.00,16.00,1.00,128.00,0.00,0.07,0.07,1.00,
ZZZZ,T0003,00:17:39,23-May-2012,,
.
. (Lines omitted for clarity)
.
You can also use commands such as viostat, netstat, vmstat, svmon, fcstat,
and seastat as long-term performance tools on the Virtual I/O Server. They can
be used in a script, or started by using the crontab command to form customized
reports.
Agents for long-term performance monitoring
Long-term performance of the Virtual I/O Server can be monitored and recorded
by using other monitoring agents. In Example 6-10, the lssvc command lists
which agents are supported. The cfgsvc, startsvc, cstopsvc, and
postprocesssvc commands are used to manage all the available agents.
Example 6-10 Listing of the available agents
$ lssvc
ITM_premium
ITM_cec
TSM_base
ITUAM_base
TPC_data
TPC_fabric
DIRECTOR_agent
perfmgr
ipsec_tunnel
ILMT
For more information about the monitoring tools, see this website:
http://www.ibm.com/developerworks/wikis/display/WikiPtype/VIOS_Monitoring

HMC Utilization Data
HMC utilization data can record information about the memory and processor
utilization on a managed frame at a particular time. You need to enable the HMC
utilization data individually on each managed frame.
The data is collected into records called events. Events are created at the
following times:
򐂰 At periodic intervals (hourly, daily, and monthly).
򐂰 When you make frame, logical partition, and configuration changes that affect
resource utilization.
򐂰 When you start, shut down, and change the local time on the HMC.
HMC utilization data collection is deactivated by default, but can be enabled by
selecting a managed frame on the HMC, and selecting Operations 
Utilization Data  Change Sampling Rate. We suggest a sampling rate of
5 minutes.
You can use this data to analyze trends, and make resource adjustments. To
display the utilization data with the HMC graphical user interface (GUI), select a
managed frame, and then select Operations  Utilization Data  View.
On the HMC command line, you can use the lslparutil command to list the
utilization data that is collected from a managed frame. You can create a script to
collect the utilization data from the HMC, and export it to a comma-separated file.
Example 6-11 shows an example of how to create a simple filter to list capped
cycles, uncapped cycles, entitled cycles, and time cycles in a comma-separated
format.
Example 6-11 The lslparutil command
hscroot@hmc9:~> lslparutil -m POWER7_1-SN061AA6P -r lpar --filter
> "lpar_names=vios1a" -n 2 -F time,lpar_id,capped_cycles,
> uncapped_cycles,entitled_cycles,time_cycles
06/08/2012
14:21:13,1,2426637226308,354316260938,58339456586721,58461063017452
06/08/2012
14:20:13,1,2425557392352,354159306792,58308640647690,58430247078431
Example 6-12 on page 104 shows how to calculate the processor utilization for
the shared processor partition, 1 minute apart. The sampling rate is a 1-minute
interval that is based on the data that is collected from Example 6-11.

Example 6-12 How to calculate processor utilization
Processor utilization % =
((capped_cycles + uncapped_cycles) / entitled_cycles) * 100
Processor utilization % = (((2426637226308 - 2425557392352) +
(354316260938 - 354159306792)) / (58339456586721 - 58308640647690)) *
100
Processor utilization % = 4.01%
Processor units utilized = (capped_cycles + uncapped_cycles) /
time_cycles
Processor units utilized = ((2426637226308 - 2425557392352) +
(354316260938 - 354159306792)) / (58461063017452 - 58430247078431)
Processor units utilized = 0.04
Virtual I/O Server Advisor
The Virtual I/O Server Advisor polls and collects key performance metrics. It
analyzes the results, provides a health check report, and then proposes changes
to the environment or suggests areas to investigate further.
Best practice is to run the vios_advisor command for a 24 hour period after the
Virtual I/O Server is configured and has an active workload that runs through it.
This practice ensures that the Virtual I/O Server has sufficient resources and that
those resources are correctly configured to handle requests during all peak
periods.
The Virtual I/O Server Advisor is a utility that is available at no charge. The tool
can be downloaded from this website:
http://www.ibm.com/developerworks/wikis/display/WikiPtype/VIOS+Advisor

Chapter 7. Security and advanced
IBM PowerVM features
This chapter provides best practices for security on your virtual environment. It
also covers best practices on the advanced features that are in the
IBM PowerVM Enterprise Edition, which includes Active Memory Sharing (AMS)
and Live Partition Mobility (LPM).
7

7.1 Virtual I/O Server security
Security is vital for all businesses today, especially because most companies are
moving from physical to virtual environments. This move eliminates physical
barriers and adopts new computing models, such as cloud computing, where
resources are shared among multiple tenants.
Section 7.1 describes best practices for security on your virtual environment.
7.1.1 IBM PowerVM Hypervisor security
The IBM PowerVM Hypervisor is secure by design, and does not have a single
security vulnerability to date. It provides isolation among the dedicated and
virtual partitions.
Your data in a virtual client is safe. You can share hardware resources, such as
processor and memory usage, among these virtual partitions. And, you can be
confident that the Power Hypervisor handles the separation of resources in a
secured way.
To date, there are no reported security vulnerabilities on the PowerVM
Hypervisor. However, as a best practice, ensure that your system is up-to-date
on the latest firmware to incorporate any new functions, or to fix any issues.
7.1.2 Virtual I/O Server network services
After installation, the Virtual I/O Server has some services which are open and
running, by default. The services in the listening open state are listed in
Table 7-1.
Table 7-1 Default open ports on the Virtual I/O Server
In most cases, the Secure Shell (SSH) service for remote login and the Secure
Copy Protocol (SCP) for copying files is sufficient for login and file transfer. Telnet
Port number Service Purpose
21 FTP Unencrypted file transfer
22 SSH Secure Shell and file transfer
23 Telnet Unencrypted remote login
111 rpcbind NFS connection
657 RMC RMC connection (used for dynamic LPAR operations)

Chapter 7. Security and advanced IBM PowerVM features 107
and File Transfer Protocol (FTP) are not using encrypted communication and can
be disabled.
Port 657 for Resource Monitoring and Control (RMC) must be left open if you are
considering the use of dynamic logical partition operations. This port is used for
the communication between the logical partition and the Hardware Management
Console (HMC).
The stopping of these services can be done by using the stopnetsvc command.
7.1.3 Viosecure command
The viosecure command allows for the configuration of security hardening and
firewall rules on the Virtual I/O Server. Firewall rules, in most cases, are
environment-specific, and are out of the scope of this document. For more
information about how to use viosecure to configure the firewall, see IBM
PowerVM Virtualization Managing and Monitoring, SG24-7590
Security hardening rules
The viosecure command can also be used to configure security hardening rules.
Users can enforce either the preconfigured security levels or choose to
customize them, based on their requirements.
Best practice when you use viosecure is to export all rules to an XML file format
and manually modify this file based on the security rule requirements. The
command, as shown in Example 7-1, exports all rules which apply to the level
high, to a file named viosecure.high.xml.
Example 7-1 Exporting viosecure high-level rules to XML
$ viosecure -level high -outfile viosecure.high.xml
The viosecure command uses the same rule set definitions as aixpert on
IBM AIX. For a full list of rules and applicable values, see this website:
http://www.ibm.com/developerworks/wikis/display/WikiPtype/aixpert
When you finish customizing the XML file, copy it back to the Virtual I/O Server
and apply it using the command, as shown in Example 7-2 on page 108.
iviosecure command rules: The full list of viosecure rules is large. To help
editing, we suggest copying the file to your desktop, and using a text editor
with XML support.

Example 7-2 Applying customized viosecure rules
$ viosecure -file viosecure.high.xml
7.2 IBM PowerSC
IBM PowerSC is a separate licensed product that can assist you in the security
and compliance of your virtual environment. PowerSC enables security
compliance automation and includes reporting for compliance measurement and
audit.
Your business might need to adhere to specific regulatory requirements or
industry data security standards, such as the Payment Card Industry (PCI)
standard. If so, it is much easier for you to meet these requirements and stay
compliant by using PowerSC in your virtual environment.
Table 7-2 lists the current PowerSC features and their benefits. Bookmark the
following link to stay current with the new PowerSC features:
http://ibm.com/systems/power/software/security/features.html
Table 7-2 IBM PowerSC security standards
viosecure command: We suggest running the viosecure command from a
console session because some rules might result in the loss of access
through the network connections.
padmin password: The maxage and maxexpired stanzas, when applied, might
result in the padmin account being disabled because of the aging of the
password. We suggest changing the padmin password before you run the
viosecure command to prevent the account from becoming disabled.
Feature Benefits
Security and
compliance
automation
Reduces administration costs for complying with industry
security standards.
Compliance reports Reduces time and cost to provide security and compliance
reports to auditors.
Preconfigured profiles
for PCI, DOD STIG,
and COBIT
Saves time, cost, and risk that is associated with deploying
industry security standards.

7.3 Live Partition Mobility
Live Partition Mobility (LPM) is a feature of the PowerVM Enterprise Edition that
can move live AIX, Linux, and IBM i logical partitions from one frame to another,
without any downtime. It can also be used to move inactive logical partitions. The
mobility process transfers the virtual I/O client partition environment, including
the processor state, memory, attached virtual devices, and connected users.
This topic describes the best practices when you implement and manage an
LPM environment.
Trusted Boot Reduces risk of compromised security by guaranteeing that
an AIX operating system image is not inadvertently or
maliciously altered.
Trusted Monitoring Ensures high levels of trust by displaying the status of all AIX
systems that participate in a trusted system configuration.
Trusted Logging Prevents tampering or covering security issues by storing AIX
virtual machine system logs securely on a central PowerVM
Virtual I/O Server. Reduces backup and archive time by
storing audit logs in a central location.
Trusted Network
Connect And Patch
Management
Ensures that site patch levels policies are adhered to in
virtual workloads. Provides notification of noncompliance
when back-level systems are activated.
Trusted Firewall Improves performance and reduces network resource
consumption by providing firewall services locally with the
virtualization layer.
Feature Benefits
Implementing LPM: See IBM PowerVM Live Partition Mobility, SG24-7460 for
information about how to implement LPM.

7.3.1 General considerations
The following list provides best practice scenarios which can benefit from the use
of LPM:
򐂰 Balancing the frame workload across multiple frames. For example, if you
need to increase the processor or memory allocation for a partition and the
frame does not have the required capacity, you can move partitions to another
frame to free up resources.
򐂰 LPM offers flexibility in repair actions and updates. For example, you can
migrate all the running partitions to another frame, while you perform
Virtual I/O Server or frame firmware updates.
򐂰 LPM can be used to migrate partitions to a new frame when a hardware
refresh is done, or to consolidate Power Systems server environments.
7.3.2 Implementing Live Partition Mobility
The following list provides best practice considerations when you implement Live
Partition Mobility (LPM):
򐂰 Before you perform an LPM operation, best practice is to run the partition
migration validation function on the HMC. This function checks that the frame,
Virtual I/O Server, LPARs, storage, and network are ready for LPM
operations. Figure 7-1 on page 111 shows the LPM validation function.

Figure 7-1 Live Partition Mobility migration validation function
򐂰 After you perform an LPM, document all changes that are made during the
operation. When you move a partition to a different frame, there might be
changes to the partition ID and the adapter slot numbers that were originally
configured.
򐂰 When possible, perform LPM operations outside of peak hours. LPM is more
efficient when the load on the network is low.
LPM: It is important to remember that LPM is not a replacement for PowerHA
SystemMirror or a disaster recovery solution. LPM can move a powered-off
partition, but not a crashed-kernel partition. Logical partitions cannot be
migrated from failed frames.

7.3.3 Storage considerations
The following storage area network (SAN) considerations must be checked when
you implement LPM:
򐂰 When you configure a virtual Small Computer System Interface (SCSI), the
storage must be zoned to both the source and target Virtual I/O Servers. Also,
only SAN disks are supported in LPM.
򐂰 When you use N-Port ID Virtualization (NPIV), confirm that both worldwide
port names (WWPNs) on the virtual Fibre Channel adapters are zoned.
򐂰 Dedicated I/O adapters must be deallocated before migration. Optical devices
in the Virtual I/O Server must not be assigned to the virtual I/O clients that are
moved.
򐂰 When you use the virtual SCSI adapters, verify that the reserve attributes on
the physical volumes are the same for the source and destination Virtual I/O
Servers.
򐂰 When you use the virtual SCSI before you move a virtual I/O client, you can
specify a new name for the virtual target device (VTD). Do this step if you
want to preserve the same naming convention on the target frame. After you
move the virtual I/O client, the VTD assumes the new name on the target
Virtual I/O Server. Example 7-3 shows the command to specify a new VTD
name.
Example 7-3 Specifying a new name for the VTD
VTD name, before LPM, on the source frame.
$lsmap -vadapter vhost1
SVSA Physloc Client Partition
ID
--------------- -------------------------------------------- ------
vhost1 U8233.E8B.061AA6P-V2-C302 0x00000003
VTD client1_hd0
Status Available
LUN 0x8100000000000000
$ chdev -dev client1_hd0 -attr mig_name=lpm_client1_hd0
VTD name, after LPM, on the target frame.
$lsmap -vadapter vhost1

SVSA Physloc Client Partition
ID
--------------- -------------------------------------------- ------
vhost1 U8233.E8B.061AA6P-V2-C302 0x00000003
VTD lpm_client1_hd0
Status Available
LUN 0x8100000000000000
7.3.4 Network considerations
The following list provides network considerations when you implement LPM:
򐂰 Shared Ethernet Adapters (SEA) must be used in an LPM environment.
򐂰 Source and target frames must be on same subnet to bridge the same
Ethernet network that the mobile partitions use.
򐂰 The network throughput is important. The higher the throughput, the less time
it takes to perform the LPM operation.
For example, if you are performing an LPM operation on a virtual I/O client
with 8 GB of memory:
– A 100-MB network, sustaining a 30-MB/s throughput, takes 36 minutes to
complete the LPM operation.
– A 1-GB network, sustaining a 300-MB/s throughput, takes 3.6 minutes to
complete the LPM operation.
7.4 Active Memory Sharing
IBM Active Memory Sharing (AMS) is a PowerVM advanced memory
virtualization technology that provides system memory virtualization capabilities
to IBM Power Systems. This technology allows multiple logical partitions to share
a common pool of physical memory.
A system administrator configures a logical partition with enough memory to
satisfy the workload and to avoid paging activity. However, such memory
demands might be needed for only a short amount of time; for example, during
workload peak times. At times, the changes in these workloads can be difficult to
predict. When a system is configured with AMS, it automatically adjusts the

amount of memory that is needed by the virtual client, depending on the
workload.
With dedicated memory, the allocation of memory to each partition is static, as
shown in Figure 7-2. The amount of memory, regardless of demand, does not
change.
Figure 7-2 Partitions with dedicated memory
With AMS, the amount of memory that is made available to a partition, changes
over the run time of the workload. The amount is based on the memory demands
of the workload.
AMS is supported by AIX, IBM i, and Linux. For more information about the setup
and requirements, see the IBM Redpaper™ publication, IBM PowerVM
Virtualization Active Memory Sharing, REDP-4470, available at this website:
http://www.redbooks.ibm.com/abstracts/redp4470.html
7.4.1 When to use Active Memory Sharing
The following three examples are perfect candidates for Active Memory Sharing
(AMS):
򐂰 Geographically separated workloads.
򐂰 Day and night workloads.
򐂰 Many logical partitions with sporadic use.
Geographically separated workloads
As a best practice, we suggest that you use AMS in an environment which
supports workloads for different regions. A suitable environment might include
Dedicated Memory
0
5
10
15
0
Time
Memory
Usage
(GB)
Americas
Europe
Asia

geographically separated workloads where logical partition memory demands
are different for day and night, or consist of many logical partitions with sporadic
use.
Figure 7-3 shows three partitions with similar memory requirements. With this
scenario, it is likely that the workloads will peak memory resources at different
times throughout a 24 hour period. High memory requirements in America during
business hours might mean that memory requirements in Asia are low.
Figure 7-3 Logical partitions with shared memory that run different regions
Day and night workloads
An environment which supports day and night workloads might be for logical
partitions which run applications during the day time and process batch jobs at
night. Figure 7-4 shows an example where the night batch process uses less
memory. The unutilizied memory can be given back to the pool for other use.
Figure 7-4 Partitions that support day and night workloads
Active Memory Sharing
0
5
10
1
Time
Memory
Usage
(GB)
Total memory used
Asia
Europe
Americas
Day and night workload
0
5
10
0
Time
Memory
Usage
(GB)
Day
Night

Large number of logical partitions with sporadic use
AMS can also be used in systems that have many logical partitions, but the use
of these logical partitions is intermittent, as shown in Figure 7-5.
Figure 7-5 Sporadically used logical partitions
7.4.2 Implementing Active Memory Sharing
When you introduce Active Memory Sharing (AMS) into a system, it is important
to plan the setup of the minimum, wanted, maximum memory, memory
weighting, and the paging devices.
For minimum memory, in most cases, 1 - 2 GB is normally sufficient. However,
the size is dependent on the server workload.
The wanted amount of memory needs to be configured to a value which is
sufficient for the workload, and to avoid paging activity.
For maximum memory, add 20 - 30% to the wanted memory value.
The maximum memory value is important for two reasons. First, if the maximum
memory value is set too low, and a system encounters a spike in workload (for
instance, natural growth, or at the end of a financial quarter), you are able to
dynamically add memory only up to the maximum value configured. Second, the
maximum memory value is used by the Hypervisor to calculate and reserve the
page tables. To avoid wasting memory, it is a best practice to carefully plan, and
keep maximum memory values to a minimum.
Infrequent use
0
5
10
0
Time
Memory
Usage
(GB)
System A
System B
System C
System D
System E
System F

Memory weight defines a factor that is used by the Power Hypervisor in
determining the allocation of physical system memory. It determines which
pages must be copied to the paging devices in case of physical memory
over-commitment. The best practice is to set a higher weight for production
systems and a lower weight for development systems. Figure 7-6 shows the
weight configuration on a logical partition.
Figure 7-6 AMS memory weight on a logical partition
Paging devices
Response from the AMS paging devices has a significant effect for clients when
there is a physical memory over-commitment. You want these operations to be
achieved as fast as possible. The best practice is to set the size of the paging
device equal to the maximum logical memory, with an exception for IBM i. In
IBM i, the paging device must be larger than the maximum memory defined in

the partition profile. It requires 1 bit extra for every 16-byte page it allocates. For
instance, a partition with 10 Gb of memory needs a paging device with a
minimum size of 10.08 Gb defined in the partition profile.
Remember, some storage vendors do not support dynamically increasing the
size of a logical unit number (LUN). If this case, the paging device needs to be
removed and re-created. The following list outlines best practices to consider
when you configure paging devices:
򐂰 Use physical volumes, where possible, over logical volumes.
򐂰 If you use logical volumes, use a small stripe size.
򐂰 Use thin provisioned LUNs.
򐂰 Spread the I/O load across as much of the disk subsystem as possible.
򐂰 Use a write cache, whether it is on the adapter or storage subsystem.
򐂰 Size your storage hardware according to your performance needs.
򐂰 Ensure that the PVIDs for the paging devices for physical volumes that are set
up by the HMC, are cleared before use.
򐂰 If you plan to use a dual Virtual I/O Server configuration, paging devices must
be provided through a SAN, and accessible from both Virtual I/O Servers.
Active Memory Expansion
If you are using AMS on AIX, best practice is to enable Active Memory Expansion
(AME). AME allows a partition to expand its memory up to a specific factor. With
AME enabled, page loaning (an AMS mechanism) tries to compress memory
page content before you move it to the paging device. When compression is no
longer possible, AMS moves memory pages straight to the paging device.
7.4.3 Active Memory Deduplication
Active Memory Deduplication is a PowerVM technology that minimizes the
existence of identical memory pages in the main memory space. When you run
workloads on traditional partitions, multiple identical data is saved across
different positions in the main memory. Active Memory Deduplication can
improve usage of physical memory. As a best practice, we suggest that you
enable Active Memory Deduplication if you use AMS and a Virtual I/O Server that
is not processor-constrained.
The Power Hypervisor reserves extra memory for page tracking and calculation
of the deduplication table ratio. Therefore, it is suggested that you keep the
maximum shared memory pool size parameter to a minimum. For more
information about setup and configuration of Active Memory Deduplication, see
Power Systems Memory Deduplication, REDP-4827.

ABI application binary interface
AC Alternating Current
ACL access control list
AFPA Adaptive Fast Path
Architecture
AIO Asynchronous I/O
AIX Advanced Interactive
Executive
APAR authorized program analysis
report
API application programming
interface
ARP Address Resolution Protocol
ASMI Advanced System
Management Interface
BFF Backup File Format
BIND Berkeley Internet Name
Domain
BIST Built-In Self-Test
BLV Boot Logical Volume
BOOTP Bootstrap Protocol
BOS Base Operating System
BSD Berkeley Software Distribution
CA certificate authority
CATE Certified Advanced Technical
Expert
CD compact disc
CDE Common Desktop
Environment
CD-R compact disc recordable
CD-ROM compact-disc read-only
memory
CEC central electronics complex
Abbreviations and acronyms
CHRP Common Hardware
Reference Platform
CLI command-line interface
CLVM Concurrent LVM
CPU central processing unit
CRC cyclic redundancy check
CSM Cluster Systems
Management
CUoD Capacity Upgrade on
Demand
DCM Dual Chip Module
DES Data Encryption Standard
DGD Dead Gateway Detection
DHCP Dynamic Host Configuration
Protocol
DLPAR dynamic LPAR
DMA direct memory access
DNS Domain Name System
DRM dynamic reconfiguration
manager
DR dynamic reconfiguration
DVD digital versatile disc
EC EtherChannel
ECC error correction code
EOF end-of-file
EPOW emergency power-off warning
ERRM Event Response resource
manager
ESS IBM Enterprise Storage
Server®
F/C Feature Code
FC Fibre Channel
FC_AL Fibre Channel Arbitrated Loop
FDX Full Duplex

FLOP Floating Point Operation
FRU field-replaceable unit
FTP File Transfer Protocol
GDPS® IBM Geographically
Dispersed Parallel Sysplex™
GID group ID
GPFS General Parallel File System
GUI graphical user interface
HACMP High Availability Cluster
Multiprocessing
HBA host bus adapter
HMC Hardware Management
Console
HTML Hypertext Markup Language
HTTP Hypertext Transfer Protocol
Hz hertz
I/O input/output
IBM International Business
Machines
ID identifier
IDE Integrated Device Electronics
IEEE Institute of Electrical and
Electronics Engineers
IP Internet Protocol
IPAT IP address takeover
IPL initial program load
IPMP IP Multipathing
ISV independent software vendor
ITSO International Technical
Support Organization
IVM Integrated Virtualization
Manager
JFS journaled file system
L1 level 1
L2 level 2
L3 level 3
LA Link Aggregation
LACP Link Aggregation Control
Protocol
LAN local area network
LDAP Lightweight Directory Access
Protocol
LED light-emitting diode
LMB Logical Memory Block
LPAR logical partition
LPP licensed program product
LUN logical unit number
LV logical volume
LVCB Logical Volume Control Block
LVM Logical Volume Manager
MAC Media Access Control
Mbps megabits per second
MBps megabytes per second
MCM multiple chip module
ML Maintenance Level
MP Multiprocessor
MPIO Multipath I/O
MTU maximum transmission unit
NFS Network File System
NIB Network Interface Backup
NIM Network Installation
Management
NIMOL NIM on Linux
N_PORT Node Port
NPIV N_Port Identifier Virtualization
NVRAM nonvolatile random access
memory
ODM Object Data Manager
OS operating system
OSPF Open Shortest Path First
PCI Peripheral Component
Interconnect
PCI-e iPeripheral Component
Interconnect Express

Abbreviations and acronyms 121
PIC Pool Idle Count
PID process ID
PKI public key infrastructure
PLM Partition Load Manager
POST power-on self-test
POWER Performance Optimization
with Enhanced Risc
(Architecture)
PPC Physical Processor
Consumption
PPFC Physical Processor Fraction
Consumed
PTF program temporary fix
PTX Performance Toolbox
PURR Processor Utilization
Resource Register
PV physical volume
PVID Physical Volume Identifier
PVID Port Virtual LAN Identifier
QoS quality of service
RAID Redundant Array of
Independent Disks
RAM random access memory
RAS reliability, availability, and
serviceability
RBAC role-based access control
RCP Remote Copy
RDAC Redundant Disk Array
Controller
RIO remote input/output
RIP Routing Information Protocol
RISC reduced instruction-set
computer
RMC Resource Monitoring and
Control
RPC Remote Procedure Call
RPL Remote Program Loader
RPM Red Hat Package Manager
RSA Rivest-Shamir-Adleman
algorithm
RSCT Reliable Scalable Cluster
Technology
RSH Remote Shell
SAN storage area network
SCSI Small Computer System
Interface
SDD Subsystem Device Driver
SDDPCM Subsystem Device Driver
Path Control Module
SMIT System Management
Interface Tool
SMP symmetric multiprocessor
SMS system management services
SMT simultaneous multithreading
SP Service Processor
SPOT Shared Product Object Tree
SRC System Resource Controller
SRN service request number
SSA Serial Storage Architecture
SSH Secure Shell
SSL Secure Sockets Layer
SUID Set User ID
SVC SAN Volume Controller
TCP/IP Transmission Control
Protocol/Internet Protocol
TL Technology Level
TSA Tivoli System Automation
UDF Universal Disk Format
UDID Universal Disk Identification
VIPA virtual IP address
VG volume group
VGDA Volume Group Descriptor
Area
VGSA Volume Group Status Area
VLAN virtual local area network

VP Virtual Processor
VPD vital product data
VPN virtual private network
VRRP Virtual Router Redundancy
Protocol
VSD Virtual Shared Disk
WLM Workload Manager
WWN worldwide name
WWPN worldwide port name

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
The following IBM Redbooks publications provide additional information about
the topic in this document. Note that some publications referenced in this list
might be available in softcopy only.
򐂰 Hardware Management Console V7 Handbook, SG24-7491
򐂰 IBM PowerVM Virtualization Active Memory Sharing, REDP-4470
򐂰 IBM PowerVM Virtualization Introduction and Configuration, SG24-7940
򐂰 IBM PowerVM Virtualization Managing and Monitoring, SG24-7590
򐂰 Integrated Virtualization Manager on IBM System p5, REDP-4061
򐂰 Power Systems Memory Deduplication, REDP-4827
򐂰 PowerVM Migration from Physical to Virtual Storage, SG24-7825
򐂰 IBM System Storage DS8000 Host Attachment and Interoperability,
SG24-8887
򐂰 IBM Flex System p260 and p460 Planning and Implementation Guide,
SG24-7989
You can search for, view, download, or order these documents and other
Redbooks, Redpapers, Web Docs, draft and additional materials, at the following
website:
ibm.com/redbooks
Other publications
These publications are also relevant as further information sources.
򐂰 The following types of documentation are located through the Internet at the
following URL:
http://publib.boulder.ibm.com/infocenter/powersys/v3r1m5/index.jsp

– User guides
– System management guides
– Application programmer guides
– All commands reference volumes
– Files reference
– Technical reference volumes used by application programmers
򐂰 Detailed documentation about the PowerVM feature and the Virtual I/O
Server:
https://www14.software.ibm.com/webapp/set2/sas/f/vios/documentation/
home.html
Online resources
These Web sites are also relevant as further information sources:
These Web sites and URLs are also relevant as further information sources:
򐂰 AIX and Linux on POWER community
http://www-03.ibm.com/systems/p/community/
򐂰 Capacity on Demand
http://www.ibm.com/systems/p/cod/
򐂰 IBM PowerVM
http://www.ibm.com/systems/power/software/virtualization/index.html
򐂰 AIX 7.1 Information Center
http://publib.boulder.ibm.com/infocenter/aix/v7r1/index.jsp
򐂰 IBM System Planning Tool
http://www.ibm.com/servers/eserver/support/tools/systemplanningtool/
򐂰 IBM Systems Hardware Information Center
http://publib.boulder.ibm.com/infocenter/systems/scope/hw/index.jsp
򐂰 IBM Systems Workload Estimator
http://www-304.ibm.com/jct01004c/systems/support/tools/estimator/ind
ex.html
򐂰 Latest Multipath Subsystem Device Driver home page
http://www-1.ibm.com/support/docview.wss?uid=ssg1S4000201

Related publications 125
򐂰 Novell SUSE LINUX Enterprise Server information
http://www.novell.com/products/server/index.html
򐂰 SCSI T10 Technical Committee
http://www.t10.org
򐂰 SDDPCM software download page
򐂰 Service and productivity tools for Linux on POWER
https://www14.software.ibm.com/webapp/set2/sas/f/lopdiags/home.html
򐂰 Virtual I/O Server home page
http://www14.software.ibm.com/webapp/set2/sas/f/vios/home.html
򐂰 Virtual I/O Server fix pack download page
http://www14.software.ibm.com/webapp/set2/sas/f/vios/download/home.html
Help from IBM
IBM Support and downloads
ibm.com/support
IBM Global Services
ibm.com/services

Index
Numerics
3rd part list application 26
A
Active Memory Deduplication 3, 118
deduplication table ratio 118
Active Memory Expansion 118
page loaning 118
Active Memory Sharing 3, 113
dedicated memory 114
desired memory 116
examples 114
implementing 116
maximum memory 116
memory weight 117
minimum memory 116
page tables 116
paging devices 117
AIX_APPCM 66
alt_root_vg command 25, 27
B
Backing up VIOS 32
commands
backupios command 34
savevgstruct command 36
viosbr command 35
Disaster recovery 32
External device configuration 33
HMC 33
IVM 33
backupios command 34–35
booting issues 92
bootlist command 25
bosboot command 92
C
cfgmgr command 92
cfgnamesrv command 37
chdev command 67, 69, 91, 112
chsysstate 48
command
viosecure 107
commands 91
alt_root_vg command 25
backupios command 34–35
bootlist command 25
bosboot command 92
cfgmgr command 92
cfgnamesrv command 37
chdev command 67, 69
chpath command 79
crontab command 35
entstat 98
entstat command 37
fcstat command 69
HMC
chsysstate 48
mkvterm 48
hostmap command 37
installios command 38, 40
ioslevel command 38
loginmsg command 38
lparstat 96
lsattr command 70, 92
lsdev command 37, 78
lshwres HMC command 72
lslparutil 103
lspath command 79
lspv command 38
lsuser command 38
lsvg command 38
manage_disk_drivers command 66
mirrorios command 25, 63
mkvdev command 75–76
motd command 38
mpio_get_config command 75
netstat command 37
nim command 40
oem_setup_env command 76
optimizenet command 37
pcmpath command 69
recfgct command 25
restorevgstruct command 41
seastat 98

svcinfo command 75
vios_advisor 104
viosbr command 35–36
viosecure command 38
crontab command 35
D
deduplication table ratio 118
desired memory 20
desired processing units 18
disk attributes
queue_depth 78
reserve_policy 78
Disk Monitoring
Commands
viostat command 96
dlpar 79, 92
DLPAR operations 42
Overwriting partition profiles 42
Renaming partition profiles 42
Virtual FC adapters 43
DS3950 66
DS4000 66, 75
DS4800 75
DS5000 66, 75
DS8000 66
dual Virtual I/O Server 29
dual Virtual I/O Server environment 27
Dual virtual switch 56
DVD 28
Dynamic logical partion operations 92
Dynamic logical partitioning 79
dynamic logical partitioning 3
dyntrk attribute 68, 92
E
eadme instructions 27
entstat command 37
F
fast_fail attribute 68
fc_err_recov attribute 68, 92
fcs AIX device 67
fcstat command 69
Fibre Channel adapter
Attributes
dyntrk 68, 92
fast_fail 68
fc_err_recov 68, 92
max_transfer_size 74
max_xfer_size 68, 92
num_cmd_elem 92
num_cmd_elems 68
Fibre Channel adapter Attributes 67
Fibre Channel adapter statistics 68
Fibre Channel Monitoring
Commands
fcstat command 99
max_xfer_size 99
num_cmd_elems 99
File backed devices 70
firmware updates 26
Fix Level Recommendation 25
Fix Packs 25
flow_ctrl 54
fragmentation 53
fscsi AIX device 67
H
hardware features 8
Active Memory Expansion 8
Active Memory Mirroring 8
RAS 8
Hardware Management Console 24
Hardware Manager Console 28
High availability 64
High Availabilty 77
HMC 100
Allow performance information collection 95
Enable Connection Monitoring 100
HMC commands
lshwres 72
hostmap command 37
I
IBM Fix Central 26
IBM i 70–71, 89
iFixes 25
installation of VIOS 24
ioslevel command 38
ISO images 26
IVM 77

Index 129
J
jumbo_frames 54
L
large_recieve 54
large_send 54
Live Partition Mobililty 67
Live Partition Mobility 3, 46, 55, 72, 89, 109
Logical Volume 74
Logical Volume mappings 77
Logical volumes 70
loginmsg command 38
Long-term performance
Commands
cfgassist command 101
cfgsvc command 102
lssvc command 102
postprocesssvc command 102
startsvc command 102
stopsvc command 102
topas_nmon command 100
topasout command 101
topasrec command 100–101
LPM 46, 55, 67, 89
lsattr command 70, 92
lsdev command 37, 78
lshwres commands 72
lspv command 38
lsuser command 38
lsvg command 38
M
max_transfer_size attribute 74
max_xfer_size attribute 68, 92
maximum memory 21
maximum processing units 18
maximum transfer unit 52, 54
maximum virtual adapter limit 21
maximum virtual adapters 72
maximum virtual processors 18
Memory Monitoring
Commands
svmon command 100
vmstat command 100
Micro-partitioning 3
migration 28
migration verification tool 110
minimum memory 20
minimum virtual processors 18
Mirroring 71, 77
mirroring 29, 63
mirrorios command 25, 63
mkvdev command 75–76
mkvterm 48
motd command 38
MPIO 77
mpio_get_config 75
MTU 52, 54
Multipathing 66, 77, 89
Multipathing for clients 66
SDDPCM 66
Storage 89
N
Naming conventions 74
netstat command 37
Network bandwidth tuning 54
flow_ctrl 54
jumbo_frames 54
large_recieve 54
large_send 54
tcp_pmtu_discovery 54
udp_pmtu_discovery 54
Network Installation Manager 24, 28
Network Interface Backup 29
Network Monitoring
Commands
entstat command 97
netstat command 97
seastat command 97
Network protocols
802.1Q 50
802.3ad 50
EtherChannel 50
Link aggregation 50
Network interface backup 50, 55, 57–58
NIB 50, 55, 57–58
trunking 50
VLAN tagging 50, 56
Networking 49
NIM 24
nim command 40
NIM server resilience 41
NPIV 3, 62, 67, 88
num_cmd_elem attribute 92

num_cmd_elems attribute 68
O
oem_setup_env command 76
Optical devices 79
optimizenet command 37
P
packet fragmentation 53
page loaning 118
page tables 116
paging devices 117
Partition startup policy 47
pcmpath command 69
physical I/O adapters 21
Physical volumes 70
Physical volumes mapping 76–77
Power Hypervisor 16
PowerHA 77, 89, 111
PowerSC 108
PowerVM
hardware planning 7
PowerVM editions 4
PowerVM features
Active Memory Deduplication 3
Active Memory Sharing 3
dynamic logical partitioning 3
Integrated Virtualization Manager 3
Live Partition Mobility 3
Micro-partitioning 3
NPIV 3
PowerVM Hypervisor 3
Shared Processor Pools 3
Shared Storage Pools 3
PowerVM Hypervisor 3
PowerVM Hypervisor Security 106
priority to processor 19
processing units 17
Q
queue_depth disk attribute 78
R
recfgct command 25
Redbooks website 123
Contact us xvii
Redundancy 90
reserve_policy attribute 78
Restoring VIOS
commands
nim command 40
NFS 38
rootvg 62, 78
rootvg on virtual SCSI clients 67
S
SAN switch 90
SDDPCM 66
AIX_APPCM 66
DS3950 66
DS4000 66
DS5000 66
SEA 51, 56, 58
SEA failover 58
SEA threading 52
server shutdown 45
server startup 45
Service Packs 25
Shared Ethernet 113
Shared Ethernet Adapter 51, 56, 58
Shared Ethernet Adapter failover 58
shared or dedicated resources 10
shared processing mode 16
Shared Processor Pools 3
Shared Storage Pool 75, 82
creating 84
requirements 82
SAN storage considerations 85
specifications 83
thin or thick provisioning 86
verifying 85
when to use 83
Shared Storage Pools 3, 70
Short-term performance
Commands
fcstat command 94
netstat command 94
seastat command 94
svmon command 94
topas command 94
topas_nmon command 95

Index 131
viostat command 94
vmstat command 94
single Virtual I/O Server 29
single Virtual I/O Server environment 27
SPT 33
startup sequence 48
Storage
DS3950 66
DS4000 66, 75
DS4800 75
DS5000 66, 75
DS8000 66
Dynamic Logical partition 92
Dynamic logical partitioning 79
fcs AIX device 67
Fibre Channel adapter attributes 67
Fibre Channel adapter statistics 68
File Backed devices 70
fscsi AIX device 67
IBM i 70–71, 89
IVM 77
Live Partition Mobility 72
Logical Volume 74
Logical Volume mappings 77
Logical Volumes 70
maximum virtual adapters 72
Mirroring 71, 77
mkvdev command 76
MPIO 77
mpio_get_config command 75
Multipathing 66, 77
Naming conventions 74
NPIV 62, 88
Optical devices 79
Physical devices 70
Physical volumes mapping 76–77
PowerHA 77
Redundancy 90
rootvg 62, 78
SAN Switch 90
Shared Storage Pool 75
Shared Storage Pools 70
Supported Solutions 66
Tape devices 75, 90
Virtual Fibre Channel 62
virtual Fibre Channel adapter 90
Virtual Fibre Channel server adapter 74
Virtual optical devices 70
Virtual SCSI 70
virtual SCSI 62
Virtual SCSI server adapter 73–74
Virtual slots 71
Virtual tape devices 70, 79
WWPN 89
Support Storage 66
svcinfo command 75
System Planing Tool, 24
System Planning Too 21
System Planning Tool 7, 33
Systems Workload Estimator 10
T
Tape devices 75, 79, 90
tcp_pmtu_discovery 54
thin or thick provisioning 86
tracing your storage configuration 22
U
udp_pmtu_discovery 54
Uncapped partitions 18
V
VIOS installation 24
viosbr command 35–36
viosecure command 38
Virtual adapters slots considerations 71
Virtual client SCSI adapter 74
virtual FC adapters 22
Virtual Fibre Channel 62
Virtual Fibre Channel adapter 90
Virtual Fibre Channel server adapter 74
Virtual I/O Server
configuration limits 5
ethernet adapter 4–5
Hardware Management Console 4
Integrated Virtualization Manager 4
IP forwarding 5
logical volume manager 6
memory 4
network and storage 11
network services 106
physical disk 4
planning 6
processor 4
requirements 4

SCSI protocol 5
security 106
single, dual, or multiple 11
sizing 8
slot numbering 12
storage adapter 4
virtual SCSI 5
Virtual I/O Server profile 16
Virtual media repository 44
Virtual optical devicesStorage
Virtual SCSI 70
virtual SCSI 22, 62, 112
Virtual SCSI server adapter 73
Virtual tape devices 70, 79
virtual target device 112
VLAN
VLAN bridging 51
VLAN tagging 56
W
weight value 19
WWPN 89

(0.2”spine)
0.17”<->0.473”
90<->249
pages
IBM
PowerVM
Best
Practices

®
SG24-8062-00 ISBN 0738437301
INTERNATIONAL
TECHNICAL
SUPPORT
ORGANIZATION
BUILDING TECHNICAL
INFORMATION BASED ON
PRACTICAL EXPERIENCE
IBM Redbooks are developed by
the IBM International Technical
Support Organization. Experts
from 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
®
IBM PowerVM
Best Practices
A collection of
recommended
practices to enhance
your use of the
PowerVM features
A resource to build
on knowledge found
in other PowerVM
documents
A valuable
refererence for
experienced IT
specialists and IT
architects
This IBM Redbooks publication provides best practices for
planning, installing, maintaining, and monitoring the IBM
PowerVM Enterprise Edition virtualization features on IBM
POWER7 processor technology-based servers.
PowerVM is a combination of hardware, PowerVM
Hypervisor, and software, which includes other virtualization
features, such as the Virtual I/O Server.
This publication is intended for experienced IT specialists
and IT architects who want to learn about PowerVM best
practices, and focuses on the following topics:
򐂰 Planning and general best practices
򐂰 Installation, migration, and configuration
򐂰 Administration and maintenance
򐂰 Storage and networking
򐂰 Performance monitoring
򐂰 Security
򐂰 PowerVM advanced features
This publication is written by a group of seven PowerVM
experts from different countries around the world. These
experts came together to bring their broad IT skills, depth of
knowledge, and experiences from thousands of installations
and configurations in different IBM client sites.
Back cover

IBM PowerVM Best Practices

More Related Content

What's hot (20)

Viewers also liked (20)

Similar to IBM PowerVM Best Practices (20)

More from IBM India Smarter Computing (20)

Recently uploaded (20)

IBM PowerVM Best Practices