What is (and is not) Live Partition Mobility

LPM is an IBM PowerVM® feature capable of migrating a running partition from one IBM Power Systems™ server to another; migration is performed without disrupting the transactions and the applications that are running on the partition, in other words transparently for the business.

LPM is not a business continuity feature. LPM improves the serviceability and maintainability of an environment running on Power Systems servers. In other words, this feature is designed to make the IT manager's life easier, when a planned maintenance is scheduled because it is capable of reducing the downtime to zero for a planned maintenance activity. Reasons for using LPM also include workload consolidation (from many servers to one), workload balancing (distribute workloads across a pool of servers), and workload resilience in response to a Predictive Failure Analysis (PFA) event.

IBM i and LPM requirements

Describing LPM requirements is not the goal of this article as the support statements are reported in IBM Redbooks®. General support statements for IBM i LPM are documented in the IBM i Technology Updates - Live Partition Mobility article.

The building blocks

The following list of building blocks is generally applicable to any LPM scenario:

• The IBM Power® server defined as source, running the IBM i partition before running LPM
• The Power server defined as target, running the IBM i partition running LPM
• The storage system hosting the IBM i logical unit numbers (LUNs): LUNs must be visible to both the source and destination servers
• The LPM orchestrator, validating and coordinating the migration operations
• The IBM i partition, running the workload

Back to top

The hardware

The source server

The source server is an IBM Flex System™ p460 Compute Node, in an IBM Flex System Enterprise Chassis called Chassis 1.

The Flex System p460 Compute Node, where the IBM i workload is running, occupies slots 3 and 4, as it is a full-wide server.

The IBM i partition is installed with the IBM i 7.1 TR4 release and the latest program temporary fix (PTF) groups, with 4 GB of RAM and 0.2 processor uncapped units. The IBM i partition is fully virtual, with two Virtual I/O Server (VIOS) partitions managing the disks and network virtualization. Storage is provided by two 25 GB LUNs, visible to both the VIOS and mapped to the IBM i logical partition (LPAR) through virtual Small Computer System Interface (VSCSI) adapters.

The destination server

The destination server is another IBM Flex System p460 Compute Node, in a different IBM Flex System Enterprise Chassis called Chassis 3.

The Flex System p460 Compute Node where IBM i will be migrated, occupies slots 1 and 2, as it is a full-wide server, too. Also in this server, two VIOS partitions are running and the same IBM i storage LUNs are already visible to them.

The storage system

All the IBM i storage space is provided by two 25 GB LUNs on an IBM Storwize® V7000 storage system. As the IBM i disks are made visible to the partition through virtual SCSI adapters, LUNs are mapped to the Fibre Channel (FC) cards owned by the VIOS.

In this context, on each VIOS on the source server:

• The two IBM i LUNs must be visible as hdisk devices.
• The two IBM i LUNs must be associated with two virtual destination devices (VTD) associated to the logical representation of the virtual SCSI adapter pointing to the IBM i partition (vhost).

		
$ hostname
sn1054a3bvios1

$ lspv -size | grep 'SIZE\|25600'
NAME            PVID                                SIZE(megabytes) 
hdisk5          00054a3b868044e2                    25600           
hdisk6          00054a3b868072a7                    25600

$ lsmap -vadapter vhost2
SVSA            Physloc                                      Client Partition ID
--------------- -------------------------------------------- ------------------
vhost2          U7895.42X.1054A3B-V1-C7                      0x00000005

VTD                   vtscsi2
Status                Available
LUN                   0x8100000000000000
Backing device        hdisk5
Physloc               U78AF.001.WZS0224-P1-C35-L1-T1-W5005076802158969-L4000000000000
Mirrored              false

VTD                   vtscsi3
Status                Available
LUN                   0x8200000000000000
Backing device        hdisk6
Physloc               U78AF.001.WZS0224-P1-C35-L1-T1-W5005076802158969-L5000000000000
Mirrored              false

		
$ hostname
sn1054a3bvios2

$ lspv -size | grep 'SIZE\|25600'
NAME            PVID                                SIZE(megabytes) 
hdisk5          00054a3b868044e2                    25600           
hdisk6          00054a3b868072a7                    25600

$ lsmap -vadapter vhost2
SVSA            Physloc                                      Client Partition ID
--------------- -------------------------------------------- ------------------
vhost2          U7895.42X.1054A3B-V2-C7                      0x00000005

VTD                   vtscsi2
Status                Available
LUN                   0x8100000000000000
Backing device        hdisk5
Physloc               U78AF.001.WZS0224-P1-C37-L1-T1-W5005076802158969-L4000000000000
Mirrored              false

VTD                   vtscsi3
Status                Available
LUN                   0x8200000000000000
Backing device        hdisk6
Physloc               U78AF.001.WZS0224-P1-C37-L1-T1-W5005076802158969-L5000000000000
Mirrored              false

On the destination server:

• The two IBM i LUNs must be visible as hdisk devices.

		
$ hostname
sn1054a5bvios1

$ lspv -size | grep 'SIZE\|25600'
NAME            PVID                                SIZE(megabytes) 
hdisk1          00054a3b868044e2                    25600           
hdisk2          00054a3b868072a7                    25600

$ lsmap -all
$

		
$ hostname
sn1054a5bvios2

$ lspv -size | grep 'SIZE\|25600'
NAME            PVID                                SIZE(megabytes) 
hdisk1          00054a3b868044e2                    25600           
hdisk2          00054a3b868072a7                    25600

$ lsmap -all
$

In some configuration, the reserve_policy attribute of the disks must be changed. For instance, in dual-VIOS environments, any disk device that is presented to a client partition through a virtual SCSI adapter must be set to no_reserve.

For partition mobility, the reserve_policy on the destination VIOS must be the same as on the source VIOS.

Back to top

IBM Flex System Manager

In an IBM PureFlex System environment, Flex System Manager (FSM) is an intelligent appliance designed for management-based tasks for PureFlex System hardware, storage, and networking. FSM is also oriented to the virtualization management and OS provisioning. When managing Power Systems compute nodes, FSM is irreplaceable because the expertise proper to Hardware Management Console (HMC) (including the capability to orchestrate the LPM), merged into the FSM.

Double-check the FSM release

To check the FSM release, use Secure Shell (SSH) to log into the FSM and enter the smcli lsconfig command. FSM must be at 1.2 release or later.

Double-check that servers are capable of IBM i Live Partition Mobility

Perform the following steps to double-check that servers are capable of IBM i LPM.

1. On the FSM Resource Explorer page, identify the source server.
2. Right-click the source server to get the navigation menu, and then click System Configuration.
3. Click Edit Host.
4. Click the Migration tab.
5. Double-check that IBM i Partition Mobility capable is true.

Repeat the above steps also for the destination server.

Define the service mover partition on both source and destination

FSM is instrumental to set up the VIOS partition as the service mover, on both source and destination servers.

1. On the FSM Resource Explorer page, identify the source server.
2. Right-click the source server to get the navigation menu and then click System Configuration.
3. Click Edit Host.
4. Select one of the VIOS as the service virtual server.

Repeat the steps above also for the destination server.

Back to top

The IBM i partition

IBM i must be installed with the 7.1 TR4 code (or later) and the latest PTF groups available have to be installed and applied.

1. On the FSM Resource Explorer page, identify the source server.
2. Right-click the IBM i partition that is running on it to get the navigation menu and then click System Configuration.
3. Click Manage Virtual Server.

IBM i partition must be fully virtual; in other words, no physical slot must be assigned to it, as any I/O resource has to be accessible through the VIOS virtualization features. The IBM i restricted I/O mode check box must be selected.

From a disk perspective, LUNs are mapped to the VIOS and therefore assigned to the IBM i partition through virtual SCSI adapters. This refers to the sections in Listing 1 and Listing 2. The IBM i partition profile must be consistent too.

As described in the previous sections, in this scenario, IBM i has two 25 GB disks available and part of the System ASP (this is not a mandatory condition, just the way I decided to test it).

From a network perspective, it is not important how many virtual Ethernet adapters are associated to the IBM i partition; the only important condition is that any bridge (SEA) and VLAN configured on the source server is consistently there, also on the destination server.

IBM i partition must not have any physical or virtual media device.

Back to top

Running LPM

My preferred method to run the LPM in a PureFlex System environment consists of using the Validation feature and then Migrating afterwards. This can be done by right-clicking the IBM i partition to be migrated, and then clicking
Operations > Mobility > Validate.

The validation task runs all the needed tasks to check whether a migration would happen correctly. After the validation happens, simply click Migrate to initiate the LPM of the IBM i partition.

Back to top

Resources

Video: IBM i Live Partition Mobility in a multi-chassis PureFlex environment

About the author

Luca Comparini is an IT Infrastructure Consultant from IBM Systems and Technology Group Lab Services Europe, located in La Gaude, France. His job role is to ease Europe, Middle East, and Africa (EMEA) client acceptance of IBM emerging technologies, thanks to expertise in virtualization, IBM systems, systems management, and cloud computing [mainly infrastructure as a service (IaaS)]. Designated as one of the European Technical Leaders on IBM PureFlex System, he started working on expert integrated systems since the first pre-GA code, driving proof of concept and first-of-a-kind implementations around Europe.

Rate this article

Comments

Back to top

Table of contents

• What is (and is not) LPM
• IBM i and LPM requirements
• Building Blocks
• The hardware
• The FSM
• The IBM i partition
• Running LPM
• Resources
• About the author
• Comments

Next steps from IBM

Rational products offer a complete collaborative lifecycle management set of tools for the IBM i platform for existing and new developers.

---

• Try: Download a free trial of Rational Developer for Power Systems, which provides an integrated development environment across the major platforms.
• Buy: Rational Developer for Power Systems

Dig deeper into IBM i on developerWorks