Source and destination servers in a partition mobility environment
Two servers are involved in partition mobility that is managed by a Hardware Management Console (HMC). The source server is the server from which you want to migrate the logical partition, and the destination server is the server to which you want to migrate the logical partition.
The source and destination servers must be POWER6® processor-based servers, or later, to participate in partition mobility. The destination server must have enough available processor and memory resources to allow the mobile partition to run on its server.
POWER7® processor-based servers with firmware at level FW760, or later, can support the Dynamic Platform Optimizer (DPO) function. DPO is a hypervisor function initiated by the HMC. DPO rearranges the logical partition processors and memory on the system, to improve the affinity between processors and memory of the logical partition. When DPO is running, mobility operations targeting the system being optimized will be blocked. To continue with the migration, you must either wait for the DPO operation to complete, or manually stop the DPO operation.
Huge pages
Huge pages can improve performance in specific environments what require a high degree of parallelism, such as in DB2® partitioned database environments. You can specify the minimum, desired, and maximum number of huge pages to assign to a logical partition when you create the logical partition or partition profile.
A logical partition cannot participate in active partition mobility if huge pages are used. However, an inactive partition migration can be performed if the mobile partition uses huge pages. The partition profile will maintain the huge page resources, but the specified number of huge page resources may not be available on the destination server, in which case the logical partition will boot without some or all these huge pages after the inactive migration.
Barrier synchronization register
The barrier synchronization register (BSR) is a memory register that is located on certain processors based on POWER® technology. A parallel-processing application running on the AIX® operating system can use a BSR to perform barrier synchronization, which is a method for synchronizing the threads in the parallel-processing application.
Shared memory pool
Shared memory is physical memory that is assigned to the shared memory pool and shared among multiple logical partitions. The shared memory pool is a defined collection of physical memory blocks that are managed as a single memory pool by the hypervisor. Logical partitions that you assign to the shared memory pool share the memory in the pool with other logical partitions that you assign to the pool.
If the mobile partition uses shared memory on the source server, the destination server must also have a shared memory pool to which the mobile partition can be assigned. If the mobile partition uses dedicated memory on the source server, it must also use dedicated memory on the destination server.
Inactive partition mobility policy
For inactive partition mobility, you can select one of the following configurations in the HMC for memory and processor-related settings of the mobile partition. If you are able to start the partition, and you select the current configuration as the mobility policy, then memory and processor-related settings are obtained from the partition state that is defined in the hypervisor. However, if you are unable to start the partition, or you select the last activated profile on the source server as the mobility policy, then memory and processor-related settings are obtained from the last activated profile on the source server. The mobility policy that you select applies to all inactive migrations, where the source server is the server on which you have set the policy.
For inactive partition mobility validation, the HMC either uses the hypervisor data or the last activated profile data to verify that the partition can be migrated to the destination server.