Processors
A processor is a device that processes programmed instructions. The more processors that you assign to a logical partition, the greater the number of concurrent operations that the logical partition can run at any given time.
You can set a logical partition to use either processors that are dedicated to the logical partition or processors that are shared with other logical partitions. If a logical partition uses dedicated processors, then you must assign processors (in increments of whole numbers) to the logical partition. A logical partition that uses dedicated processors cannot use any processing capacity beyond the processors that are assigned to the logical partition.
By default, all physical processors that are not dedicated to specific logical partitions are grouped together in a shared processor pool. You can assign a specific amount of the processing capacity in this shared processor pool to each logical partition that uses shared processors. Some models allow you to use the HMC to configure multiple shared processor pools. These models have a default shared processor pool that contains all the processor resources that do not belong to logical partitions that use dedicated processors or logical partitions that use other shared processor pools. The other shared processor pools on these models can be configured with a maximum processing unit value and a reserved processing unit value. The maximum processing unit value limits the total number of processors that can be used by the logical partitions in the shared processor pool. The reserved processing unit value is the number of processing units that are reserved for the use of uncapped logical partitions within the shared processor pool.
You can set a logical partition that uses shared processors to use as little as 0.10 processing units, which is approximately a 10th of the processing capacity of a single processor. When the firmware is at level 7.6, or later, you can set a logical partition that uses shared processors to use as little as 0.05 processing units, which is approximately a 20th of the processing capacity of a single processor. You can specify the number of processing units to be used by a shared processor logical partition down to the 100th of a processing unit. Also, you can set a shared processor logical partition so that, if the logical partition requires more processing capacity than its assigned number of processing units, the logical partition can use processor resources that are not assigned to any logical partition or processor resources that are assigned to another logical partition but that are not being used by the other logical partition. (Some server models might require you to enter an activation code before you can create logical partitions that use shared processors.)
You can assign up to the entire processing capacity on the managed system to a single logical partition, if the operating system and server model supports doing so. You can configure your managed system so that it does not comply with the software license agreement for your managed system, but you will receive out-of-compliance messages if you operate the managed system in such a configuration.
Automatic redistribution of work when a processor fails
If the server firmware detects that a processor is about to fail, or if a processor fails when the processor is not in use, then the server firmware creates a serviceable event. The server firmware can also unconfigure the failing processor automatically, depending upon the type of failure and the unconfiguration policies that you set up using the Advanced System Management Interface (ASMI). You can also unconfigure a failing processor manually using the ASMI.
When the server firmware unconfigures a failing processor, and there are no unassigned or unlicensed processors available on the managed system, the processor unconfiguration can cause the logical partition to which the processor is assigned to shut down. To avoid shutting down mission-critical workloads when your server firmware unconfigures a failing processor, you can use the HMC to set partition availablity priorities for the logical partitions on your managed system. A logical partition with a failing processor can acquire a replacement processor from one or more logical partitions with a lower partition-availability priority. The managed system can dynamically reduce the number of processors used by shared processor partitions with lower partition-availability priorities and use the freed processor resources to replace the failing processor. If this does not provide enough processor resources to replace the failing processor, the managed system can shut down logical partitions with lower partition-availability priorities and use those freed processor resources to replace the failing processor. The acquisition of a replacement processor allows the logical partition with the higher partition-availability priority to continue running after a processor failure.A logical partition can take processors only from logical partitions with lower partition-availability priorities. If all of the logical partitions on your managed system have the same partition-availability priority, then a logical partition can replace a failed processor only if the managed system has unlicensed or unassigned processors.
By default, the partition-availability priority of Virtual I/O Server logical partitions with virtual SCSI adapters is set to 191. The partition availablity priority of all other logical partitions is set to 127 by default.
Do not set the priority of Virtual I/O Server logical partitions to be lower than the priority of the logical partitions that use the resources on the Virtual I/O Server logical partition. Do not set the priority of IBM® i logical partitions with virtual SCSI adapters to be lower than the priority of the logical partitions that use the resources on the IBM i logical partition. If the managed system shuts down a logical partition because of its partition availability priority, all logical partitions that use the resources on that logical partition are also shut down.
If a processor fails when the processor is in use, then the entire managed system shuts down. When a processor failure causes the entire managed system to shut down, the system unconfigures the processor and restarts. The managed system attempts to start the logical partitions that were running at the time of the processor failure with their minimum processor values, in partition-availability priority order, with the logical partition with the highest partition-availability priority being started first. If the managed system does not have enough processor resources to start all of the logical partitions with their minimum processor values, then the managed system starts as many logical partitions as it can with their minimum processor values. If there are any processor resources remaining after the managed system has started the logical partitions, then the managed system distributes any remaining processor resources to the running logical partitions in proportion to their desired processor values.
