Shared Processor Pool is a familiar term to all of you. It's a unique technology that has been an integral part of the power industry for years. But where there's light, there's shadow.

“Shared processor pools do have some limitations. Dedicated processor partitions can not be assigned to a shared processor pool and therefore any unused cycles shared by those partitions are available to any shared processor partition in the system. Shared processor pools have no impact to the affinity placement for the assigned shared processor partitions and can’t be used to group related partitions. Shared processor pools also don’t help to prevent noisy neighbor impacts.”

And this is exactly where the Resource Groups concept comes into play.

Power11 introduces a new feature called Resource Groups that provides isolation for assigned cores by physically limiting utilization to partitions assigned to the group. Users configure the quantity of cores assigned to a Resource Group, as well as assign both dedicated and shared processor partitions.

Use Cases:
- Consolidation across multiple lines of business
- Isolation of production workloads from test/dev workloads
- Improved application performance by grouping workload tiers into resource groups
- System-level isolation in multi-server consolidation scenarios
- Improved performance by mapping Shared Processor Pools into Resource Groups

Values Summary:
- Separate workloads from department A from department B (Private Cloud)
- Separate production from test and development workloads (critical workloads receive dedicated resources, preventing 'noisy neighbor' issues and guaranteeing consistent, predictable performance)
- Grouping tiered workloads
- System Level Isolation when consolidating workloads
- Improve workload running in Shared Processor Pools (can provide improved shared processor partition efficiency by utilizing resource group specific shared processor dispatching optimizations in the hypervisor)
- Better control over CPU resource (Assigning dedicated processor partitions to a Resource Group allows for limiting the set of partitions that can consume donated core cycles to those in the group)
- Improved Virtual Processor Dispatching (when configured appropriately, resource groups can prevent affinity issues caused by off node virtual processor dispatches)
- Workload Affinization