Why use resource groups?

Start of change Resource groups are a way of limiting or guaranteeing general purpose resource capacity. A resource group is a named amount of CPU capacity on general purpose processors that you can assign to one or more service classes. For most systems, you can let workload management decide how to manage the resources in the sysplex and not use resource groups. You set performance goals for work and let workload management adjust to meet the goals. End of change

In some cases, however, you might want to use a resource group, for example, to limit the service that a certain service class can consume. Note that it is recommended that each resource group is assigned to only one service class. Using a resource group is appropriate, for example:

If you have an SLA that charges for a fixed amount of CPU capacity and you want to ensure that no more is used.
If you want to ensure that some work with discretionary goals receives some minimum amount of capacity.

Generally, it is recommended to evaluate whether using a resource group fulfills your requirements best, or whether you let workload management take care of managing the resources.

Note:

The sysplex capacity values of the resource groups apply to general purpose processors only and not to specialty processors. WLM manages resource groups based on their consumption of general purpose processor capacity. End of change

Start of change For a resource group, you specify either a minimum or a maximum amount of general purpose processor capacity in unweighted CPU service units per second, or both. End of change

Figure 1 shows some examples of resource groups.

Figure 1. Resource groups

In Figure 1, the BATHOG service class is assigned to the LIMIT resource group. BATHOG might include work that consumes processing capacity in huge amounts, so you assign it to a resource group, and limit it to a maximum of 800 CPU service units per second. Also, the goal assigned to BATHOG is discretionary.

The TSOMED service class, on the other hand, is associated with the PROTECT resource group because, according to an SLA, it is contracted to be guaranteed a minimum amount of processing capacity. So the PROTECT resource group is assigned a minimum of 1000 CPU service units per second. Then, when there is sufficient TSOMED work in the system, and TSOMED is underachieving its goal, TSOMED is guaranteed at least 1000 CPU service units per second. Note that if TSOMED is surpassing its goal, then the minimum capacity setting has no effect. Workload management then processes work from both service classes, making sure that TSOMED gets its minimum amount of capacity, and making sure BATHOG does not consume more than its assigned maximum. Keep in mind the service class goal: if BATHOG is assigned a stringent goal, the goal may never be achievable within the LIMIT resource group capacity. You should determine whether the resource group capacity fulfills your purpose, or the goal. Using both controls could introduce some conflict, and the resource group controls will prevail.