This chapter describes provisioning that based on CPU consumption, which provides the means to promptly react to high CPC-wide CPU consumption.

Customers can define Capacity Provisioning Policies for provisioning On Off Capacity on Demand (OOCoD) processor capacity through scheduled activations for anticipated capacity bottlenecks. The policies also can be defined for workload-based activations that are driven by WLM service classes that exceed certain PI limits for unexpected capacity shortages.

In situations where customers consider the central processor complex (CPC)-wide CPU consumption as the primary and sufficient evidence for a processor capacity shortage and cannot identify suitable WLM service classes that can trigger provisioning, the new provisioning-based on CPU consumption provides the means to promptly react to high CPC-wide CPU consumption.

Specifications for how Provisioning Manager must react to high CPC-wide CPU consumption require the following definitions in Capacity Provisioning:

Defines a Capacity Provisioning management domain that specifies CPCs with processor capacity to be managed and systems in those CPCs that are used to observe the CPU consumption.

Defines scopes and rules for changes of Processor Capacity, where the following conditions apply:

The new Policy supporting Utilization Based Provisioning can be defined only in z/OSMF for z/OS V2R2.

This section provides a step-by-step basic configuration of Capacity Provisioning.

To use Capacity Provisioning-based on CPU usage, managed CPCs must be specified, including a valid OOCoD Record ID. Also, at least one system on the managed CPC must be defined to observe the CPC-wide CPU consumption metrics.

Figure 9-1 shows a sample CPC and Systems configuration from z/OSMF.

Before defining your policy, you must consider and answer the following questions regarding CPU and capacity management:

The information about how much more Processor Capacity is needed is used to define the maximum extra capacity the Provisioning Manager is allowed to manage and to set up step-by-step increments for utilization-triggered or workload-triggered management.

Use Maximum Processor Scope and Rule’s Processor Scope options to define how much more processor capacity can be provisioned at most. These settings apply to all defined scheduled activations, workload-based activations, and utilization-based activations. Combinations of scheduled activations, workload-based activations, and utilization-based activations never exceed the maximum activation level.

Figure 9-2 shows a sample configuration for Maximum Processor Scope tab.

Then, define where and what kind of capacity must be managed in which situation by using the Utilization Condition tab. This tab can be defined with Workload Condition.

Figure 9-3 shows a sample Utilization Conditions definition.

Next, define where capacity must be managed in one of the following components:

The Provisioning Manager automatically observes suitable systems that are running on that CPC to monitor CPC-wide CPU consumption data.

Also, define the following type of processor that must be managed:

Finally, define the following situations that qualify for provisioning and deprovisioning extra processor capacity, as shown in Figure 9-4 on page 29:

The CPC-wide CPU consumption level of the specific processor type that must be exceeded to trigger provisioning.

The minimum time during which the CPC-wide CPU consumption must exceed the provisioning utilization to trigger provisioning.

The CPC-wide CPU consumption level of the specific processor type must fall below this limit to trigger deprovisioning of more processor capacity.

The minimum time during which the CPC-wide CPU consumption must fall below the deprovisioning utilization to trigger deprovisioning.

Unlike the workload condition, the utilization condition-driven activations are triggered by any workload that increases up the CPC-wide CPU consumption.

Provisioning Manager messages on the console about processor capacity changes it starts and of their successful implementation. It does not inform about the starting action or policy element. Example 9-1 shows a sample message.

The following reports can be created to monitor Capacity Provisioning activation:

The following console commands provide more information about CPU consumption, processor capacity, and policy. This information also can be gathered from z/OSMF:

Displays the current Processor Capacity of a CPC.

Displays the active policy’s utilization conditions.

Displays the CPU consumption and detected demands for increasing processor capacity.

Displays utilization report for all observed CPCs.