Your Workload Management (WLM) service definition for the observed systems ensures that monitors run at a higher priority than the work being monitored. The Provisioning Manager, together with the infrastructure it uses, must be prioritized so that they perform reliably even in a capacity-constrained situation. On the runtime systems, ensure that the Provisioning Manager started task CPOSERV is classified appropriately.
Either classify the Provisioning Manager and the supporting subsystems into SYSSTC, or give them an aggressive single period velocity goal at an importance level that is higher than the work that is defined in your provisioning policy. In particular, the following subsystems and address spaces are required:
Network latency in the CIM protocol can be minimized by specifying numeric IP addresses or by defining entries in the hosts file for the observed systems.
Other contention factors that could affect the ability of the Provisioning Manager to operate efficiently must be minimized. The Provisioning Manager and the CIM servers rely on the z/OS® UNIX file system. Severe contention must be relieved by appropriate tuning measures to allow Capacity Provisioning to operate reliably.
The consumption of resources (processor or virtual storage) by the Provisioning Manager largely depends on the domain and policy definitions. The following factors increase resource consumption by the server:
Most Provisioning Manager processing is zAAP-eligible. zAAPs are used automatically on systems configured with them and no additional definitions are required.