Defining service classes for kernel work

A service class is a group of work that has the same service goals or performance objectives, resource requirements, or availability requirements. For workload management, a service goal and, optionally, a resource group is assigned to a service class.

When you define a service class for forked child address spaces, it should normally have three performance periods because it must support all types of kernel work, from short interactive commands to long-running background work. You can set duration values using the service-units-per-second value reported in the RMF™ Monitor I Workload Activity report.

The next example is a sample service class for forked child processes. Change these values as appropriate for your installation.

* Service Class OMVS - OMVS forked child processes

  Base goal:

    #  Duration   Imp  Goal description
    -  ---------  -    ----------------------------------------
    1   2000      2    Response Time  80%  1 second
    2   4000      3    Response Time  60%  2 seconds
    3             5    Execution velocity of 10

Also, define a service class for daemons. This service class should normally have only one period with a velocity goal higher than the velocity goals of other forked child processes.

* Service Class OMVSKERN - OMVS startup processes
  Base goal:

    #  Duration   Imp  Goal description
    -  ---------  -    ----------------------------------------
    1             1    Execution velocity of 40

Your installation might have other special classes of users. If so, you might want to define other service classes for kernel work.

If you have used the PRIORITYGOAL statement in the BPXPRMxx member of SYS1.PARMLIB to enable the nice(), setpriority(), and chpriority() functions, additional service classes for kernel work must be added. See Controlling dispatching priorities for details.