Promoting critical jobs

An overview of the mechanism that is behind the promotion of critical jobs, or of their predecessors, running on IBM Z Workload Scheduler Agents, including pools and dynamic pools.

The promotion of a critical job is the process whereby the execution of a job in the critical path, that is about to miss its deadline, is provided with more resources. Thus, it is redirected to a more powerful or faster dynamic agent, to idle workstations in a pool or dynamic pool, to an agent allocated to the execution of critical work.

As soon as IBM Z Workload Scheduler determines that a job flagged as critical is not going to meet its deadline, it automatically updates the definition of the job, and of as many of its predecessors as necessary (that is, all the jobs in the critical path of the critical job), with the tws.job.promoted and tws.job.resourcesForPromoted promotion variables. These variables set the dynamic workload broker component to perform the load balancing required to guarantee that the job is run on time.

If the critical job is to be processed in a dynamic pool, the variables must be added manually to the job definition. For details, see Promoting jobs scheduled on dynamic pools.

To control the priority allocation of system resources to the jobs in the critical network that must be promoted, you must configure the PromotedNice or PromotedPriority property of the native job launcher section in the JobManager.ini file of your IBM Z Workload Scheduler Agents.

When a native job is promoted, the TWS_PROMOTED_JOB environment variable is automatically added with value set to YES in the job definition. You can possibly take advantage of the value (YES or NO) of this variable in the script/program run by the native job, to implement a prioritization mechanism.