Your Browser is not longer supported

Please use Google Chrome, Mozilla Firefox or Microsoft Edge to view the page correctly
Loading...

{{viewport.spaceProperty.prod}}

Class scheduler

&pagelevel(5)&pagelevel

Within the framework of job management the class scheduler handles the following functions:

Starting of the batch jobs

As shown in figure 17 in section "Job management", all job schedulers pass the jobs they have released for execution to the class scheduler.
In collaboration with the task management system, a job released by the job scheduler is started immediately by the class scheduler.

Monitoring job class limits

The JMU statement DEFINE-JOB-CLASS and the CLASS-LIMIT operand serve to define an admission limit per job class (maximum number of jobs to be started in this class). Relative to the entirety of job classes, systems support thus specifies a job mix. The class scheduler is responsible for checking that this mix is not violated. When a job is passed to it, the class scheduler checks whether the CLASS-LIMIT has been reached. If this is the case, the job is rejected (unless it is a job with the start attribute IMMEDIATE).
As soon as the situation changes, e.g. as a result of job termination or a MODIFY-JOB-CLASS command to increase the limit, the relevant job scheduler is notified of this event and a job is passed to the class scheduler.

Buffering jobs in the case of system saturation

If the paging area or the main memory is saturated, the class scheduler still accepts jobs from the various schedulers, but buffers them in class-specific queues.
The inclusion of jobs in these queues is based on the FIFO principle. This principle does not apply to jobs with the start attribute IMMEDIATE (known as express jobs). These jobs come first in any queue. If the class scheduler receives more than one express job, the job that has arrived last takes precedence in the queue.

Starting these jobs after clearing the saturation state

Once the saturation state has been eliminated, the class scheduler starts the jobs by selecting a job class with the lowest M value whose class optimum has not yet been reached, using the following algorithm:

M = (C + 1) / W

where:

M

Precedence

C

Number of jobs currently running in the job class

W

Weight (urgency) of job class

The class scheduler starts the job that comes first in the queue.

After every job start and job termination the class scheduler must recalculate the precedence.

If the software product PCS is used, the class scheduler is linked to the performance controller.