T 3.75 Inadequate control of the batch jobs in z/OS

z/OS operating systems are still used to a large extent for running batch jobs. A batch job comprises one or more single steps (job steps).

The inputs to a batch job are either one/several file(s) or corresponding control cards, which are added using the Job Entry Subsystem (JES2/3). The output is also managed using the Job Entry Subsystem.

The control of the batch jobs primarily comprises starting, monitoring execution and checking the result (mostly in the form of a return code). Depending on the return code, it is often necessary to start other batch jobs. The greater the number of jobs and the complexity of executing the jobs, the greater the probability of an error is.

Manual control

During the manual execution of batch jobs there is always a risk that problems will occur during the execution of the batches due to human error. Along with the sequence of execution in time, the dependencies of the batch jobs on each other are affected. With an increasing number of batch jobs to control, the complexity of the entire batch chain therefore increases drastically and will result in an ever increasing number of errors. Manual control therefore has its natural limits.

Time delays, for example, can result in an online process running after the batch job not starting at the right time, or file backups colliding with the online process.

Computerised control (job scheduler)

If a computerised process (job scheduler) is used, the execution of the jobs will be ensured. However, errors can occur if the instructions to this job scheduler have not been properly tested and there are errors in the instructions. Incorrect reactions from the job scheduler can also be caused by incorrectly defined automation during the batch processing.

Example:

© Federal Office for Information Security. All rights reserved.