Understanding the common batch window problem

In many environments, the batch window is a predefined time of day that VSAM spheres can be offline, unavailable for online CICS® processing.

For example, a batch window can be defined every night between 10:00 PM and midnight. During the batch window, a number of tasks might be performed against the offline VSAM data sets, including:
  • Backup creation
  • Report generation
  • Execution of batch jobs against the VSAM data sets

In most scenarios, the batch window is restricted to a predefined time interval. If an error occurs during one of the tasks run during the batch window, the time it takes to recover from the error might prevent the completion of all required tasks before the end of the batch window. In most environments, this could prove very costly to business.

For example, consider the following scenario:
  • An environment has a batch window between 10:00 PM and midnight. During this time interval, backups of all VSAM spheres must be taken and a batch job that updates the VSAM spheres must be run. However, in this example scenario, all backups completed successfully but the batch job encountered an abend during step six.
Figure 1. Batch window error
This shows a sample batch job where the first five steps have been successful. Step 6 failed and steps 7 and 8 are still pending. The next figure shows how you would recover from this scenario.
To recover from this scenario, the user would have to perform the following steps:
Figure 2. Batch window error recovery without CICS VR batch backout
This diagram follows from the previous one. First you restore the VSAM spheres from their latest backups, then you forward recover all the changes that took place for steps 1 to 5. Fix the error in step 6 and resubmit the batch job for steps 6 to 8.
  1. Restore the VSAM spheres from their latest backups
  2. Forward recover the VSAM spheres up to the time that the batch job step, step six, that was being run when the abend occurred started
  3. Fix the problem that caused the abend
  4. Continue execution of the batch job, starting at the batch job step, step six, that was being run when the abend occurred

While this solution successfully recovers the VSAM spheres from the abend that occurred, for some users the amount of time required to restore and forward recover the VSAM spheres would cause processing to exceed the allocated batch window time frame. Therefore some users have elected to use the batch backout process in preference to forward recovery.

To help avoid missing the batch window time frame, and to ease the process of recovering from abends that occur during batch processing, you can use the CICS VR batch backout function. CICS VR batch backout removes all updates that were made to VSAM spheres from a batch job step, or multiple steps, if required. For example, the user could recover from the same batch error that was previously described by performing the following steps:
Figure 3. Batch window error recovery with CICS VR batch backout
In this diagram using batch backout removes all the updates that were made in job step 6 where the error occurred. The error is fixed and the batch job can be resubmitted for steps 6 to 8.
  1. Use CICS VR batch backout to remove all updates made by the job step, step six, that was being run when the abend occurred
  2. Fix the problem that caused the abend
  3. Continue execution of the batch job, starting at the batch job step, step six, that was being run when the abend occurred
Using CICS VR batch backout allows you to recover from this batch job error without:
  • Restoring the VSAM spheres from backups
  • Forward recovering the successful updates made by steps 1-5

Removing the steps listed allows for a quicker recovery from the batch job error. Therefore increasing the chances of successfully completing all required batch window processing in the allocated batch window time frame.