Topics in this domain

Job Scheduling and Batch Processing

Operations · 2 min read

The night shift

A bakery runs to a strict timetable overnight: mix the dough, let it rise, bake, cool, pack. Bake before it’s risen, or pack before it’s cooled, and the batch is ruined.

Most transaction processing, interfaces, backups and reports run as scheduled batch jobs, often overnight. If the wrong job runs, a job runs twice, or one fails without anyone noticing, data can end up wrong or missing in every downstream system.

Load transactions, calculate interest, post to ledger, reports and backup, each waiting for the previous job.

Job scheduling software runs the jobs automatically in the right order and at the right time. It enforces dependencies (a job waits until the one before it finishes successfully), reduces manual operator intervention, and logs everything that runs.

The controls

Controls over the schedule, over who can change it, and over failures.
ControlDetail
Authorised scheduleOnly approved jobs are in the production schedule
Change managementNew or changed jobs and schedules are approved and tested
Restricted accessFew people can add, change or delete jobs; operators can run but not edit
DependenciesJobs wait for the ones before them to complete successfully
Monitoring and alertsFailures and unusual run times are flagged
Exception handlingFailures investigated; restarts from checkpoints
Rerun authorisationReruns approved and logged, with checks for duplicate processing
Ad hoc jobsJobs run outside the schedule need the same approval and logging
LogsJob logs kept and reviewed
Output checksRun-to-run totals and reconciliations confirm nothing went missing

A scheduled job can run with very powerful rights, so access to the scheduler is effectively privileged access. Review it like admin access.

Operators

Operators run the approved schedule and respond to failures. They shouldn’t be able to change job definitions, program code or input data, and their console activity should be logged.

Reruns

When a job fails halfway through, rerunning it from the start can process the first half twice.

Example. An interest accrual job fails halfway through. An operator reruns it from the start without authorisation. Half the accounts receive interest twice. A rerun procedure requiring approval, a restart from a checkpoint, and a run-to-run total check would have prevented it.

How the exam asks about it

If the question asks…The answer is usually…
Main benefit of automated job schedulingReduces errors from manual intervention; enforces sequence
Control over changes to the production scheduleChange management with approval
Risk of unauthorised rerunsDuplicate processing
Who should be able to modify job definitionsA restricted, authorised group, not operators
Evidence that jobs ran as scheduledScheduler logs reviewed for exceptions

In short

  • Job scheduling software runs batch jobs automatically in the right order and at the right time, reducing operator error.
  • Only authorised, approved jobs should be in the production schedule; changes to schedules go through change management.
  • Access to the scheduler is restricted, because a scheduled job can run with powerful privileges.
  • Job outcomes are monitored; failures are investigated and reruns are authorised and logged.
  • Job logs and exception reports are reviewed to detect unauthorised or failed jobs.
  • Dependencies between jobs must be defined so later jobs don't process incomplete data.

Exam traps

  • Operators shouldn't be able to change job definitions or programs, only run approved schedules.
  • A rerun without authorisation can duplicate transactions.
  • Ad hoc jobs submitted outside the schedule need the same approval and logging.
  • Scheduler access is effectively privileged access; review it like admin access.

Slides

Click the slide and use the arrow keys. F toggles full screen.

CISA · Domain 4 · Operations

Job Scheduling and Batch Processing

The automated jobs that run overnight: why the schedule must be controlled, who may change it, how jobs depend on each other, and what happens when one fails.

1 / 8