Batch processing

Batch processing means collecting work up and running it in one go, unattended, rather than responding to a person in real time. It is one of the two ways work happens on a mainframe; the other is online transactions.

A classic example is a bank's overnight run. All day, the online systems take deposits, payments and withdrawals. Overnight, a long sequence of batch jobs reads that accumulated activity, applies interest, produces statements, generates regulatory reports and prepares files for other institutions. By the time branches open, the books balance.

Batch work is described in JCL, submitted as a job, and runs in steps. Nobody sits and watches it. Instead it leaves behind output and a return code for each step, which is how you find out afterwards whether it worked.

The term sometimes carries a whiff of the past, but the pattern is everywhere in modern computing too. Anything you would call a scheduled task, a cron job or a data pipeline is batch processing under a different name.