What is Batch Processing?

What is Batch Processing?
Many “experts” in the world of IS punditry seem unaware that the AS/400 handles two types of information processing quite well. The traditional scenario involves large volumes of short, fast processing associated with input from 5250 clients or other client/server devices. A similar scenario involves the processing of message-driven jobs that handle one input request at a time.
The second (and lesser known) type of processing that the AS/400 does well is batch. Some business people still believe large volumes of batch work are best left to big, isolated (“glass-house”) mainframes. But in fact, the AS/400 can complete a lot of work at a much lower cost per unit than some of the big guys can achieve.
OS/400 defines several job types, including type B for batch. Another job type is I for interactive, which identifies jobs that work with input data originating from a 5250 workstation or another device that emulates a 5250 data stream. The WRKACTJOB (Work with Active Jobs) command shows an active job’s type designation.
The distinction between type-B and type-I jobs ensures the routing of a work unit to the proper subsystem (e.g., QBATCH for batch, QINTER for interactive). Also, when a type-I job uses the CPU, the amount of processor cycles consumed are debited against the machine’s capacity to perform interactive work. (This value is called the Interactive Commercial Processing Workload (ICPW) rating.) However, when a type-B job uses the CPU, the amount of processor cycles consumed are debited against the machine’s total capacity. (This value is called the Processor Commercial Processing Workload (PCPW) rating.)
Any type of job can process database or other types of data. To process database data, the job may read a record, change it, and write it back to the file; read some data, process it, and write a new record back to the file from which it read the data; or originate new data and add it to either an existing file or a new file. My focus here is improving the performance of batch jobs that process database data (e.g., read, update, add, delete) on a single system (or within a single partition on an logical-partition (LPAR) system).