HIPERBATCH (high performance batch processing) is the term given to the use of a special new variant of PFA caching.
Batch processing often involves a series of processing steps for individual files. For example, in one processing step a (temporary) file is created which in a subsequent step is read again as an input file and used as such. Between two such processing steps, the file is usually closed and reopened.
When a file buffered with PFA is closed
in the case of a write cache, the data belonging to the file in the cache is written back to the disk
the cache management data of this file is released by the software cache handler DAB (i.e. cache in the main memory)
the data belonging to this file which is still in the cache is invalidated.
As long as there is no follow-up processing for this file, this is a highly effective approach. In a subsequent access, however, the file must first be paged back into the cache in a “transient phase” until the application profits from read hits. In the case of batch processing, there is no real need to write the data back to disk for reasons of data security, since the run can be repeated in the event of an error.
This is where the HIPERBATCH concept comes in. A new CLOSE parameter (ADD-FILE-LINK command or CLOSE macro) can be used to specify that at the time of closure the data in the cache should not be written back and – more importantly – not invalidated. A subsequent open operation on the same file can use the data in the cache immediately. The effect is a noticeable acceleration of batch processes with file follow-up processing steps.
The CLOSE parameter can be specified as follows:
Command: ADD-FILE-LINK ...,CLOSE-MODE=*KEEP-DATA-IN-CACHE
Macro: CLOSE <fcb>,*KEEP-DATA-IN-CACHE