Joined: 21 Nov 2009 Posts: 58 Location: California
To clarify, Prefetch & Multi-fetch do not reduce I/O, and do not deal with "repeatable access to same data."
They are methods by which Adabas will read-ahead for sequential commands. For each physical call to Adabas, multiple records are returned to Natural, thereby reducing inter-region communication between Adabas and Natural. Adabas and Natural do essentially the same amount of work as without Prefetch/Multi-fetch. It's the overhead that is reduced.
When applied properly, Prefetch/Multi-fetch can reduce elapsed time by as much as 80%. When applied improperly, elapsed times can increase by an order of magnitude.
The best scenario is a single, long file access; for example, a data extract or report.
If multiple file accesses are nested, the benefits of Prefetch/Multi-fetch will be reduced or even eliminated.