by dick scherrer » Wed Dec 24, 2008 12:18 am
Hello,
Creation and usage of alternate index entries usually invloves some trade-off.
There is additional overhead on adds and deletes, but usually a significant reduction in overhead when reading. A very simple example would be the case where there was a file with 5,000,000 records and 9 of them have a particular value for (let's say) PART_NO. With an alternate index, the 9 records could be found directly. With no alternate index, 5,000,000 records would need to be read. If there were many processes/inquiries that needed to read records for some specific part_no, the alternate index might be critical for these inquires to be able to run.
Each alternate index will take some amount of processing overhead as well as a bit of dasd storage. You would need to weigh the pros and cons for each particular situation. When possible, it is helpful to prototype with "real" volumes to help with the decision.
Hope this helps,
d.sch.