Not sure how deleted records helps concurrent inserts.
Perhaps also some application uses deleted records to 'preallocate' space.
Not very sure but I seem to remember (long time ago), a comment from an
IBMer about that one of the advantages of a file with REUSDLT(*YES) was
improved performance at the moment of inserting a record, as it was faster
to just insert it into a "slot" in, say, the middle of the file, that going
all the way to the end and insert it there (and extend the file, I suppose).
So I suppose that for a process that, say, delete some hundred/thousands of
records and then create new ones in their place, it would be better NOT to
reorganize the file...
I think that I agree with Chuck that (*maybe*) it would be better to do
RGZPFMs in a very selective way. Of course, every shop is different, and
your mileage can vary...
IBM Certified Systems Expert — eServer i5 iSeries