I have an NM ksam file which is used for generic retrieval. I can't use
Image B-tree functions because the ksam file is built as the union of
several datasets.

Currently this file has 9 indices. Basically there are 4 big fields (50-60
bytes) used as indices and 4 additional indices beginning several bytes into
these 4 fields. The 9th index is 8 bytes and is the common element tying
many records together, mostly for maintenance of the file.

Would it be better to leave this thing as it is:
  240 byte records, 9 indices, 250,000 records
Or would it be better to have smaller records with 2 indices and a lot more
records:
  60 byte records, 2 indices, 4 million records

Regards,
John Pickering

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *