We went through a similar process recently. This required us to schedule 24
hours of down time to do a DB reorg. Unfortunately this is not an online
process. We saw the numbers change from 100GB DB that was 33% used to a
100GB DB that is 22% used. Our goal was to set the maximum size to 50GB. We
were able to recover 50GB after this process and now this is a 50GB DB with
45% used space. We are running on the S/390 with version 4.2.

The length of this process leads me to believe we need smaller DB sizes. A
24 hour period without backups is not acceptable. The alternative seems to
be to keep adding TSM servers, but an initial look at that model would lead
us to between 75 and 100 TSM servers to keep the DB size in the 10GB range.