You may want to investigate an entirely different approach to both
optimizing your use of big tapes (Ultrium, SuperDLT...) in the library,
and to disaster backup. Run a second TSM server. It could even be a
second image on the same physical machine. (In our case, it will be on a
second machine in a physically remote location from the main server.)
Send your Database backups to "Virtual Volumes" on the other server,
which then handles them as ordinary Client Archive Files, keeps track of
them in its database, and migrates them to its tapes (stacking as many
on one tape as will fit) when its disk archive storage pool gets full.
You can also use this for your backup copy of the client storage pools.

Your TSM server can also be a regular backup/restore client of this
second server, so if it has a failure, you could restore the OS and TSM
executables from it as an ordinary TSM client.

This is obviously an oversimplification of how this works, but it is all
described in the TSM Administrators Guide for the V4.2 server. This new
feature appears to offer some interesting advantages for both improving
disaster recovery capabilities, and for reducing the number of tape
volumes needed for prudent database backup.

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu