We just picked up an eval copy of TSM HSM and hadn't thought about automating it yet. Personally I'm of the mind that HSM jobs should be run on a monthly basis or "as needed", particularly after the first migration is complete (in our case) since the primary goal for use is to free disk space only.
We're in a situation here where file servers are begin added as needed when disk space gets low and the server is as big as the server boyz will allow (for recovery reasons). We in Storage would like to prevent this "business as usual" habit of growing out the file server population like this and are looking for HSM to help us "stop the madness".
If this HSM product works, we'd migrate off all files not accessed for 90 days or more to start. We're looking for answers from IBM now for these questions.
1. The age of the data migrated seems to be set in the mgt class. Can we assign different management classes to a single file space (not sure I asked that properly),
2. If a migrated file expires, does the stub file go with it?
3. When a user deletes a stub file, does the file get deleted from the filespace?
4. Our first test migration was a volume with 100GB of user files. I chose to migrate all but those file that had been accessed with in 90 days. 75GB was written to tape, but it took 2-3 days to do that. I got 500K/s max during the job (to 3590E drives). I doubt the tape drive was the bottleneck and need confirmation that this is normal performance for that much data.
5. I haven't found the "calculate space saving" option to be of much use. The first time you run it you'd think it would give you a before and after view of your disk but it doesn't. Instead it tells you only what's going to be moved and not the total picture of "this is what you have", "this is what you'll end up with". The screen is probably reporting the info it was designed to give, but it should really give you more information.