Please excuse the stupid Lotus Notes quoting.

We have DLT7000s here in a Stk 9714.  Each tape holds 35Gb raw and the 9714
is only a 100 tape library so it's not really practical for us to run node
based colocation. We do have one node though with 53 15Gb filesystems
(Imaging system).  I've enabled filespace colocation for just those 53
filespaces but limited the tape pool to 26 tapes.  It's my understanding
that ADSM will write the first 26 filespaces to the 26 tapes and then go
back over the tapes writing a second filespace to each tape, and so on.
It's not that much worse than true filespace colocation but the tapes are
better utilised.

I haven't had it running long enough to see exactly what will happen when
the data expires and reclamation starts.  If anyone has done this I'd be
interested in your experiences.

Carl.






Paul Zarnowski <vkm AT CORNELLC.CIT.CORNELL DOT EDU> on 22/08/98 08:31:00 am

Please respond to "ADSM: Dist Stor Manager" <ADSM-L AT VM.MARIST DOT EDU>

To:   ADSM-L AT VM.MARIST DOT EDU
cc:    (bcc: Carl Makin/CBR/IPAustralia)
Subject:  Re: Suggestion for ADSM enhancement to reduce the need for
      collocation




[re colocation]
every day during migration from disk to tape (we use collocation).  This
increased wear on tapes means that we have to replace tapes more frequently
than we might otherwise have to.  We're still uncertain about this, but it
seems a likely possibility.  Given that new technology tapes aren't cheap,
this can run into a significant expense.  If there were a way to lessen
this problem, it would result in a savings for us in replacement tapes.