> I have been implementing TSM for AIX with a 3494 VTS recently.  We have
> just finished testing at the first site so decided to delete all the data
> from TSM and start afresh when we went live next week.  After the deletion
> it all looks fine in TSM.  We had around 4,000 virtual volumes in use and
> they have all now been returned to scratch.  So we can see them with q
> libvol but not with q vol.  This is as I would have expected.  On the VTS
> side I am having problems fully understanding what is happening.  There has
> basically been no effect on the tape utilization there.  Thus I have to
> assume that while TSM has returned the VVs to scratch category and thus
> deleted all pointers to the data on those VVs, the VTS does not accordingly
> update it's DB to say the VVs are now empty.  If it did this I would expect
> it to recalculate the utilization of the VTS physical volumes which ought
> to trigger a big reclamation.
>
> Is this behaviour correct or am I missing something?

I have only used the VTS in a mainframe environment. If I remember correctly,
IBM chose to have the contents of virtual scratch volumes behave just like the
contents of real scratch volumes. That is, the contents of a scratch volume
remain intact and readable until the volume is rewritten. As a result,
returning a large number of virtual volumes to scratch status will not, in and
of itself, trigger a flurry of VTS reclamation. Each time the production ADSM
environment uses one of the scratch volumes for the first time the VTS will
discard the data written on the volume during testing. This should eventually
result in the VTS reclaiming all the space occupied by test data.