>> On Mon, 11 Jun 2007 23:50:20 +0100, Neil Schofield <neil.schofield AT 
>> YORKSHIREWATER.CO DOT UK> said:


> - We still need to take into account the overhead of the reclaimable space
> on a virtual tape. This can be managed by varying the reclamation
> thresholds, but not eliminated.

With a pure disk VTL, you can keep reclamation thresholds more
stringent than would otherwise be the case, modulo pending status.

It's entirely possible to pack huge tracts of land into PENDING, if
you set reclaim threshold low enough.


> - A pending delete volume will still occupy an underlying disk capacity
> equivalent to its size.

Yes.  You want this. :)

> - Since the conversion of a pending delete volume to a scratch tape takes
> place purely in the TSM database, a virtual scratch tape will also occupy
> the full disk space on the VTL. until it is re-used.

This I don't know about.

> So am I correct in thinking that in the whole scratch, filling, full,
> reclaim, pending, delete, scratch lifecycle of a storage pool volume, the
> only time that we get the benefit of the over-provisioning is when it's
> filling?

If the volume continues to use space when it's returned to scratch,
then it's only at the -first- filling cycle you get that.  That
doesn't seem quite right, methinks.


- Allen S. Rout