I hope this isn't a stupid question but I can't seem to find a way to
achieve the following: I have 3595 L12 library with 2 drives and 120
cartridges - all scratch. It is defined as a single device class to ADSM.
Currently this is the device that the db backups are also being written to.
I have in place a script to delete old backup volumes so that they don't
consume all the volumes in the library so that there are scratch tapes
available for migration from the backup storage pool. The system has been
in place for 6 months now and is at about 80% capacity with 525 filespaces
defined. My average restore times are getting too long now - about 10 to 15
tape mounts on average for a full restore and I was thinking about turning
collocation on to improve restore performance. My understanding of the
process leads me to guess that the first nights incremental after doing
this is going to consume the remaining scratch tapes - there are only a
handful of tapes in a filling state and 100+ nodes get hit each night, the
first night might not start spreading across additional tapes but the
second night definitely would. My question is, how can I ensure that there
are sufficient tapes available for the database backups with collocation
on? Is it possible to define the L12 as two devices - one with 10 tapes and
the other with 110, for instance, and use the 10 tape device for db backups
and the 110 tape device for the migration destination?

Thanks in advance for any suggestions

Garin Walsh
The Sabre Group