Our TSM server (RS6000 S7A) is attached to an STK silo
with 22 tape drives.  The problem that we are
experiencing, beyond library and drive reliability, is
with the apparent migration and recall process.
Because we have many tape drives, we have dedicated 8
drives to HSM only.  We have two HSM enabled
filesystems, /hsmprod and /hsmuser which are local to
the TSM server.  The filesystem of most concern is
/hsmprod (production).  This filesystem is 200GB and
has 90GB written to it nightly.

The problem we are encountering is if we miss a backup
or two the filesystem is full.  The backup process is
relatively fast, we write to disk and to multiple tape
drives.  However, the migration process is extremely
slow.  It appears that this process will only mount
one tape drive for migration.  Yet when users try to
recall data, it will mount multiple tapes and spawn up
to 20 recall processes (as per maxrecalldaemons=20).
Are there any settings to increase the number of
sessions or daemons the migration process will spawn.

Thanks in Advance

Don Avart

__________________________________________________
Do You Yahoo!?
Yahoo! Auctions - buy the things you want at great prices
http://auctions.yahoo.com/