Backing up to a disk storage pool of some kind will help. Plain JBOD
disks are fine. The closer you get to 300gb of space in your disk
storage pools, the fewer tapes you will use. The TSM Administrators
Guide says that the ideal is to have your disk storage pool large enough
to accomodate all data backed up in an average day. Even if you can't
achieve that, and migration has to kick in several times during your
backup window, you will still achieve more efficient tape utilization by
backing up to disk and letting migration write the tapes.

Go through your activity log and look at tape mounts, and see if you can
tell when and why they are being mounted. Do a

select session_id,client_name,mount_point_wait,input_vol_access,-
output_vol_access from adsm.sessions

periodically to see which client backup session has which tape mounted
and you can probably figure this thing out. That tape that only gets
0.7% full might already be mounted when the next client starts to back
up that might fill it efficiently, so a new tape is mounted.

If you've been following this list, you know I've been struggling with
disk storage pools filling up. One side effect I see is that, when
client nodes back up directly to tape, I get a LOT more "Filling" tapes.
These are absorbed by themselves when I do my job, watch disk storage
pool fullness, control migration thresholds properly, etc. In theory you
should have the same number of Filling tapes as your MIGPROCESS setting,
although usually in practice there are 1 or 2 more.

You have to wonder why it seems to act as though collocation were in
effect, when client nodes back up directly to tape. The only answer I
have is, "That's just how it works." The system was designed to be run
most efficiently with large disk storage pools.

Roger Deschner      University of Illinois at Chicago     rogerd AT uic DOT edu
======I have not lost my mind -- it is backed up on tape somewhere.=====


On Fri, 17 Oct 2003, Cecily Hewlett wrote:

>Please help me.
>
>I am backing up 7 nodes, running TSM 5.1 on AIX 5.1 and AIX 4.3.3.
>Total data backed up = +- 300 GB.
>I am using an LTO3583 library, with 3 x 3580 drives, across a SAN.
>
>Backup times are great, but TSM is being v. wasteful with tapes
>using up to 7 tapes every night, some of them only have 0.7%
>utilization.
>
>I am not using collocation, or a copy pool.
>I tried to use a diskpool on my shark and then move the data to tape,
>but , SANergy is not supported by IBM.
>
>Does anyone have any suggestions.???
>
>Cecily Hewlett
>