Ours is:

       Device Class Name: BACKDISK
  Device Access Strategy: Sequential
      Storage Pool Count: 1
             Device Type: FILE
                  Format: DRIVE
   Est/Max Capacity (MB): 4,096.0
             Mount Limit: 20
        Mount Wait (min):
   Mount Retention (min):
            Label Prefix:
            Drive Letter:
                 Library:
               Directory: G:\DISKVOLS
             Server Name:
            Retry Period:
          Retry Interval:
                Twosided:
                  Shared:
      High-level Address:
        Minimum Capacity:
                    WORM: No
         Scaled Capacity:

We are just piloting an all disk pool - on this particular server we are
storing small files.  Clients backup to a disk stgpool, this migrates to a
sequential file stgpool with a maximum size threshold and then this migrates
to a LTO Tape stgpool.

We currently only have internal storage on our TSM server (HP Proliant ML
530 with 12 internal storage bays).  So we have 12 X 146GB drives.  The disk
stgpool and seq file storage pool is about a 1TB RAID5 array striped across
the 12 drives.

It is possible to pre-define the volumes with DEFINE VOLUME (I have not
tried this).

IBM is looking at allowing a stgpool to span multiple file systems as a
potential enhancement for sequential access disk pools.

-----Original Message-----
From: Levinson, Donald A. [mailto:dlevinso AT ACSALASKA DOT COM]
Sent: August 3, 2004 5:03 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: D2D vs. tape backups with TSM?

Can anyone tell me how they defined their sequential FILE devclass and
how the underlying hardware is setup? It seems like the best way to do
this is to have multiple physical disks and spread the TSM volumes
across the disks in a way to encourage spreading the i/o across the
spindles. But because the devclass uses either the specified directory
or the TSM default directory to build the volumes automatically then it
appears you would be forced to either have a large software or hardware
stripe or RAID volume or else an insane collection of storagepools to
get enough space in the FILE volumes to make them usable.
Maybe it is possible to manually create and checkin the FILE volumes so
that I can create them in specific filesystems?

-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU] On Behalf Of
Rushforth, Tim
Sent: Tuesday, August 03, 2004 12:15 PM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: D2D vs. tape backups with TSM?


I did some restore tests previously and found with my multi-session
restore
tests some of the restores were waiting for a long time for volumes that
were in use.  It was this that made me think that collocation was not
good
for this.  At that point in time I was using 25 GB volumes, I've since
switched that to 4 GB volumes.

I just did a quick test - restoring 46.1 GB, 217,310 objects

With no collocation, the data was spread across 36 volumes.  With
collocation it was spread across 14 volumes.

No collocation restore and 8 mount points: 35.5 minutes, 22.2 MB/sec
With collocation and 1 mount point: 68.9 minutes, 11.4 MB/sec
With collocation and 8 mount points: 34.8 minutes 22.9 MB/sec

So in my case as long as the data is spread out among a number of
volumes
(with collocation or not) multi-session restore works with a higher
throughput than single session restore.

So now I'm beginning to think that collocation may be good as long as
the
volume size is fairly small.

My environment:
TSM Server 5.22 on Windows 2003
TSM Client 5.229 on Windows 2003
The volumes are defined on SCSI disks (not ATA)
Client compression is on

It would be interesting to see what other people are experiencing.

Thanks,

Tim Rushforth
City of Winnipeg
-----Original Message-----
From: Otto Schakenbos [mailto:otto.schakenbos AT TELEFLEX DOT COM]
Sent: August 2, 2004 8:15 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: Re: D2D vs. tape backups with TSM?

Food for thought on multiple sessions and filebased backup:
If you got 2(or more) virt. tapes on one disk(array) and the client has
mounted them both and is pulling data from them this will probarly be
slower in throughput then when just reading from one virt. tape at a
time since sequantial reading is normally faster (esp. with ata) then
random.
In other words multiple mount points means the heads of your disk has to
move longer distances then when using a single mountpoint.
Thats why we have collocation turned on and only allow single sessions
on our file type backup pool.

maybe we are doing it all wrong anyway...

regards

Otto Schakenbos
System Administrator

TEL: +49-7151/502 8468
FAX: +49-7151/502 8489
MOBILE: +49-172/7102715
E-MAIL: oschakenbos AT teleflex DOT com

TFX IT-Service AG
Fronackerstrasse 33-35
71332 Waiblingen
GERMANY

****************************************************************************
*******
3/8/2004
This transmittal may contain confidential information intended solely for
the addressee. If you are not the intended recipient, you are hereby
notified that you have received this transmittal in error; any review,
dissemination, distribution or copying of this transmittal is strictly
prohibited. If you have received this communication in error, please notify
us immediately by reply or by telephone (collect at 907-564-1000) and ask to
speak with the message sender. In addition, please immediately delete this
message and all attachments. Thank you. ACS