Here is the backup info for one of our tsm db's . . . .

tsm db size:  25,505,789 used pages
tsm:  tsm 5.1.7
os:  AIX 5.1
processor:  7038-6m2    2p2g
tape drives:  3590h
disk subsystem:  emc symmetrix with raid-S (yuck, but is working well)
backup duration:
    2.5 hours to virtual volume on another tsm server at remote site
    1.5 hours to local disk
restore duration:  around 1.5-2 hours if from local disk

I've performed about 4 restores.  If restoring from a local disk backup I
expect
the restore to run just a little longer than the backup.  If restoring from
a virtual
volume on a remote tsm server, the restore basically depends on the network
bandwidth available.  In general, virtual volume restore is about the same
time
as the backup, or faster if the network and remote tsm server aren't busy.


Sorry, I don't have much idea about why your restore is running so slow.  A
guess is that
the writes to disk can't keep the dlt drive streaming, causing it to
shoe-shine.  But you
indicate that the disks are new and can stream data well.  This to me
implies that for some
reason the disk writes aren't sequential.

If you have the resources and time, try a backup to local disk, take the
backup to another
server with similar disk setup, and perform a restore.  Try and have the
backup on different
drives than the db.  This would tell you what a restore would take without
disk/tape interaction.

Questions for experts:

1)  Does a tsm db backup read the db blocks sequentially down the data
files, or is it
in some type of logical block order that causes random reads all over your
db files?

2)  Does a tsm db restore write the blocks sequentialy down the files, or,
does it put the blocks
back in some logical order that causes random writes?


Rick





                      Richard Sims
                      <rbs AT BU DOT EDU>             To:       ADSM-L AT 
VM.MARIST DOT EDU
                      Sent by: "ADSM:          cc:
                      Dist Stor                Subject:  Re: Why do database 
restores take so long?
                      Manager"
                      <[email protected]
                      .EDU>


                      01/21/2005 08:02
                      AM
                      Please respond to
                      "ADSM: Dist Stor
                      Manager"






Some numbers from another site, with hardware specifics.

My experience of late is that a db backup of 5737933 pages takes about
15 minutes, AIX 5.2, p650, 3590E drives, SCSI, with disks on modern
SSA, RAID 1+0 (mirroring and striping), raw logical volumes. Restoral
takes about an hour. I would imagine that the restoral would take
longer if the db volumes were JFS, given sync processing.
In ADSM 3.1 on an R3U AIX 4.1 system, mostly JFS volumes on older SCSI,
same 3590E drives, the backup used to take 1.5 hours.

    Richard Sims




-----------------------------------------
The information contained in this message is intended only for the personal
and confidential use of the recipient(s) named above. If the reader of this
message is not the intended recipient or an agent responsible for
delivering it to the intended recipient, you are hereby notified that you
have received this document in error and that any review, dissemination,
distribution, or copying of this message is strictly prohibited. If you
have received this communication in error, please notify us immediately,
and delete the original message.