We're running Oracle, not DB2 - but 'large' is 'large', and our results
might help.

Our production database is running at 90 GB and takes about two hours to
back up, using three concurrent sessions to three tape units. We run
this direct to tape, and not through a disk pool. In addition, our test
environment DB is also 90 GB; this backup runs about five hours, running
two sessions to tape (this one uses 100 Mbit ethernet, not the SP
switch).

My ADSM server is a wide node (old style - 77 MHz); we've got two
Quantum DLT-7000 tape units on each of two SCSI boards. So far, we've
been able to run all four drives at full rated speed with this hookup,
and are planning one more SCSI board and two more tape units for later
this year. Any further expansion will require us to use a second server,
or, more likely, to go to a high node for more I/O slots.

Tom Kauffman
Sr. Technical Advisor
NIBCO, Inc.

>----------
>From:  Terry Allan[SMTP:Tallan1308 AT EMAIL.MSN DOT COM]
>Sent:  Thursday, February 26, 1998 5:42 PM
>To:    ADSM-L AT VM.MARIST DOT EDU
>Subject:       ADSM on SP2 backing up large DB2 database on mutiple nodes
>
>Is there a high-water mark for the number of nodes than can be backed up by
>1 server?
>I gather there can be multiple ADSM streams backing up the same DB2
>database.
>At what point do you have to move to a second server?
>I (as you can gather) am not an SP2 giant (or DB2 talking to ADSM, either).
>