ADSM-L

How do you handle time-critical backups?

1999-02-22 14:00:46
Subject: How do you handle time-critical backups?
From: "Kauffman, Tom" <KauffmanT AT NIBCO DOT COM>
Date: Mon, 22 Feb 1999 14:00:46 -0500
I just ran into an ugly situation this weekend. We restored our production
Oracle database to a test server -- and hit a read error on one of the five
DLT tapes (about 60 GB on each tape). I ended up re-trying this tape on all
drives, to no avail. And, of course, without the 9.2 GB that failed, the
entire database was unusable.

There were *no* errors logged when the tape was created.

It takes about 4.5 hours to run the backup; I get one 12-hour window per
week for an off-line (and the bulk of the 12-hour window is dedicated to OS
upgrades, Oracle reorgs, and anything else that may require a shutdown).

How do I avoid this problem in the future? Is there *any* platform which
supports dual-write or mirrored tape (I could do this on GCOS-3 15 years ago
. . .)?

Ideally, I'd like to verify the validity of the backup before turning the
system over to anyone else - but the window just isn't there (we don't even
have enough window to copy this -- the originals are marked 'unavailable'
and shipped off-site).

TIA

Tom Kauffman
Sr. Technical Analyst
NIBCO, Inc.