Oracle Backup Strategy re: online redo log backup

From: Nate Jones <nathan.jones_at_bi-tech.com>
Date: 19 Mar 2002 14:25:05 -0800
Message-ID: <cde4fecc.0203191425.4de465ad_at_posting.google.com>


Hello everyone,

I have been trying to improve my backup strategy, which right now involves putting all tablespaces into Hot backup mode (none of these clients are 24/7) and then backing up the entire file system via some UNIX util like fbackup.

First, thank you Tom, Howard, Jonathan (and others) for you continued help in these forums and your websites. I think I would have drowned in the Oracle pool a couple years were it not for your help!!! BIG thanks!

Secondly, (now that I have buttered you up) I was reading some of the doc on Howard's website
(http://www.hjrdba.com/backuppdfs/inconlinelogs.pdf) and I think I have missed something. In one of the pdf's about the online redo logs, Howard makes very clear that they should never be backed up or restored, particulary the current log. That all makes sense.

What I what am not clear on is the procedures during a restore. Let's say, for the sake of argument, that all of my dbf and rdo files exist on one drive (with the redo logs being multiplexed on a separate one) and that drive fails. I am left with one copy of the current redo log, and no dbf files.

So, my restore procedures would be to (once the file system has been created again) restore all of the dbf files and then, using the backup controlfile (to trace done every night), recreate the control files, correct? But dont the redo logs get recreated during when recreating the instance? Or do they know to automatically use the current log that existed before the instance crashed?

Howard's pdf make it seem that as long as you dont ever overwrite the redo logs, you can come up current to the point of failure. I guess I am looking for a little reassurance.

Thanks in advance!!

Nate Jones
Sungard Bi-Tech
nathan.jones_at_bi-tech.com Received on Tue Mar 19 2002 - 23:25:05 CET

Original text of this message