ADSM-L

Logmode (Was: Recovery Log size)

2001-06-25 05:55:20
Subject: Logmode (Was: Recovery Log size)
From: Reinhard Mersch <mersch AT UNI-MUENSTER DOT DE>
Date: Mon, 25 Jun 2001 11:55:51 +0200
> At 08:35 AM 6/22/2001 -0600, Kelly J. Lipp wrote:
> >Doesn't this bring back the issue of Roll-Forward vs. not?  I guess I'm
> >still in the not camp with no compelling reason to join the other.  Also,
> >for those in the other camp, I'm thinking if your environment is so large
as
> >to fill a 13 GB log in a 24 hour period perhaps the environment is busting
> >at the seems in other areas and should perhaps be split anyway.
> >
> >Even if larger logs were supported, how long would a db restore take with
> >roll-forward enabled?  I'm thinking way too long.
>
> I don't believe this is a pure issue of database size.  Rather, I think it
> is the combination of performance characteristics of the server, the
> network, and the client - the entire environment.  roll-forward makes sense
> when the time it takes to run an incremental backup is much less than what
> it takes to run a full backup.  If you can afford to run a full backup
> every day, then go for it.  Your recovery time will be short.  The time it
> takes depends not just on DB size, but also tape speed, processor speed,
> processor memory, etc.  When we upgraded our server, we found that database
> backups ran faster - no surprise there.  We run roll-forward because we
> don't have enough wall clock time to perform a full backup every day, but
> we want the protection.  While recovery time is an issue, backup time can
> be too and it is in our environment.

Paul,

why does logmode=normal imply performing full backups every day?
I guess, your statement has to be seen in the context of Kelly's remark
regarding restore times; clearly, having to restore just one full backup
would be advantageous. We also cannot afford to run full backups every day,
so we run daily incrementals. We nevertheless live with logmode=normal,
because, as Lindsay Morris already stated, roll-forward probably gives you
nothing, if the DB is mirrored.

The only situation, where roll-forward seems important to me, is when
the DB is not mirrored. The DB must be big then, because otherwise it
would be better to spend the extra log space needed for roll-forward for
the DB mirror.

Reinhard

--
Reinhard Mersch                        Westfaelische Wilhelms-Universitaet
Reinhard Mersch                        Westfaelische Wilhelms-Universitaet
Zentrum fuer Informationsverarbeitung - ehemals Universitaetsrechenzentrum
Roentgenstrasse 9-13, D-48149 Muenster, Germany      Tel: +49(251)83-31583
E-Mail: mersch AT uni-muenster DOT de                       Fax: 
+49(251)83-31653
<Prev in Thread] Current Thread [Next in Thread>