Re: ORACLE Backups

From: Roderick Manalac <rmanalac_at_oracle.com>
Date: 21 Aug 1994 21:16:19 GMT
Message-ID: <338g73$f2v_at_dcsun4.us.oracle.com>


Keith_at_keithl.demon.co.uk (Keith Leng) writes:
> I agree that it is good practice, and it is advice that I would
> follow myself. But I still think that the result would be the
> same whether the start-up was done before a backup or after a
> restore. I don't understand what the manual is saying in that
> quote. I would be grateful to anyone who can enlighten me.
> The backup includes the control and log files so I can see no
> reason why the recovery procedures wouldn't work after a restore.
> As far as I know, the timestamps on the various header blocks
> must be consistent with each other, but have no relation to
> "the current point in time".
>

In theory, what you're saying is correct. It becomes a question of whether you want to pay the price of doing instance recovery before making the backup or after restoring a full snapshot of the instance. I can think of a couple of snags.

First, if an upgrade of Oracle occurred between the time of backup and restore, it may be necessary to also restore the old version of Oracle to do the instance recovery. Hopefully, it will not be necessary in the future, but customers who performed a shutdown abort of a 6.0.34 instance, then installed the 6.0.36 maintenance release, were bitten by the fact that the redo log format was changed.

Also if something goes wrong with the backup/restore of the redo logs (e.g. bad tape), the whole backup could be considered useless (unless archived versions of the logs exist).

There are probably a couple of other issues that make restoring from a shutdown abort inflexible that I haven't thought of yet.

Roderick Manalac
Oracle Corporation

Disclaimer: Speaking for myself not my company. Received on Sun Aug 21 1994 - 23:16:19 CEST

Original text of this message