Re: log_archive_interval

John Wood wrote:
> First, I would like to thankyou for all the advises you guys gave in this
> post.
>
> I read in some books saying that it would be more efficient for Oracle if we
> can configure that the LGWR would do a log switch when CheckPoint. As the
> default CheckPoint timeout is 1800, does that mean we better have about 30
> minutes for the log switch interval ?

The rate at which you wish to switch logs is largely a function of how much data you are prepared to lose in a worst-case scenario. The current log has never been archived... so if you only switch logs every two hours, potentially 1 hour 59.99 minutes-worth of data might be lost if the current log were to be deleted or corrupted.

> I plan to have to two Archive Log destination, one will be mandatory at
> local drive, say drive E; the other will be optional, at a network mapped
> drive, say V on a different node. This is just to serve as a backup of the
> Archive Log files, as it is recommanded to have two different copies of
> archive log files.

I know what it is intended to serve as. And I know what's recommended. And I'll simply restate that it is NOT recommended, nor actually supported, to archive to a remote destination unless you do it with the log_archive_dest_n='service=xxx' configuration. Mapped network drives don't count.

Don't do it. It will potentially cripple your database's performance, and in any case will invalidate any support contract you might have.

> This way I don't have to back up the database using
> RMAN's command 'backup database plus archivelog', I just need to do 'backup
> database' to save some time in the backup process.

It is a false economy.

The one thing you don't muck about with in Oracle is backing up properly. And mucking about with archiving so that you can muck about with backup commands is just asking for trouble.

Stick to what's tried, trusted, supported and (most importantly) safe.

Regards
HJR Received on Wed Dec 01 2004 - 01:12:27 CST