Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.misc -> Re: Performance issues when running in archive redo log mode

Re: Performance issues when running in archive redo log mode

From: Brian Peasland <peasland_at_edcmail.cr.usgs.gov>
Date: Wed, 3 Nov 1999 14:11:30 GMT
Message-ID: <38204292.8918F249@edcmail.cr.usgs.gov>

> One of the consultants that we are working with claims that their tests have
> consistently shown that there is a significant performance hit when ORACLE 7
> (running on UNIX, different flavors, one of them being NCR UNIX), is set to
> operate in archive redo log, as oposed to non archive redo log, mode (the
> performance hit is said to be 30-40%). This confuses me, because I had the
> obvious misconception that the only difference (as long as you do not
> actually issue "alter tablespace begin backup") is that you do not re-use
> the redo logs, but "archive" them, so that they can be used later in the
> recovery process. I certainly did not expect such a big hit, which basically
> stops us from using normal on-line backup procedures, even though
> availability has to be 7x24. For performance reasons as well, the consultant
> recommends not using replication either, which leaves us with the only
> choice of a proprietary application replication scheme, in order to provide
> high availability and database maintenance capabilities (provided by the
> consultant). Can anyone confirm that these claims make sense?
> Thank you for your response!

From a more technical standpoint: In Archive mode and non-Archive mode, the redo logs are being written to. When you are running in archive mode, the ARCH process is responsible for archiving filled redo logs. This process was created to ease the burden on the RDBMS as a whole. While the ARCH process will add some overhead to your overall system, the resources required aren't that significant.

One reason that I can think of that they may have experienced performance issue is if they didn't have enough redo log groups. Or maybe the redolog groups were too big. If the system has written to all redo log groups and ARCH hasn't had enough time to archive the logs, the system will "hang" and wait until ARCH has written a log to disk. Once a log is archived, it is now available for use by the system. If the redo logs are very big, it can take a long time to archive the log. If you only have a small number of log groups, you may also run into this problem. Adjust the size and number accordingly.

I recently converted a system from non-Archive mode to Archive mode and hardly noticed any difference in my database's performance.

HTH,
Brian Received on Wed Nov 03 1999 - 08:11:30 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US