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

Home -> Community -> Usenet -> c.d.o.server -> Re: Scalable Performace - Inserts/Updates

Re: Scalable Performace - Inserts/Updates

From: Nuno Souto <nsouto_at_nsw.bigpond.net.au.nospam>
Date: Wed, 20 Dec 2000 18:10:57 GMT
Message-ID: <3a40f0cb.3970377@news-server>

On Wed, 20 Dec 2000 14:23:12 GMT, Brian Peasland <peasland_at_edcmail.cr.usgs.gov> wrote:

>According to the Oracle8 DBA Handbook by Kevin Loney, page 86, "Each
>database must have at least two online redo log files available to it.
>The database will write to one log file in a sequential fashion until it
>is filled, then it will start writing to the second redo log file."

What's that got to do with groups? It's writing redo log files in sequence, not in groups.

>
>This is especially true if you are running in Archive mode. If there is
>only one redo log group as you have suggested, then you will have to
>wait for ARCH to finish archiving the redo log group before you can
>write to it again. If there is activity on you db, and after the redo
>log fills and needs archiving you will ALWAYS receive the "checkpoint
>not complete" message while ARCH is copying your one and only redo log
>group.
>

Brian, you're completely confused. Redo log files are written sequentially within each group, and at the same time (in parallel or "mirrored" if you prefer) between groups.

Archiver kicks in after each redo log file is finished with, not after the group is finished with. It copies file by file, not "all files in a group". If you have two groups and therefore two redo logs get switched, then archiver will copy both of those. But not by group.

Groups exist to prevent "accidents" in systems that don't have disk mirroring facilities. That's what they were introduced for. Back I believe in V6 or V7 can't remember exactly which one, was a long time ago. With things like EMC's, they are completely redundant.

For something like this requirement (fastest possible commit) the worst thing you can do is create multiple groups. One (the default) and that's it. Within this one create as many redo log files as you want or need, of course. But let the EMC do its job, don't try to replace or supplement it by adding software-based redundancy.

Cheers
Nuno Souto
nsouto_at_bigpond.net.au.nospam
http://www.users.bigpond.net.au/the_Den/index.html Received on Wed Dec 20 2000 - 12:10:57 CST

Original text of this message

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