Re: Archive Log Creation Rate??

From: Richard Foote <richard.foote_at_bigpond.com>
Date: Thu, 02 Oct 2003 13:03:08 GMT
Message-ID: <gmVeb.133690$bo1.108644@news-server.bigpond.net.au>

"Charles Davis" <cdavis10717_at_comcast.net> wrote in message news:a-GdnTG0_8tG0eaiU-KYuA_at_comcast.com...
> Hi, all.
>
> I am looking for opinions about sizing ReDo logs vs the rate at which
> Archive Logs are created.
>
> Oracle suggests archiving once every 20-30 minutes.
>
> Others suggest archiving more frequently to minimize potential data loss
and
> for a shorter recovery/warm start when reading the most recent archive
log.
>
> Either approach affects the ReDo sizes specified for the average database
> taking into account the usual logged update activity.
>
> I was cutting a 200MB archive log every 5 minutes, then I increased ReDo
to
> 1GB to archive less frequently, more in line with what Oracle suggested.
>
> What's your opinion or experience? I'd love to hear it.
>

Hi Charles,

One thing I would suggest is to read up and research how Oracle now implements incremental checkpointing (since 8i). It kinda negates many of the arguments regarding how one should size redo logs with regard to both instance recoveries and checkpointing issues. I remember discussing all this a while ago so a quick search on google should provide some information.

The key parameters for this are fast_start_io_target (8i) and fast_start_mttr_target (9i).

Cheers

Richard Received on Thu Oct 02 2003 - 08:03:08 CDT