Re: redo log stats

Another misleading naming convention

   'redo log space requests' - is only for disk space related waits, not the sga redo log buffer

To simplify things, I'd say don't look at the table (for redo tuning)

     v$sysstat
but look at

    v$system_event
This table gives you all you need to know (v7.3+ for before v7.3 see http://www.geocities.com/oraperf/redo/log_file_space_switch.html)

Here is a tuning example
http://www.geocities.com/oraperf/redo/redo_tune_rules.html

On how to use

   v$system_event
see
http://www.geocities.com/oraperf/seminar.html

Best
Kyle

In article <JQ556.1122$Wo1.3641_at_nntpserver.swip.net>,   "Mr Taki" <nospam_at_localhost.spamwarn> wrote:
> Hi all,
> I've seen conflicting "facts" regarding the Oracle statistic called
> 'redo log space requests' and what actions to take to resolve the
 problem.
> Some say it pertains to waiting for redo log buffer space (in SGA)
 and some
> say to waiting for redo log file space (file), each with different
 actions.
> (Btw, my "gut" tells me that it is the latter, i.e. waiting for a
 complete
> log
> switch so that the next file/group can be used. But then I can think
 of a
> few different problems causing this, like too small files, or DBWR
> performance problem.)
>
> Can somebody bring forth explanation to the actual fact, whether it
 is the
> redo log file or the log buffer, and some source reference then I'd be
> really glad :-)
>
> Regards
> /Fad
>
>

Sent via Deja.com
http://www.deja.com/ Received on Thu Jan 11 2001 - 17:50:20 CST