Re: redo log stats

"Jonathan Lewis" <jonathan_at_jlcomp.demon.co.uk> wrote in message news:979110606.27978.0.nnrp-14.9e984b29_at_news.demon.co.uk...
>
> When the log buffer is too smal, you start
> to get log buffer waits which are not due
> to log switches - remember I said 'one of the
> reasons for .... is log switches' When the
> buffer is small, it can get filled very rapidly
> and then everything waits whilst it is written
> out, then restarts.
>

Didn't this part of the thread focus in only on the case of requests "during" a log switch since four posts ago? :-) Well, for a proper context, perhaps one must look also at other event stats (waits).

> >> Having a too big redo buffer is more likely to lead to high
> >> values in v$session_wait for 'log sync' and 'log write'.
 

> >
> >I was under the impression that log sync waiting is posted by
> >COMMITing transactions only.
> >

(Probably I was stuck here on that log switch, hmm, still.)

>
> Correct - but if the buffer is big, and lots of items go into it
> before a single commit occurs, then a lot has to be written
> out before the sync is complete. In this case I meant
> 'large amounts of time' rather than 'large counts' when I
> used the expression 'high values'.
>

No good for response time.

In your experience, what redo log buffer sizes/size range are feasible?

Regards
/Fad Received on Wed Jan 10 2001 - 20:02:01 CST