Re: performance problems after increasing redolog size

Hi Bud,

The performance problems are not because of the change in size of the redo logs but probably because of the fact you have now started to archive the buggers and poor little, struggling, abused, I-cant-cope-I want-out-of-here drive D is feeling somewhat stress about it.

To test, simply put the database back to noarchivelog mode leaving the redo logs at 10M and see if performance improves.

Also, based on the frequency of log switching I would look at making the logs somewhat bigger still.

Finally, storing your files and redo logs (both online and archived) on the same drive is always somewhat questionable from both a performance (as you're experiencing) and from a recoverability point of view.

Get those new disks of yours ASAP.

And two may not be enough !!

Good Luck

Richard

"Bud Socks" <bud_socks_at_gmx.de> wrote in message news:b47e2h$1t3puv$1_at_ID-2182.news.dfncis.de...
> Hi there,
>
> first of all thank you very much for the numerous messages.
>
> Of course i owe you some more informations:
>
> We have 4 Oracle 8.1.7 instances running on a single cpu (PIII750) Winnt4
> machine with
> 1 GB ram.
> There are two drives created on a raid level 5 (drive c:, d:).
> No other internal drives for e.g. raid 1 are available !!!!
>
> All Oracle files including datafiles, redologs, archives and even the
> controlfiles are located
> on drive d: !!!!!!
> (don`t beat me - i know it's sheer lunacy! - but i have to administrate it
> and did not set
> it up that way)
>
> The client application running on top is a "small" CRM-solution for a
> call-center !
> Two of the oracle instances are the productive databases - that means each
> of it has its own
> CRM application & call-center !
>
> Both of them are having the archive mode & automatic archival enabled plus
> one more
> test-database running in the same mode (but low/less user activations).
The
> last database
> is very infrequently visited - no archive mode enabled !
>
>
> As i have taken over this machine there were just 3 redolog groups with
just
> one 1MB member !
> So far the archive mode was still disabled.
>
> Almost frequently the client-server connection has gone lost - several
times
> a week.
>
> By having a view on the alertlog i noticed heavy logswitching.
> Therefore i decided to modify the redolog configuration as a first step on
> my troubleshooting path.
> (due to sequrity/recovery issues i have enabled the archive mode as well)
>
>
> 2 times a week (not scheduled - just triggered by a user) an import from
> another host system is
> started. The amount of datas do almost have the same size (thats what i
was
> told).
> As the redolog size was set to 1MB this import has taken two hours
runtime.
> Now after the modification (4 members / 10MB each) the import takes more
24
> hours.
>
>
> Unfortunately i don`t have the init<sid>.ora available now - cause i´m not
> at work.
>
> But please find below a part of the oracle alert log taken yesterday.
>
> What problems do you see ?
>
>
>
> By the way i have requested 2 hd`s to build up a raid1 to place all
redologs
> on this logical disc - although
> it is just a interim solution.
>
>
>
> Thank you very much
>
>
> Bud
Received on Thu Mar 06 2003 - 08:03:58 CST