Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: redo logs tuning?
"Andrew Zhilenko" <andrew_at_nextra.com> wrote in message
news:39B29400.64F07199_at_nextra.com...
> Hello!
>
> I have installed Oracle 8.1.6.2 database on Sparc E220
> with Solaris 7. Right now I mostly do some PL/SQL development.
> I have "Spotlight on Oracle" running almost all the time. And
> almost all the time it complains, that something is wrong
> with the REDO log buffers:
> - Log buffer space waits is XX % of active waits
> - Log switch wait time is is XX % of active waits
> - Average redo log write time is 65.50 ms
> ....
> and explanation says that this operation should be much faster.
> (For example my average time is 65 ms, and it says that "modern
> disks should have this value less than 20 ms")
>
> I have LOG_BUFFER=160K, three redo log files 2M each, no
> archivelog. Can I improve situation somehow?
>
> WBR, Andrew
> --
> Andrew Zhilenko, AZ283-RIPE, System Administrator
> Nextra Czech Republic s.r.o.
> tel.+42-02-96159436 fax. +42-02-96159422
> e-mail: andrew@nextra.com http://www.nextra.cz
> Vaclavske namesti 4 - Praha 1 - 110 00
> Czech Republic
IMO, 2M is too small, and you should better go as a minimum for 4 redolog
files of 5M each.
You can verify how often a switch occurs by checking v$log_history. Most
likely it is quite frequent (shouldn't be more than one in 15 minutes).
To change the redologfiles you need to drop and to recreate them. If a
redolog file is active you can force a switch by issuing alter system switch
logfile.
Also, don't forget to check log_checkpoint_interval in init<sid>.ora. In the
5M situation it should be 10000.
Hth,
Sybrand Bakker, Oracle DBA Received on Sun Sep 03 2000 - 14:03:34 CDT