Re: Oracle8 panic! - lousy performance/high load

From: Ronald <devnull_at_ronr.nl>
Date: 19 Nov 2001 09:13:25 -0800
Message-ID: <67ce88e7.0111190913.7fc67b92@posting.google.com>

Tom Karlsson - Remove NOSPAM <tk_at_babylon5.dkNOSPAM> wrote in message news:<Pine.LNX.4.21.0111131904310.24342-100000_at_sith.ammulti.dk>...
> Hi Jim :)
>
> On Tue, 13 Nov 2001, Jim Kennedy wrote:
>
> > You have 1 disk drive for data. Your IO is highly limited.(probably about
> > 60 to 70 io's /sec) You need to spead the load.
>
> Actually, disk IO seems quite healthy. atsar shows this:
>
> Linux foobar 2.4.5 #2 SMP Thu May 31 12:45:22 CEST 2001 i686
> 11/13/2001
>
> 00:00:00 device read/s rdKb/s write/s wrKb/s rdwr/s
> _disk_

Hi Tom,

I would check v$sqlarea for high numbers on buffer_gets compared to rows_processed and executions. I suspect table scans of tables that are cached. This way you can find the badly performing sql. Using sql*plus and the autotrace feature you can easily debug the statement.

Ronald.

---

http://ronr.nl/unix-dba
Ben je een ervaren dba en wil je dit ergens anders laten zien, mail me dan. Received on Mon Nov 19 2001 - 11:13:25 CST