Re: ksuosstats global area

  • From: Tanel Poder <tanel@xxxxxxxxxx>
  • To: Joel.Patterson@xxxxxxxxxxx
  • Date: Wed, 10 Nov 2010 20:00:40 +0200


You need to show us some numbers and how you measured them! I don't really
understand the issue - is it in CPU usage or latch contention?

If it's CPU usage, then the normal pattern I do is

1) check v$sesstat and v$sess_time_model counters (with snapper) to see
whether the CPU is used mainly for parsing, running the query (logical IOs
or sorting) or something else.

2) if the above don't give enough evidence, then I would look into which
Oracle functions are executed the most directly using process stack sampling

Tanel Poder

On Tue, Nov 9, 2010 at 9:08 PM, <Joel.Patterson@xxxxxxxxxxx> wrote:

> I’ve been using DBOptimizer and using Tanel Poders scripts.
> SID 0 is using a lot of CPU (in a particular test case).  If the latch was
> named ‘shared pool latch’ I’d have an explanation, but LatchProfX is
> revealing that latch ‘ksuosstats global area’ is at the top and ‘shared pool
> latch’ is not listed.

Other related posts: