RE: Database Weirdness........

--openmail-part-096b8b58-00000001

Content-Type: text/plain; charset=US-ASCII; name="BDY.TXT"
Content-Disposition: inline; filename="BDY.TXT"
Content-Transfer-Encoding: 7bit

Steve,

I don't know much about perf tuning, but I would ignore the rdbms/smon/pmon timers as they are just backgound wait events for forground tasks (at least at this point in time). You should instead concentrate on looking at V$SYSTEM_EVENT and V$SYSSTAT before and after these events. The following query will quickly give you the top CPU consumer in AIX:

#! /bin/ksh
#
# Print details of the top 20 CPU process
#
# sorts by cpu %, then memory usage %
#

ps u | sed 1q; ps gu | sort -nr +2 +3 | sed 20q

'ps aux' is also an equivalent.

Tie this in with 'vmstat' (to look at CPU, Paging stats) and 'iostat' (for disk-wise IO stats). Why don't you just log these values once every hour using a wrapper and try and correlate them. It is essential to tie OS and DB stats together. You could also look at Network stats using 'netstat'. Please bear in mind that the first set of values from vmstat/iostat are cumulative upto that point, while subsequent sets are for the measurement period!

In any case, are you using Async I/O and if so, what are your MAX/MIN Servers? What is your disk and controller config? And what else are you running on the Server? What is the Hardware config? How big is your RAM/swap space and how much of it is used (lsps -a)?

IBM has some good perf. measurement/analysis documents at 'http://www.redbooks.ibm.com' (free to all). Search for 'RS/6000 Performance'.

Hth,
John Kanagaraj
Brunei Shell Petroleum
http://www.geocities.com/john_sharmila

God so loved the world that He didn't send a committee! (See John 3:16 for details)
** Opinions expressed here are solely mine and not necessarily those of my employer **

--openmail-part-096b8b58-00000001

Content-Type: application/ms-tnef; name="WINMAIL.DAT"
Content-Disposition: attachment; filename="WINMAIL.DAT"
Content-Transfer-Encoding: 7bit

<binary content removed -- do not send binaries to the list> Received on Thu Jun 29 2000 - 20:34:17 CDT