RE: Oracle, Siebel and rule-based optimization

[Cunningham, Gerald]

Title: Message

Hi there,

 

We run quite a bit of Siebel here, against everything from 7.3.4 to 8.1.7. They all run rule-based. We have set optimizer mode to CHOOSE and analyzed tables in the past - performance got dramatically worse!

 

You can probably get some individual statements to run better cost-based, but the app as a whole does run better rule based.

 

One thing to watch out for - Siebel creates all objects using default storage clauses ("to save space" I was told by a Siebel consultant, since there are typically hundreds of unused tables). Go with LMT to be safe. I learned the hard way, with a database that contained more than 2 MILLION extents. Had to rebuild with LMT... even coalescing a TS took days... (learned some internal stuff tho')

 

 

- Jerry

 

 

-----Original Message-----
From: Vergara, Michael (TEM) [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 30, 2003 12:51 PM
To: Multiple recipients of list ORACLE-L
Subject: RE: Oracle, Siebel and rule-based optimization

Michael:

 

Would you please post a summary of findings here on the list?  I

too have a Siebel implementation going on, and we are

experiencing performance issues, and have set all the

Siebel-specified initSID.ora parameters.

 

I am in California, and the database is 'across the pond' in

Europe, so I am doing this remotely, and about 8 hours out of

phase with my users.  Any knowledge transfer will be appreciated.

 

Thanks

Mike

 

-----Original Message-----
From: Michael Fontana [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 30, 2003 7:41 AM
To: Multiple recipients of list ORACLE-L
Subject: Oracle, Siebel and rule-based optimization

I would like to ask a question of this esteemed and knowledgeable audience about running
SIEBEL applications on Oracle.  Siebel insists that their CRM application will only run with
acceptable performance if rule-based optimization is chosen, yet, as we all know, there are
documented examples of poor performance, and strong recommendations to the contrary from Oracle.

At one point, I recall reading an article from a knowledgeable consultant describing what must be done
to get Siebel to run with optimizer_mode=choose.  Any assistance in reaching this goal, or comments,
would be more than appreciated.  We're still in the implementation phase, and are already receiving
poor performance on certain processes, even using the recommended rule based optimization.  I
predict it will only get worse!