Re: Re: SQL performance in prod
Date: Fri, 20 Jul 2018 10:30:12 +0200 (CEST)
Message-ID: <2007398760.7787.1532075412925.JavaMail.webmail_at_bluewin.ch>
Hi Stefan et al,
I do not like the 10053 idea. It means looking into a CBO decision protokoll that did lead to a wrong plan. So have wrong conclusions based on wrong estimations. I rather want to know what made the optimizer stumble. The fastest approach would be cardinality feedback. http://www.centrexcc.com/Tuning%20by%20Cardinality%20Feedback.pdf
BTW: http://www.centrexcc.com will also show a good article on how to read 10053.
Regards
Lothar
----Ursprüngliche Nachricht----
Von : contact_at_soocs.de
Datum : 20/07/2018 - 10:09 (GMT)
An : carlospena999_at_gmail.com
Cc : oracle-l_at_freelists.org
Betreff : Re: SQL performance in prod
Hey CP,
yes, there are several great sources for this.
Understanding the CBO principles:
- Cost-Based Oracle Fundamentals by Jonathan Lewis: https://www.apress.com/de/book/9781590596364
- Some of the newer CBO behavior by Alberto Dell'Era: http://www.adellera.it/static_html/investigations/
Reading CBO traces:
- Chasing the optimizer by Mauro Pagano;: https://de.slideshare.net/MauroPagano3/chasing-the-optimizer-71564184
I guess the latter is the one you are looking for as it explains an analytical approach to digest any CBO trace file, decomposing it into smaller pieces and make it easy to analyze.
Best Regards
Stefan Koehler
Independent Oracle performance consultant and researcher
Website: http://www.soocs.de
Twitter: _at_OracleSK
> Cee Pee <carlospena999_at_gmail.com> hat am 19. Juli 2018 um 22:49 geschrieben:
>
> Nice thread, we get same kind of problems too. Is there a book or paper on how to read the 10053 trace file, like "basics of 10053"?
>
> Reading Tim's response, I think it is a good time someone wrote a book on that topic.
>
> CP
-- http://www.freelists.org/webpage/oracle-l -- http://www.freelists.org/webpage/oracle-lReceived on Fri Jul 20 2018 - 10:30:12 CEST