Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> RE: Resetting execution plan

RE: Resetting execution plan

From: Laimutis Nedzinskas <Laimutis.Nedzinskas_at_landsbanki.is>
Date: Thu, 30 Nov 2006 16:34:40 -0000
Message-ID: <5A8896FB2AFC5445A7DCFC5903CCA6B06FEBA6@W03856.li01r1d.lais.net>


Oracle can get away with that.  

Actually, it could be that Oracle had realized what was happening and was changing the direction already.
Instead of repeating the old mantra "gather statistics" Oracle 10g documented/introduced many useful hints which enable to control CBO quite well. For example, I was impressed by the ability to hint not a particular index but any available index on specified columns. Naming subqueries (qb_name) is another nice feature of 10g.      


From: oracle-l-bounce_at_freelists.org
[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Wolfgang Breitling

At 08:36 AM 11/30/2006, Laimutis Nedzinskas wrote:

	Ahhhr. 
	It sucks so much this empiric time-variant CBO thing. Out of
always bad and sometimes bad I prefer always bad.          

        Ask any electronics engineer what problem he prefers: the constant one or no-no-but-yes-and-no-and-yes-and-no. I hope Oracle will realize it one day.

Don't count on it. In my view it is only going to get worse in Oracle 11 when they activate the object cache hit statistics (it's already in 10 but disabled by default) and other dynamic CBO features.          

Fyrirvari/Disclaimer
http://www.landsbanki.is/disclaimer

--
http://www.freelists.org/webpage/oracle-l
Received on Thu Nov 30 2006 - 10:34:40 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US