Re: Which plan is better - what COST really means ?
- From: "Jonathan Lewis" <jonathan@xxxxxxxxxxxxxxxxxx>
- To: <mgogala@xxxxxxxxxxxxxxxxxxxx>, <cary.millsap@xxxxxxxxxx>, "'Oracle-L (E-mail)'" <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 14 Jan 2005 22:00:32 -0000
Note in-line
Regards
Jonathan Lewis
http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/seminar.html
Public Appearances - schedule updated Dec 23rd 2004
Cary Millsap wrote:
>If I recall correctly, 10g has some operational memory built into the =
>CBO
>that allows the CBO to use operationally measured response times for a =
>given
>query as an input into computing the next plan.
>
I think you may be remembering the explanation I gave
in one of my Oracle 11! demonstrations to explain how
the HAL optimizer could change its mind in mid-stream.
When 9i came up with v$segstat and v$sql_plan_statistics,
it became obvious that Oracle Corp. could start writing
an extension to the optimizer that could use results from
previous queries, and object-level cache hit rates as the
basis for changing its estimated cost of a query.
--
//www.freelists.org/webpage/oracle-l
Other related posts:
