RE: CPU COSTING

  • From: "Khedr, Waleed" <Waleed.Khedr@xxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Fri, 23 Apr 2004 10:25:46 -0400

So for reads (i/o0 it measures the logical reads that might get =
translated to physical or not. It's affected by the buffer cache, and =
this makes sense.

So as you said, any changes/tuning to oica and oic (opt_ind_cache) might =
need to be altered again.

Waleed

-----Original Message-----
From: Jonathan Lewis [mailto:jonathan@xxxxxxxxxxxxxxxxxx]
Sent: Friday, April 23, 2004 10:08 AM
To: oracle-l@xxxxxxxxxxxxx
Subject: Re: CPU COSTING



"CPU Costing" specifically estimates the CPU time=20
needed for all operations, so there is a 'CPUCycle
count' for a logical I/O, a CPUCycle count for comparing
locating a row in a block, a CPUCycle count for
skipping over each column in a row to find the right
one and so on.

Loosely, when CPUCosting is enabled it's part of
enabling the CPU cost model, normally controlled by=20
parameter _optimizer_cost_model, which can be set
to IO, CPU, or CHOOSE.

So in fact, the I/O costing changes at the same time,
so that (in particular):
    cost of t/s =3D blocks to scan / recorded MBRC *
                        mreadtim / sreadtim +=20
                        CPU cost.

I've got a paper coming out on OTN in a couple of
weeks that talks about it a bit.


Regards

Jonathan Lewis
http://www.jlcomp.demon.co.uk

The Co-operative Oracle Users' FAQ
http://www.jlcomp.demon.co.uk/faq/ind_faq.html

April 2004 Iceland  http://www.index.is/oracleday.php=20
June  2004      UK - Optimising Oracle Seminar
July 2004 USA West Coast, Optimising Oracle Seminar
August 2004 Charlotte NC, Optimising Oracle Seminar
September 2004 USA East Coast, Optimising Oracle Seminar
September2004 UK - Optimising Oracle Seminar

----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to:  oracle-l-request@xxxxxxxxxxxxx
put 'unsubscribe' in the subject line.
--
Archives are at //www.freelists.org/archives/oracle-l/
FAQ is at //www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Other related posts: