Re: db_file_multiblock_read_count and performance

Hi Ryan,

> 2. bounce the instance to flush all the cache.

if you try to objectify the test by flushing the cache your test table is probably much too small to se any significant effect.

If you have system statistics gathered, consider the

(sreadtim - mreadtim/mbrc) * number of blocks in the table

as a theoretical accelaration in *miliseconds*!

On my PC I need to read some 30.000 blocks (8K) to reach a difference in order of seconds while using mbrc 1 and 128

hth

Jaromir D.B. Nemec
http://www.db-nemec.com

• Original Message ----- From: <ryan_gaffuri_at_comcast.net> To: "Post, Ethan" <Ethan.Post_at_ps.net>; <Oracle-L_at_freelists.org> Sent: Monday, December 06, 2004 8:50 PM Subject: RE: db_file_multiblock_read_count and performance

--
http://www.freelists.org/webpage/oracle-l