Re: IO performance question

david hill wrote:

>Hi guys,
>
>I'm trying to investigate performance of one our db's where db file
>sequential read are 90% of waits
>
>

This means that your application is reading blocks from indexes. Is performance of the
application unaccpetable? If so, you should set the event 10046, level 8 in the session
executing the application and see what took the longest (tkprof with sort=exeela will give you that)
and tune that particular statement.

>Oracle 9.2.0.6, on a RHEL 3, dual Xeon box
>Everything's on a raid 5, 7+1, Hitachi SAN
>
>

Wow! RAID-5. You should join the battle against any raid five (http://www.baarf.com). You should also
read few articles from www.hotsos.com

>So with a few test's here what I came up with
>I can do a full table scan I can read 225,525 disk reads in 40 seconds
>
>
>

Full table scans don't wait for db file sequential reads. They wait for db file scattered reads. In this case
you may want to not use index.

-- 
Mladen Gogala
Oracle DBA
Ext. 121


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