Re: Why "Separating Data and Indexes improves performance" is a myth?

From: Nuno Souto <dbvision_at_optusnet.com.au>
Date: Sat, 24 Apr 2004 05:23:03 +1000
Message-ID: <015901c4296a$66c8e2d0$9b00a8c0@dcs001>

Original Message ----- From: <Paula_Stankus_at_doh.state.fl.us>
> So in laying out disks for a new database - perhaps the only thing worth =
> separate out is the redo based on I/O contention - right? =20
>

Not really. If you have a "mostly-read" database, why separate the redo if you rarely write to it? Waste of time and disk.

I guess the real issue in this whole argument of separate I/O is that there is no such thing as a silver-bullet approach to splitting files in an Oracle database. The nature of EACH SPECIFIC case dictates what should be spread across devices and what should not.

To try and find a one-size fits all rule is just condemned to failure.

The problem is: Find out if you have I/O contention. The solution is: IF you have I/O contention, find out where and spread THAT load.

If that means you have to allocate more disk to the system tablespace, or to the redo, or to a single table, or to a group of indexes, or whatever is your problem area, then so be it.

Trying to make a rule of thumb out of something that has got no hands is really hard... ;)

Cheers
Nuno Souto
in sunny Sydney, Australia
dbvision_at_optusnet.com.au

Please see the official ORACLE-L FAQ: http://www.orafaq.com

To unsubscribe send email to: oracle-l-request_at_freelists.org put 'unsubscribe' in the subject line.

--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------

Received on Fri Apr 23 2004 - 14:42:10 CDT