Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: separate data/inidex

Re: separate data/inidex

From: Niall Litchfield <niall.litchfield_at_dial.pipex.com>
Date: Fri, 26 Apr 2002 13:04:13 +0100
Message-ID: <3cc9423d$0$8514$cc9e4d1f@news.dial.pipex.com>


"D.Y." <dyou98_at_aol.com> wrote in message news:f369a0eb.0204251308.6c241085_at_posting.google.com...
> If the data or index you want to read is always next to the disk head
> then you've achieve the best possible configuration for performance.
> Well, in a real application this almost never happens. Separation of
> table and index segments was intended to keep disk heads from jumping
> between segments. However, the access pattern of multi-user applications
> is inherently random, and that normally negates whatever performance you
> hope to gain by separating segments. Let's say your application simply
reads
> index then table, and reads index again then table again, with no
interference
> from other sessions, and you have very limited cache ... Can you still say
> separating table and index won't make a difference?
>
> So the old rule (I didn't invent it) is not just a myth, it has its
merits.

In a single user environment without file systems, caching controllers, system processes (CKPT for example writing to each datafile ever 3 seconds). maybe. But then one wouldn't be using ORACLE for that environment anyway.

--
Niall Litchfield
Oracle DBA
Audit Commission UK
Received on Fri Apr 26 2002 - 07:04:13 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US