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: Using multiple buffer pools

Re: Using multiple buffer pools

From: Nuno Souto <nsouto_at_optushome.com.au.nospam>
Date: Sat, 28 Jul 2001 10:08:53 GMT
Message-ID: <3b628c04.2455118@news>

On Fri, 27 Jul 2001 13:06:24 -0400, Chuck Hamilton <chuck_hamilton_at_yahoo.com> wrote:

>I currently have everything in the DEFAULT pool with about a dozen
>segments assigned to KEEP. The KEEP pool size is slighly larger than
>the sum of the blocks used by those objects. Those objects are tables
>that are scanned very frequently but are small in size. Is this a good
>usage of multiple buffer pools?
>--

Sure. In my case, I make the KEEP much smaller than the sum of blocks of all tables assigned to it. All I'm interested in is getting the sub-set of blocks that is being used at the moment to stay in memory for as long as possible. I get 99.99993% cache hit rate this way and am quite happy with it.

Most of my volatile tables (temp tables that get dumped-in for reports, calculations, whatever) are in RECYCLE, which is really small but not as small as recommended in the manual. I get in excess of 75% hit rate there.
DEFAULT gets the remaining and hoovers around 96-98%.

Before (one buffer pool) I was getting consistently <80%... This way I ended up with much better buffer use without having to commit oodles of memory to buffers (same global number as before).

Cheers
Nuno Souto
nsouto_at_optushome.com.au.nospam Received on Sat Jul 28 2001 - 05:08:53 CDT

Original text of this message

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