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: count(*) vs. a cursor when determining the existing record on primary key

Re: count(*) vs. a cursor when determining the existing record on primary key

From: Ryan Gaffuri <rgaffuri_at_cox.net>
Date: 17 Jun 2003 09:48:38 -0700
Message-ID: <1efdad5b.0306170848.46fe3e1c@posting.google.com>


Billy Verreynne <vslabs_at_onwe.co.za> wrote in message news:<bcmf5u$3bg$1_at_ctb-nnrp2.saix.net>...
> Slight off the subject..
>
> Select count(*) on a table with 83 million rows (table size close to 20GB).
>
> Time? 18.07 seconds.
>
> Reason.
> a) Table is partitioned.
> b) Local bitmap indexes exist.
> c) CBO using the bitmap index in parallel for the select count(*).
> (no hints used)
>
> Not all problems are nails. And there's more than just the hammer in the
> Oracle toolbox. ;-)

the guy had a count and a group by. not every has partitioned tables. not everyone has bitmaps.

even with that. a simple index lookup is faster, uses less memory, less consistent gets... i dont understand your point here at all. Received on Tue Jun 17 2003 - 11:48:38 CDT

Original text of this message

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