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: Full table scan and select count(*)

Re: Full table scan and select count(*)

From: <ctcgag_at_hotmail.com>
Date: 21 Apr 2003 16:15:13 GMT
Message-ID: <20030421121513.405$al@newsreader.com>


Rick Denoire <100.17706_at_germanynet.de> wrote:
> I was trying to find the best value of db_file_multiblock_read_count
> in order to improve full table scans.
>
> So to test the effect of this setting, I started issuing the command
> select count(*) from tabq;
> tabq hast several Million rows.
>
> I got different results that surprise me a lot:
> PARAMETER Command Time
> 64 as above 12 sec
> 64 /*+ FULL(TABQ)*/ 57 sec

Oracle can count the number or rows using an index, as long as at least one of the columns in the index is not null.

Xho

-- 
-------------------- http://NewsReader.Com/ --------------------
Usenet Newsgroup Service              New Rate! $9.95/Month 50GB
Received on Mon Apr 21 2003 - 11:15:13 CDT

Original text of this message

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