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 -> REPOST: Re: Can joins be less efficient?

REPOST: Re: Can joins be less efficient?

From: Ilya Zvyagin 2147483647 <ziv_at_fct.ru>
Date: Fri, 25 Jan 2002 12:45:58 +0300
Message-ID: <3$--$%%%%-_---%%_$@news.noc.cabal.int>

"Vikas" <g_vikas_at_hotmail.com> wrote in message news:30e95949.0201241658.3eeec754_at_posting.google.com...
> Actually, in some cases where you have large tables (think DW with
> 100s of Millions of rows) the join is sometimes more costly than the
> code you are mentioning here. Depends on
> partitioning/indexing/cpus/optimizer/etc....

No, never for ONE row if tables have indices. It might be consequences of bad statistics or bad query plan but this could be managed.

The only case I know is two tables without any indices.

A - Na records
B - Nb records

select B.* from A, B where A.key = B.FK and A.val = @someval

will cost Na * Nb records of IO
and

select @a_key = A.key from A where A.val = @someval select B.* from B where B.FK = @a_key

will cost Na + Nb records of IO

This message was cancelled from within The Unacanceller's glorious new software, Lotus 1-2-3 For Rogue Cancellers. Received on Fri Jan 25 2002 - 03:45:58 CST

Original text of this message

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