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: Tuning a full outer join

Re: Tuning a full outer join

From: Mikito Harakiri <mikharakiri_at_ywho.com>
Date: Wed, 4 Jun 2003 17:17:03 -0700
Message-ID: <O4wDa.10$W71.189@news.oracle.com>


Can I suggest using oracle syntax

select --+ use_merge(x y) merge_aj
  *
from x,y where x.table_name = y.table_name(+)

? Then plan doesn't seems to have any antijoin.

"Richard Kuhler" <noone_at_nowhere.com> wrote in message news:EGvDa.60612$x67.2680144_at_twister.socal.rr.com...
> Daniel Morgan wrote:
> <snip>
>
> > It would appear that you never ran DBMS_STATS and produced statistics
for
> > the CBO.
>
> I was showing that example only as a demonstration of how you apparently
> can't influence the optimizer for the 'anti-join'. I'm not trying to
> tune that query specifically. My real database is well analyzed and
> indexed.
>
> > And I'd be far more concerned about the apparent lack of indexes, or at
> > least their usage,
> > then how Oracle might choose an execution path.
>
> My actual query relies on results that can't be indexed (they come from
> pipeline table functions).
>
> Anybody have an answer for my actual question? How do you influence the
> 'anti-join' method of a full outer join query?
>
>
> Richard
>
Received on Wed Jun 04 2003 - 19:17:03 CDT

Original text of this message

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