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: OCP - Performance Tuning

Re: OCP - Performance Tuning

From: tingl <one4all_at_all4one.not>
Date: Wed, 29 Jan 2003 00:16:56 GMT
Message-ID: <Y3FZ9.978$Wu1.104304@newsread1.prod.itd.earthlink.net>


I think B is the correct answer. C would be correct if FIRST_ROW is not specified.

"Don Burleson" <don_at_burleson.cc> wrote in message news:998d28f7.0301281145.4a901fce_at_posting.google.com...
> It sounds like the question is about why Oracle chooses a
> nested-loop(index access) over a full-table scan access (e.g.
> sort-merge join).
>
> To me, a or b are not correct as there are no absolutes with the CBO!
>
> C - The sort parameters CAN influence this decision because is the CBO
> knows that a disk sort is involved, when he will favor index access.
>
> D - The number of rows in the table also influences the CBO. For
> example, even with optimizer_index_cost_adj=1 and first_rows, Oracle
> will do a sort-merge join on a small table, and an index access on a
> large table.
>
> Hence I vote for C and D. . . .
>
Received on Tue Jan 28 2003 - 18:16:56 CST

Original text of this message

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