Re: hint injection in 10g??

From: Richard Foote <richard.foote_at_bigpond.com>
Date: Sat, 24 Jan 2004 13:15:23 GMT
Message-ID: <LduQb.26071$Wa.11347@news-server.bigpond.net.au>

"Andrey" <adm_at_mlm.ru> wrote in message news:butdq6$6ia$1_at_gavrilo.mtu.ru...
>
> Really, Oracle must revert back to rule-base optimizer and do not mess
with
> stupid outlines and something-horrible-injection. Very simple to make -
> user-written pre-sql-parsing, where with simple SQL-code user can alter
> statments, passed to server. This can return to Oracle it's former
> reputation as managable and predictable server.
>

Hi Andrey,

Revert back to RBO, I think not.

From 9.2, (and in most cases a good deal earlier), I would suggest that those that don't understand "why" they have an execution plan and associated performance, don't have a reasonable understanding of "how" Oracle works and "what" is required of them from a manageability perspective.

The CBO works extremely well (with the exception of the "odd" bug ;) if we give it a chance. And yes, that includes misusing these "stupid" outlines which generally should only be used for "migrating" like purposes.

If you really want to go back to the RBO, I have a few Oracle6 disks you might like to borrow ;)

Cheers

Richard Received on Sat Jan 24 2004 - 07:15:23 CST