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: hint injection in 10g??

Re: hint injection in 10g??

From: Andrey <adm_at_mlm.ru>
Date: Mon, 26 Jan 2004 12:06:06 +0300
Message-ID: <bv2lal$pg5$1@gavrilo.mtu.ru>

"Paul Drake" <drak0nian_at_yahoo.com> wrote in message news:1ac7c7b3.0401240737.680dda6f_at_posting.google.com... > "Andrey" <adm_at_mlm.ru> wrote in message

news:<butdq6$6ia$1_at_gavrilo.mtu.ru>...

> > "Venkat" <0002_eveev_at_oohay.moc> wrote in message
> > news:YqiQb.21$jN5.340_at_news.oracle.com...
> > > on 1/23/2004 3:19 PM Daniel Morgan said the following:
> > >
> > > > Domenic G. wrote:
> > > >
> > > >> Someone was telling me that there's a new feature in 10g, similar
to
> > > >> stored outlines that does something like this:
> > > >>
> > > >> Suppose you have a query coming in to the database that you simply
> > > >> can't change, and suppose you have cursor sharing set so that
oracle
> > > >> substitutes bind variables in place of literals.
> > > >>
> > > >> Now, let's say you want to alter the execution plan using any hint,
> > > >> but cannot inject that hint into the original application source
code,
> > > >> apparently there is some feature that now lets you substitute a
very
> > > >> specific "cooked" execution plan in place of the one the optimizer
> > > >> generates using a text match.
> > > >>
> > > >> I know this is very much like stored outlines, but with stored
> > > >> outlines, you can only switch between the high level optimizer
modes
> > > >> ... first_rows, rule, all_rows, and generate the outline.
> > > >>
> > > >> Does this make sense, or was I misinformed? Can this type of
> > > >> fine-grained control be done with stored outlines?
> > > >>
> > > >> Example ...
> > > >>
> > > >> You want to inject a /*+ INDEX(...) */ or /*+ DRIVING_SITE ... */
at
> > > >> the back end because you can't access the compiled application code
to
> > > >> stick it in there.
> > > >
> > > >
> > > > Check out the 10g features at http://otn.oracle.com.
> > > >
> > > Particularly,
> > >

> >
http://otn.oracle.com/products/manageability/database/pdf/twp03/TWP_manage_automatic_SQL_tuning.pdf
> >
> >

> > 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.
>
> Andrey,
>
> do you use babelfish to write this stuff?


Please clarify what is babelfish. If I misuse something (really I do), please write to my personal mail. Thanks.

As for automatic tuning, all what is seems to be truthful in this document (if document or its part can be truthful) TWP_manage_automatic_SQL_tuning.pdf , it is 1) Oracle tries to get rid from third-party applications and 2) all this stuff with autooptimization really makes one thing - it search for absent indexes. After all requered indexes are present, all newly created optimizing modes are needless.

> Pd Received on Mon Jan 26 2004 - 03:06:06 CST

Original text of this message

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