Re: Oracle hints

From: joel garry <joel-garry_at_home.com>
Date: Wed, 30 Jan 2008 11:18:25 -0800 (PST)
Message-ID: <84499a0e-f827-4d41-96d2-c2d6f6541abf@e10g2000prf.googlegroups.com>


On Jan 30, 11:01 am, bdbafh <bdb..._at_gmail.com> wrote:
> joel garry wrote:
> > On Jan 29, 6:05 pm, Peter Teoh <htmldevelo..._at_gmail.com> wrote:
> > > For a one time query, and if I were to make many of these, caching in
> > > the V$SQLAREA is not necessary.
>
> > > Is there any Oracle hint to allow me to do that?
>
> > > Thanks.
>
> > v$sqlarea is just a view.  Oracle has to hard parse the new query
> > anyways.  There really isn't anything sensible one could do to avoid
> > that.
>
> > If there are many singular queries, they'll just get aged out fast.
> > If you are reusing other sql just after it is getting aged out, you
> > need a bigger shared area anyways.
>
> > Sybrand probably understates the issues with cursor_sharing.
> >http://download.oracle.com/docs/cd/B19306_01/server.102/b14211/memory...
> > Look at the bulletins and bugs found when searching for cursor_sharing
> > on metalink.
>
> > I can't always tell if Mladen is joking.
>
> > Can't help but think it is a valid wish to be able to tell Oracle
> > "hey, don't even bother trying to keep this query around.  And no
> > peeking!"
>
> Joel,
>
> do you mean that in addition to the
>
>  /*+ cursor_sharing_exact */ hint,
>
> to have a hint along the lines of:
>
> /*+ cursor_sharing_exact_then_flush */
>
> -bdbafh

Not really, since the idea is not to share at all because it is unique. Hopefully, this could help Oracle not have to go through all it's checking. I have no idea if this is realistic, on the surface it seems simple enough.

jg

--
@home.com is bogus
If wishes were horses
beggars would ride
If turnips were watches
I'd wear one by my side.
Received on Wed Jan 30 2008 - 13:18:25 CST

Original text of this message