Re: sga
Date: Fri, 12 Jun 2009 11:01:08 -0400
Message-ID: <4A326DB4.6090100_at_tufts.edu>
I can attest that this technique works, at least in a problem we were having with an application that wasn't using bid variables. Increasing the sga or shared pool didn't help. Decreasing it made the problem manageable until the next release of the app came along which used more bind variables.
Martin Berger wrote:
> Brian,
>
> I read this as 'all the SQLs are unique, and really big' - am I right?
>
> if they are unique, they MUST be hard parsed every time.
> In such a case, it might be worth reducing the Library Cache, to reduce
> the time every process needs to search through it until it does not find
> a preparsed object and parses itselve.
>
> just an idea, totally untested, still without enough understanding of
> the setup.
>
> br
> Martin
>
>
> On Fri, Jun 12, 2009 at 15:37, Zelli, Brian <Brian.Zelli_at_roswellpark.org
> <mailto:Brian.Zelli_at_roswellpark.org>> wrote:
>
> The form for the application does build some really crappy dynamic
> sql, 2 pages worth. The vendor said "oh you have to add some
> indexes" but that doesn't look to be the case. I seem to be
> leaning toward the cursor issue myself. I'll see what it is set to.
>
>
>
>
>
-- http://www.freelists.org/webpage/oracle-lReceived on Fri Jun 12 2009 - 10:01:08 CDT