From: Sriram Kumar <>
Date: Thu, 24 Jan 2013 15:05:27 +0530
Message-ID: <>

Hi Chris,

     cursor_sharing in 11gr2 is not recommended/supported any more. cursor sharing is a bad workaround that allows developers to write SQL's without bind variables. performance is then dba's responsibility.

from a change management perspective, I would like to keep all the environments same. have faced some bugs, cursor_sharing=force gave wrong results (in 10gr2). sure you would see them in metalink now also.

best regards

sriram kumar

On Wed, Jan 23, 2013 at 12:13 AM, <> wrote:

> I'm curious how many of you guys have set CURSOR_SHARING=FORCE from EXACT
> in a development or test database (or even production) and what led you to
> set it?
> My only experience with CURSOR_SHARING comes from my PeopleSoft days when
> SIMILAR was thought to be a "good" thing and then found out it wasn't (at
> least I believe that's the way it went).
> I've never played around with FORCE.
> The reason I'm asking is that we have 4 QA instances that share the same
> server and one of these QA boxes has the requirement to match production as
> far as memory pools, parameters etc.
> This has the adverse effect of forcing the other 3 into smaller RAM
> buckets. I'm getting 4031 errors intermittently and [of course] it's
> related to non-shared sql blowing my sql area and library cache sizes all
> to he**.
> (Keeping in mind this is a very immature application - much work going on
> in redesign - while that is happening though I really would like to keep
> the 4031 errors down to a minimum)
> Regards,
> Chris Taylor
> Oracle DBA
> Parallon IT&S
> --

Received on Thu Jan 24 2013 - 10:35:27 CET

Original text of this message