Re: inefficient sql

  • From: rjamya <rjamya@xxxxxxxxx>
  • To: "Taylor, Chris David" <ChrisDavid.Taylor@xxxxxxxxxxxxxxx>
  • Date: Wed, 30 May 2012 10:36:15 -0400

Chris,
Capturing binds is a good idea, perhaps I should think about that as well.
Using binds while comparing execution stats, hmm perhaps we can then use
this data to identify if we suddenly ran into some data-skew ? if such
patterns arise, dbms_stats settings can be tweaked as well if possible.

I should stop thinking now ...

Raj

On Wed, May 30, 2012 at 10:31 AM, Taylor, Chris David <
ChrisDavid.Taylor@xxxxxxxxxxxxxxx> wrote:

> Good point by Raj.  I have an internally developed snapshot capture that I
> can run over a 24-hour period, or a week or whatever I desire.
>
> The upshot of this data I capture is that I can run a report off of it to
> list SQL statements by executions and elapsed time per execution.  That way
> I can focus on the SQLs that get executed the most that take the most
> elapsed time per execution.  (Focusing on what the users are waiting on)
>
> It's ugly (at least to me) but it's handy as I also capture all the bind
> variables for the executions.  The binds snapshot table can grow very large
> very quickly.
>
>


--
http://www.freelists.org/webpage/oracle-l


Other related posts: