Re: Need a consultant to find the performance issues

From: Dave Herring <gdherri_at_gmail.com>
Date: Tue, 25 Sep 2018 12:46:23 -0500
Message-ID: <CAFN=diCycaDskczgvd_PKdGOdrhL+VxeH+knCrXGHbDL3mR9Dg_at_mail.gmail.com>



While I don't have a recommendation for a consultant (or company) but I do suggest that you and your team rework the expectations of the candidate. As you listed it you quite possibly may get good results but you risk the effort being spent on "improvements" which may end up being "throw more hardware at the solution and we're done" vs. really analyzing the perceived issue and addressing that first, regardless of what the solution may be.

For example, my assumption is that you have 1 or more production issues that are negatively affecting production and need additional help in addressing them. In light of that I believe it'd be important to look for someone who:

  • Can effectively analyze the current state and clearly define what the problem is/are.
  • Have expertise in knowing how to track down issues at various levels.
  • Apply understanding of the problem(s) and address the highest impact first, then proceed to the next highest impact.
  • Accurately demonstrate / prove the impact of any changes made.

Maybe I'm being to picky or playing with semantics but I see too often changes suggested/made that involved almost no understanding of the real issue and that masked the underlying problem which eventually came back. For example, I frequently see tickets come in complaining about performance and the app team wants statistics gathered on x number of tables. Stats are then gathered and suddenly performance is better (for a while). Were the underlying objects really stale? Did the change in statistics cause a different xplan to be chosen or did just flushing the statement (as a result of stats being gathered) cause a new/better plan to be chosen? Without knowing this, the next thing you know is any time a performance issue arises everyone rushes to gather stats regardless of what the real issue is.

Dave

On Tue, Sep 25, 2018 at 9:27 AM Joan Hsieh <joanhsieh08_at_gmail.com> wrote:

> Hi listers,
>
> We are in need of recommendations for an oracle consultant or a consultant
> company who can manage to solve the following issues that we have had on
> the production in the past year.
>
> The qualified candidate would be needed to be able to do many things
> including:
>
> - Review our existing architecture, make suggestions for
> improvements for both the production hardware and software
>
> - Suggest what patches and settings would better improve
> performance
>
> - Review past outages, the upgrades and improvements we have made
> and assess them
>
> - Find the root cause/causes of performance problems that can't
> be adequately addressed by simply “fix this sql” followed by “fix that
> sql”, as the problems are more widespread than that.
>
>
> I think that’s the general idea of the scope of expertise we are looking
> for. Thanks for your help.
>
> we are located in Marlborough, MA
>
> Joan H
>

-- 
Dave

--
http://www.freelists.org/webpage/oracle-l
Received on Tue Sep 25 2018 - 19:46:23 CEST

Original text of this message