Re: Data Dictionary Hit Ratio - myth or fact?

  • From: "Allan Nelson" <anelson77388@xxxxxxxxx>
  • To: "Oracle-L Freelists" <oracle-l@xxxxxxxxxxxxx>
  • Date: Wed, 19 Mar 2008 07:33:44 -0500

Just make it an occasion for a mammoth Method-R project.

"Well, we just have to find out what is running slow because the BCHR and
DDHR are indicative of slow end user experience.  Let's find out who is
suffering the most pain and work at it from that end."

or alternatively

"old bad scripts produce old bad advise.  This script does not look at the
AWR, ASH, BDUMP DEST, OPEN CURSOR, and OPEN_LINKs parameters and so are
hopeless"

Allan

On Tue, Mar 18, 2008 at 6:41 PM, Anjo Kolk <anjo.kolk@xxxxxxxxxxx> wrote:

> Yepp (pun intended),
>
> I have lost years of my life fighting this stuff. In a way it is kind of
> dissappointing to see that it is still going on.
>
> For this customer, let me make a prediction:
> 1) The data dictionary cache hit ratio is responsible for all their
> problems
> 2) They will spend a considerable amount of time and effort (and money) on
> this
> 3) They will probably in the end just increase some memory parameters and
> the hit ratio will go up
> 4) They will feel very good about it
> 5) Performance hasn't hasn't changed a bit
>
> Hmm, I should do a white paper this. Yepp, I will
>
> Anjo.
>
>
> Op 3/18/08 10:12 AM, Tony Sequeira <tony@xxxxxxxxxxxxxxx> schreef:
>
> > I help support a 10.2.0.3 Database running on MS Windows Server 2003
> > SP1.
> >
> > Someone has some old performance scripts that displays the data
> > dictionary hit ratio amongst others.  This comes in in the mid to high
> > 80s.  The BCHR on this system is high 90s.
> >
> > I have received a request to sort this out.
> >
> > Quote
> > Data Dictionary Hit Ratio should be > 90% Check Oracle Memory
> > assignment.
> > End
> >
> > I am aware of the BCHR and the arguments against using it, but I cannot
> > find anything specific on the DDHR.  I'm thinking of sending them a link
> > to Kolk's YAPP Method.
> >
> > Any pointers?
> >
> > Of course, they may be right, and this may need addressing.
>
>
> --
> //www.freelists.org/webpage/oracle-l
>
>
>

Other related posts: