Re: Manually changing existing histograms

From: <breitliw_at_centrexcc.com>
Date: Wed, 15 May 2013 10:00:22 -0600
Message-ID: <20130515100022.vgavgeel8kk8o4gk_at_webmail.telushosting.com>



The public examples are in the papers/presentations on my website ( www.centrexcc.com ). However, I just had a look at "Active Statistics" slide 43 and realized that the example is old. There are subtle changes introduced in 10.2.0.4 and later.
The link to Jonathan's blog is probably the best reference.  I have an example somewhere where I created a fake FH histogram for a column with way more than 254 distinct values (i.e. you couldn't gather an FH ) which gave accurate estimated rowcounts for a few select values and a default rowcount for the rest. It also pointed out the difference between the "old" FH setup and the "new". I'll post that when I find it.

On Wed, 15 May 2013 17:11:42 +0200, Martin Klier <usn_at_usn-it.de> wrote: Hi Wolfgang,
>
> thanks a lot - it wasn't easy to guess the right searchwords. :)
>
> First thing I got:
> http://jonathanlewis.wordpress.com/2010/03/23/fake-histograms/
>
> Can you give the list a link to your examples? Just collecting ideas at
> the moment, wasn't able to reproduce my case so far, with and without
> the FH.
>  

  --
Wolfgang Breitling
Centrex Consulting Corporation

--
http://www.freelists.org/webpage/oracle-l
Received on Wed May 15 2013 - 18:00:22 CEST

Original text of this message