Re: How to analyze a odd column distribution

Aahh,

you hadn't said anything about the table being partitioned in your 
original post. How is it partitioned?
As to your histograms, it appears you could get away with far fewer 
buckets than the default 75 since the only popular value is 2999-12-31.
Secondly, in partition P2 that seems to be the ONLY value. If that is 
the case then
a) the histogram is at the very least useless
b) as you case shows may very well be harmful
and therefore
c) only collect the histogram for partition P1


Alfonso León wrote:

> This is an example 
> 
> first there is the statistics and histograms and at the end the
> explain plan for the queries, as you can see if I specify the
> partitions then I get the real number, but that is not the case when I
> don't specify the partitions.
> 

-- 
Regards

Wolfgang Breitling
Centrex Consulting Corporation
www.centrexcc.com
--
http://www.freelists.org/webpage/oracle-l

Other related posts: