Re: Vagaries of the CBO: Out-of-Range Predicates
- From: "Anand Rao" <panandrao@xxxxxxxxx>
- To: "Wolfgang Breitling" <breitliw@xxxxxxxxxxxxx>
- Date: Tue, 19 Dec 2006 07:02:02 +0530
Hi,
Thanks Wolfgang, you are right i got a bit entangled in the terminology and
hence ended up formulating a wrong question.
Yes, selectivity is what i should be worried about and ultimately the cost
factor. because, i my case CBO is choosing the wrong index due to the NDV
being 1 and i don't have histograms.
so, looks like i have resign myself to the fact that an NDV hack is the only
way to make CBO choose the right index (without hinting the query).
many thanks,
anand
On 18/12/06, Wolfgang Breitling <breitliw@xxxxxxxxxxxxx> wrote:
At 12:11 AM 12/18/2006, Anand Rao wrote:
adding one more question,
3) How does CBO calculate the density in cases of bind variables when the
value is out of range?
or is this a RTFM / RTFB ... ??
density is a "property" of the column and is roughly related to the
probability of finding a particular value in the rows of the table. It is
calculated when statistics are collected, *not* at parse time. density is
in no way dependent on bind variables - or literals for that matter. You may
be mistaking density with selectivity.
Regards
Wolfgang Breitling
Centrex Consulting Corporation
www.centrexcc.com
Other related posts:
