Re: dimension table
- From: Greg Rahn <greg@xxxxxxxxxxxxxxxxxx>
- To: kp0773@xxxxxxxxx
- Date: Tue, 8 Dec 2009 14:21:51 -0800
What troubleshooting have you done - give us some context? Where do
you perceive the problem to be and why? Are the cardinality estimates
representative?
(BTW - not having the SQL or at least the predicate section of the
plan will make it nearly impossible to give much valuable feedback)
Can you simplify this down to a 2 table query like such:
select count(*)
from
SALDB.DIM_DATE d
SALDB.FACT_PGP_OPP f
where
f.date_key = d.date_key and
d.quarter = 1 -- or whatever filters there are on these 2 tables
You can find the filters in the predicates section of the dbms_xplan output.
How many rows come back? 181,000? How many rows come back when the
2nd dim_date is added? (rowsource #25)
The plan shows a very low ROWS estimate (<14) for every row source
after the 3rd join (DIM_DATE-> FACT_PGP_OPP -> DIM_DATE). Given the #
of NL joins, if those first joins produce a much larger row set, that
plan will be quite slow.
On Tue, Dec 8, 2009 at 12:52 PM, Kar <kp0773@xxxxxxxxx> wrote:
> Thanks Guys . . having a tough time with this table.
> Attaching the plan . wont be able to put the query for nda issues.
> appreciate any help.
--
Regards,
Greg Rahn
http://structureddata.org
--
//www.freelists.org/webpage/oracle-l
Other related posts:
