I think you talk about the process step of the cube?
Oracle is sensible to the where clause order, but you can't specify it in
your cube.
you can optimize your cube to remove any relationships during the process
step.
but be carefull to take a backup before, in case of bad data quality.
you have to insure that your dimension tables will contains all the keys
used in the fact table.
in this case AS2000 execute a "simple" select col1, col2 col3... from
facttable without join to other tables. this will remove your need to order
the where clause.
if after you have optimized your cube the joins still here you have 2
choices:
1. review your database and dimension / cube design to have a good schema
2. create a view in your oracle database which return all the required
columns and use this view instead-of the current table in the cube. in this
case you'll control the order of the where clause.
"OfraA" <ofraa (AT) clal-ins (DOT) co.il> wrote
Quote:
Hi,
I have two similar cubes. One cube query run very fast on the Oracle
DB, and the other runs very slow. The only difference is the order of
the Where conditions.
The Oracle DBA told me it makes the difference.
How can I influence the order of the Where condition?
I'm working with OLAP 2000 SP4
Thanks in advance
Ofra |