[Mondrian] How does mondrian choose aggregate tables

Venkatesh U venkatesh20 at gmail.com
Fri Feb 4 11:44:59 EST 2011

  I have a situation where mondrian is not picking the aggregate table only
in a particular case.

Fact table structure:

agg_tab_1 structure:

As per the business situation, the fact "fact_2" is only valid with the
dimension combination dim_1,dim_2. i.e fact_2 in the actual fact table will
be null and will never be available.I have defined a calculated member with
the following formula

calc_member c1 = [Measures].[fact_1] / [Measures].[fact_2],[dim_1].All

with some basic testing done with more than 50000 rows, mondrian always
calculated the value [Measures].[fact_2],[dim_1].All from the aggregate
table and [fact_1] from fact_table. But while trying the same thing with
less rows 20000 in fact table and 72 rows in aggregate table, the mondrian
tries to calculate [Measures].[fact_2],[dim_1].All from the fact_table
though an aggregate table with value is available.

I am not sure, if i had made sense. To be simple, Requested for a measure
with an index of dimensions dim_1,dim_2, I always want the mondrian to pick
up the values from the appropriate aggregate table. Is there any setting
which can help me?

We use a columnar database, so its a better idea to pick up each fact from
appropriate aggregate table if the measures have different levels of
dimension. Please feel free to question in case you dont understand my
scenario. Appreciate any kinda of help. Thanks in advance.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.pentaho.org/pipermail/mondrian/attachments/20110204/1a2aeb7c/attachment.html 

More information about the Mondrian mailing list