Semi additive measures

What is the best way to handle semi-additive measures?

From my search I found:
1)Use an aggregation function on the measure and train users to not aggregate on the particular dimension
2)Use “None” as the aggregation function and manually aggregate in the report
3)Use a self-restricting join and force the users to pick the dimension value and therefore restrict everything using this prompt
4)Break up the measures into 2 different fact tables/contexts - problem is you can’t drill between incompatible objects, right?

Seems like all 4 of these approaches have a downside. I’m tempted to use approach #4 but need to be able to create KPIs in the report that use measures in both fact tables and drill from the KPI to the different measures.
Thanks,
Nile


Nile :us: (BOB member since 2004-02-12)

This post talks about semi-additive measures!

http://neverknewthat.wordpress.com/2008/10/02/multipass-sql-20/


dtaylor92 (BOB member since 2008-10-03)