Header/Item extractor data modelling question

Question

Hi,
I have installed the Business Content for CRM Activities. In the Business Content they have a separate extractor and cube for header and item level.
However, for our reporting requirements we would like to use one cube. We don't consider a multicube as a valid solution since the item/header cubes have ofcourse not all fields in common what will lead to the famous not-assigned behaviour in Bex.
Therefore I'm trying to enrich each item line with all header fields. I don't want to write a user-exit to enhance the item extractor on datasource side since this will be a lot of work and also not sure this would be delta-compatible. I have bad experiences with this on a previous project.
So I would like to find a solution on BW (data modelling) side but at the moment I don't see any good alternatives. The only thing that I can think of is to build one ODS/master data object (transaction number) that contains all header information and to read this ODS/Master data object while loading the item lines from the ODS to the cube. This way I can copy all header fields to the item line. But this will force me to do FULL loads between the item ODS and the "global" cube and ofcourse I would like to avoid this. (if a header fields changes I want this to be changed in all item lines as well ofcourse)
Does anyone have better solutions ?
Thanks

Answer 1

What I finally did is the following :
1) Create a new cube with all characteristics and keyfigures from header/item level.
2) Load Header ODS to new cube
3) Load Item ODS to new cube but via a startroutine that enriches the item datasources with all header characteristics that are not yet available on item level.
4) Make a query and put constant selection on a restricted keyfigure that contains the following :
a) Product (item level) ==> CONSTANT SELECTION enabled
b) Nr. of activities (header level)
In Bex everything looks now fine with the nr. of activities header keyfigure, even when I do a filtering on material that still only exists on item level thanks to the constant selection but my result row is acting a little bit dissapointing. I hope that can be solved.
Let me give an example : (i'm even excluding the header line in this query)
1) all fine
   Transaction nr ; Nr. Of activities
   6000           ; 1
   ====================================
Resultrow           1
2) drilldown to material
   Transaction nr ; Material ; Nr of activities
   6000           ; 12D      ; 1
   6000           ; 13D      ; 1
=====================================================
Resultrow                      2
==> As you can see the resultrow changes to 2 after doing the drill-down.
Note that I am using "calculate result as summation". If I use "nothing defined" the result will be fine in this case but then it is not correct in the case where I do filterings on e.g. a material number. It will then show the number of activities for the WHOLE query.
Is there anything I can do to get the resultrow right in all circumstances using constant selection ?
Thanks
Message was edited by: Double U