Analysis Services has a limitation of one distinct count measure for each measure group; in the past, it was a limitation for the single cube. Each distinct count measure cause an ORDER BY clause in the SELECT sent to the relational data source during processing. I’d like a different approach, but probably this is already a good compromise between all the needs.
Now in SSAS2005 it seems to be a little inconsistency in the user interface of BI Studio. If you try to create a measure group manaully, editor don’t allow you to create more than one measure group from the same fact table. If you use the “New Measure…” context menu (and related wizard) it creates a new measure group for each distinct count measure.
What I’d like to know is if the presence of a distinct count measure in the same measure group containing several other measures cause a performance loss for all the measures similarly to AS2000: in that version I create each distinct count measure in a separate cube, joining all the measures in a single virtual cube; in that way I had no performance loss (caused by potentially less aggregations) when queries contain only “standard” measures. The “New Measure” dialog box creates a new measure group for each distinct count measure. It seems that not very much has changed from the previous version.