Hi,
In my fact table I have two degenerate dimension (OrderNo, ItemNo). The grain is at billing level, so 1 OrderNo can have many rows, same with the ItemNo.
Question, how do I best represent these as measures in my cube (OrderNo Count and ItemNo count). I can do a distinct count to create both measures, however I'm trying to avoid distinct counts, one reason this will create two new measure groups, so I assume processing and query performance will decrease (my fact table is roughly 200million rows).
These are important measures for me because many questions are about about the production... how many orders last month, how many orders for supplier x and so on...
Are there alternatives to distinct counts?