We're creating a MultiProvider over 4 multi-million-row cubes, and a Bex query where many users will need to drill down to each level of Profit Center Hierarchy with 5-second response time. I'm starting to create an aggregate at each Profit Center hierarchy level. Does that sound reasonable?
Could be reasonable, but I think I would want to know at the point they are drilling to the Profit Center level, how much of the data has already been filtered out? That is, are they looking at a very small subset of the rows in the fact table or a larger chunk?
How many InfoProviders are in the MP? You will gain the parallel processing feature of using a MP, so you may find that performance is adequate without the multiple aggregates.
Another possibility is loading up the entire results into OLAP cache which would provide better performance than aggregates.
Add a comment