Richard Tkachuk has just posted a new whitepaper on the
www.sqlserveranalysisservices.com site called “
Loading Aggregate Data in Yukon”. It discusses several scenarios when the data is already precomputed at the non-leaf cells and needs to be loaded there instead of being aggregated from the leaves. The modeling approach Richard took is similar to what was done in Analysis Services 2000 - i.e. using parent-child hierarchies with Data Members and then using MDX to copy values from Data Members cells to the non-leaf cells. The MDX expressions themselves are simple, but what makes this whitepaper interesting, is that it demonstrates some of the new Analysis Services 2005 MDX concepts, most interesting of which is auto-recursion resolution - i.e. where Analysis Services 2000 would've raised infamous “Inifinite Recursion” error, Analysis Services 2005 knows how to resolve it. The whitepaper is conviniently labeled “v1”, which hints that this is just a first iteration. Indeed, the technique demonstrated in the current document can be futher refined and optimized, and I also think we can expect additional scenarios to be covered.