Hello, Im finding it a little "extra" work is needed to get a complicated tabular model designed vs traditional OLAP cubes and it also seems more clunky to work with. For example renaming fact/dims tables/columns there is no auto-refresh of the data source view, you have to go to each table properties and refresh there and then modify your affected formulas. Having say 10 facts and 20 dims there are a lot of "inactive" relationships created by the designer which results in custom measures to be created for each of these, why did they do this? E.g. CountofInactiveRelationField:=CALCULATE(COUNT(Table[column]),USERELATIONSHIP(Table[column] ,LookupTable[LookupColumn])). If we dont do this then the counts are not correct for this inactive field. Traditionally to get a simple count rolled up for each dimension there is on standard count measure and the rollup happens automatically in OLAP cube. This happens to be a pain where there are 10+ formulas for each of the facts!
Is this the design experience developers are having out there? or is there some workarounds for this? It seems memory/compression speeds for data loading/retrieve is offset by bad design principles needed to be applied to get it.