HANA Aggregate Question

[former_member182921]

We just conducted a BW to HANA migration and our team has a question that we hoped someone on this Wiki could help with.  We know that HANA does not keep aggregates but yet we see a difference in statistics data between DSO and Infocubes.  Can someone explain why?

Please not that the cube is using $X aggregates and DSO is not using $X in HANA, which we do not know why

[henrique_pinto]

BW is a different thing. If your cube is on a lower granularity than the DSO, BW will keep the aggregated fact table on HANA DB. Notice that even the in-memory optimized cubes are not virtual models, they do persist data (though in a simpler fashion than a non-HANA BW system - only one fact table instead of E/F tables, no DIM tables, etc.).

If you want to go for 100% virtual modeling on BW on HANA, check out the composite providers. They can be built on top of your DSOs directly, and they won't persist any intermediate result (true logical DW).