Working with a client recently, we came across some issues with reporting against Dynamic Cubes and determining when to use a Detail Filter or Slicer. Many of us have heard and been taught that when we report against OLAP cubes like Transformer we should generally try to use a Slicer instead of Detailed Filters, that it would provide better performance. We got on a call with an IBM rep who had deep knowledge of the architecture of Dynamic Cubes and he shared some info that reset the old way of thinking.
IBM says that when you report specifically against Dynamic Cube you should generally choose Detailed Filters as your first option. Apparently, the Dynamic Cube engine is setup to properly interpret that filter so that it does not bring back all the data. This makes sense since it is a ROLAP engine.
When using a Slicer, the Dynamic Cube engine will bring back all the data from the database or caches and simply ‘mask’ the data not filter them out.