Thanks Shrikant.
Indeed, performance is my main concern.
Generally speaking, 16 dimensions seems fair enough for me (I don't think that I need to use it all in my company). However, my concern is which kind of dimension I use. For example: if I would like to present the material dimension in the BPC, I can do it by adding the material (MATNR) dimension or by using material groups (MATKL) dimension .
[Just to emphasize, in my case, at the material level, we have about 100,000 different materials, while in the material groups level we have only 1,500 different material groups].
Obviously, without any performance issue, I prefer to use the specific material dimension in order to have better drilldown option, but I'm not sure that I can do it without affecting the performance.
Therefore, I'm searching for some best practice regarding to cube sizing/perfomarnce tradeoff.
Thanks,
Yaron