Amir,
HANA doesn't force you to do anything.
If you want to use HASH partitioning to distribute the partitions over multiple hosts, that's your choice.
Concerning point 3: I did not say that there is no point or reason to distribute the data.
All I'm saying is that you don't want to transport a lot of data between the hosts.
You want to aggregate first.
To me the question is: what makes you believe that partitioning solves your problem at hand?
Could it be sufficient to distribute tables over the different nodes without partitioning them?
SAP BW powered by HANA is very reserved about partitioning of it's database objects, so I wouldn't rush into partitioning if I wouldn't know what exactly I get out of it and how to get it.
- Lars