I have a cube with 150 million rows. I have it partitioned to around 6
million rows per partition. All attributes and hierarchy members are set to
'default' for aggregation type except some that I set to 'none' because I
know I don't want them.

Now, if I put the partition count to 6 million I can only make the following
aggregation set:

16 aggs 35% 13.2 MB

If I lie and put six hundred thousand instead of 6 million in as the
partition count, then I get this different aggregation set:

77 aggs 10% 174.4 MB


What is up with that? In this case I have the time and storage space to
make many, many aggs. How do I do that? What is bad about lying about the
rowcount, anything?