Tuning Cube Processing Performance

I wrote about performance tuning on SSAS about 5 months ago. That was both on processing performance and query performance. This time I would like to write only about processing performance but in a bit more detail.

So the usual case is something like "my cube processing takes 5 hours, can we reduce it to 2 hours?" Now what do we need to do? The wrong thing to do is blaming the hardware :) The right thing to do is to systematically find out which part of the cube takes a long time to process.