I have been doing quite a lot of work on tuning several SSAS Multi-dimensional cubes.
I have used usage based optimisation to create better aggregations and this has really helped improve speed massively.
I really recommend this approach!
However I have found with one of our dimension hierarchies the top levels are very fast but once you start trying to drill down to the bottom level it can be very slow at bringing the data back
I have setup proper dimension attributes relationships but have had to leave the relationships as flexible as the data does change in this dimension.
I was wondering if anyone has any tips.