Clearing the Analysis Services cache
If you have a poorly performing query, you will review both the design of your cube and the MDX query, and test the performance of different scenarios. Having the same initial conditions for every test is a must. Only then can you truly measure the effect of any changes you make.
The problem in preserving the initial condition lies in the fact that Analysis Services caches the result of each query, making every subsequent query potentially run faster than it normally would.
Normally, caching is a great thing. Hitting a cached value is a goal we're trying to achieve in our everyday cube usage because it speeds up the result. Here, however, we're trying to do the opposite - we're clearing the cache on purpose to ...