I have an MDX query that takes a long time to run. I've looked at it with MDXScriptPerformance, but because its a composite of many calculations, it seems to be the volume of the calculations that is slowing it down. The calculations include percentiles which use ranks, and the cellset is about 350k in size.
Currently it takes about 35 secs to run and I've used SQL Sentry to try and diagnose the bottleneck. It looks like the formula engine, but I really don't know what to do next. For the sort of performance improvement I need its almost as if I need a caching mechanism involving the DataMart. As it stands the ADO .NET driver is timing out in SSIS when I'm trying to push results from the cube back to the mart.
Sean.