Parallelism and MDX Queries (by Chris Webb)
A lot of my work involves tuning MDX queries, and over the last few months I've come across several situations where calculation-heavy queries take a long time to execute but for the most part only seem to use one CPU on a multi-processor server. When I first saw this I made some enquiries about what was going on and found that this was indeed a feature of the AS calculation engine: only one query can be run per session at any one time, and for that query the work of the formula engine (unlike the storage engine) can't be parallelised. Hohum.
However, then I realised that there are some scenarios where you can inject parallelism into a query.
Tags: mdx, performance