SSAS Many to many optimization technique #1 (By Erik Veerman)

This has been something brewing for a few weeks that I am just getting to writing down… and yes, naming this idea #1 does imply that I have another idea :) so stay tuned.  Before diving into the details, the summary is that this approach when applied against the AdventureWorksDW database was able to take the intermediate measure group rowcount down from approx 16,000 to only 19.  Wow… because the size of the intermediate measure group has a big impact on performance.

 

Lets call this the “Many to Many Matrix Relationship Optimization” approach for SSAS M:N relationships… I’ve been working on this with Dan Hardan (Msft) on this because of a need to make the Many to Many relationships perform better.

 

As a background, there are many kinds of many to many relationships, but if I can summarize two of the common kinds…