Performance of Distinct Count in DAX

If your database has a few millions rows, DISTINCTCOUNT is easy and efficient. However, when you face billions of rows or hundreds of millions of different values, then performance starts to be an issue. In such a case, you have to find different solutions like partitioning, distinct value reduction, rephrasing of the DAX query and, for each trial, you need to dive into the query plans, to get a good understanding on how to best leverage the VertiPaq engine. You will see all of that in this session. At the end, you will NOT have learned how to write the best formula for any distinct count calculation, but you will have the knowledge to find the best formula for your specific scenario.
Presented by Marco Russo at SQLBits XII
Tags (no tags)