Forum Discussion
Performance semantic model measure vs custom measure
Super User
Credential wrote:
I don't think it has anything to do with Power BI. I think it's more because of how the Query Optimizer optimizes the measures.
I was suggesting that it likely does have to do with the query optimizer but this is, in fact, related to Power BI updates. When you run a query in DAX Studio, it connects to the Power BI tabular analysis server, either a .pbix you have open locally or a dataset in the service (it doesn't have its own DAX engine).
But we only use SSAS tabular 1500 on premise. What does this have to do with Power BI updates?
The measure is defined once in the tabular model itself and once in the query itself. Both queries are sent to SSAS tabular 1500 and evaluated there via the FE and SE, right?
What I do not understand?
- AlexisOlson
OK. I may not have fully understood the specifics of your question. Let me make sure we're on the same page. Is the following accurate?
You are using Power BI with a live connection to an SSAS tabular model (not Import or DirectQuery). You have a measure that performs better as a Report Level Measure (as defined here) than the same measure defined within the SSAS model.
Yes, that's correct.
- AlexisOlson
That is indeed quite strange. I can't think of a good reason for a Report Level Measure to be faster than the same measure defined in the tabular model. Can you verify that it's not related to caching?