When to create a measure in AAS and when in the power bi layer?

[Anonymous]

What would be the criteria to decide if it is better to create a measure in AAS or power bi, when it looks like the tendency from Microsoft is to put more power into Power Bi Premium (i.e. Large Dataset Format) and less into AAS.

I hold a power bi premium license and also AAS dev access.

Also I am connecting live to AAS.

Can you tell me pros and cons ?

[Kjoebie]

Create measures in AAS when you want to define them once and reuse them many times. Connecting to AAS from PBI enables users to use the measure but not alter them. The good thing is that you have a central place for all company wide verified and certified measures. When a measure needs to be changed this request will be send to the AAS team (through a governance process). Users connection with PBI to AAS are able to create there own measure in a report. But these measures are hard to manage as the report developer is (more or less) responsible for it, unless those measure become part of a governance process.

Another thing with AAS cubes is that end users cannot create calculated tables so they a limited in extending reports with specific calculation or other datasources.

If you want to leverage the full force of self service capabilities of PBI than don't use AAS and create reports in PBI desktop and publish them to the service.

[dhruvinushah]

Hi @Anonymous ,  Great question. 
I'm not an expert on semantic tabular models, but per my experience from SSAS and Power BI, the measure you create in power bi is delivered back to the Cube and runs on the Cube itself. I am assuming AAS does the same thing except maybe a faster connection as it is Azure Cloud connection between Power BI and AAS. 
Hope this helps. 
Refer this for more explaination: @v-lionel-msft : Measures in report vs cube performance? - Microsoft Power BI Community.