Forum Discussion
Microsoft EmployeeShare your thoughts on DirectQuery for Power BI datasets and Azure Analysis Services (preview)
Advocate Vtessahurr Congrats on your "insane amazing" Guy in a Cube debut! 😀
I hope to use this feature to break up/modularize my models.
I have a large model in the Service, with multiple fact tables that do not have the same refresh frequency requirement. Currently, I'm forced to refresh all fact tables every 10 minutes even though some only need to be refreshed daily, because the PBI XMLA endpoint won't support automating that with SSIS or SQL Agent. That costs unnecessary time and resources.
With this new feature, I will just make each fact table a single-table "model" that will be refreshed on its own schedule. I will have a "main" model that is just the smaller dimensions, refreshing like every 10 minutes. Then I can just plug and play with the different fact tables as needed. My POC on this idea worked fine today. I just hope it performs when I scale up to 500 million rows.
Either way, I'm not planning on putting this in production until out of preview... Too many limitations outstanding ( I use "ALLSELECTED" a lot, and not being able to delete the objects once added is kind of annoying). But thanks for letting us try it, and for continually making me actually look smart to all of my customers and bosses! GA cant get here soon enough. In the meantime. Happy holidays!
Most Valuable ProfessionalIn your scenario I would definitely get the refresh issue fixed instead of making a workaround using this new feature. I forseas you will get into performance problems when you start joining tables from so many different models. The general performance implications when using DQ also applies when using it with Analysis Services: https://docs.microsoft.com/en-us/power-bi/transform-model/desktop-composite-models#performance-implications
Do you have the latest MSOLAP and ADOMD.NET client libraries installed on your SSIS machine?