Skip to main content
cancel
Showing results for 
Search instead for 
Did you mean: 

July 7 - July 17 | Round 2 of the Power BI Dataviz World Championships. Don't miss your chance! Learn more

Reply
djurecic
Super User
Super User

Direct Lake Views

Hi All,

 Trying to get some clarification around DirectLake and views- specifically what views are not supported?

 

According to this article (https://learn.microsoft.com/en-us/power-bi/enterprise/directlake-overview)

 

  • Currently, Direct Lake models can only contain tables and views from a single Lakehouse or Data Warehouse.

and

 

  • Tables based on TSQL-based views cannot be queried in Direct Lake mode. DAX queries that use these model tables fallback to DirectQuery mode.

I'm looking for some clarification on  the difference is between "views" in the first sentence and "TQL-based views" in the second.

 

Thanks,

DJ

1 REPLY 1
freginier
Solution Sage
Solution Sage

There's no real contradiction between those two sentences - they describe two different things: what you can put in the model vs. how it's queried at runtime.

 

- Model composition (sentence 1): when you build a Direct Lake semantic model, the objects you add can be tables or views exposed by the Lakehouse/Warehouse SQL endpoint. So a view is allowed as a model object - you can include it.

- Query execution (sentence 2): Direct Lake gets its speed by reading the underlying Delta/Parquet files in OneLake directly. A TSQL-based view has no physical Delta table behind it - it's just a SQL definition. So when a DAX query touches a model table that is actually a view, there's nothing for Direct Lake to read, and that query falls back to DirectQuery (the view's SQL runs against the SQL endpoint).

 

So "not supported" really means "not served in Direct Lake mode". You can add views, but querying them won't use Direct Lake - you get DirectQuery behaviour for those tables instead. If you need Direct Lake performance on that logic, materialise the view as a Delta table (a Lakehouse table, or a warehouse table built by a pipeline/notebook) and point the model at that table rather than the view.

Helpful resources

Announcements
FabCon and SQLCon Barcelona 2026

FabCon & SQLCon – Barcelona 2026

Join us in Barcelona for FabCon and SQLCon, the Fabric, Power BI, SQL, and AI community event. Save €200 with code FABCMTY200.

60 days of Data Days Carousel

Data Days 2026

Join Fabric Data Days 2026: 60 days of free live/on-demand sessions, challenges, study groups, and certification opportunities.

Power BI DataViz World Championships carousel

Power BI DataViz World Championships - June 2026

A new Power BI DataViz World Championship is coming this June! Don't miss out on submitting your entry.