DirectLake over OneLake with combination with DirectQuery tables

[jaryszek]

Hello Guys,

Can I create a composite model on the current existing DirectLake over the OneLake remote semantic model?

I want to connect to Fabric Warehouse and make the new table a DirectQuery? 

It will be working?

Best,
Jacek

[v-ssriganesh]

Hi @jaryszek,
Thank you for posting your query in the Microsoft Fabric Community Forum.

Adding a DirectQuery table via Tabular Editor may appear to work at the metadata level, Microsoft only supports composite models (Direct Lake + DirectQuery) when the semantic model is created and owned as a composite using supported tools, not by extending an already published remote Direct Lake on OneLake semantic model.

The best approach is to Create and own the semantic model in Fabric, including Direct Lake (OneLake) tables and Warehouse tables, letting Fabric manage the storage modes (Or) create a new local semantic model as a composite model that includes Direct Lake and DirectQuery tables.

Thanks, @Jaywant-Thorat for sharing valuable insights.

Best regards,
Ganesh Singamshetty.

[jaryszek]

Hello,

thank you. 

Please explain what this means: "when the semantic model is created and owned as a composite using supported tools"

Tabular Editor 3 has option to import tables from a warehouse. 

Best,
Jacek

[v-ssriganesh]

Hello @jaryszek,
I mean that the semantic model is initially built and authored using Microsoft supported modeling experiences, where composite models are explicitly supported and documented.

According to the documentation:

• Composite models (mixing storage modes) are supported when the model is created using Power BI Desktop or supported Fabric web modeling experiences:
  Use composite models in Power BI Desktop - Power BI | Microsoft Learn
• DirectLake semantic models are designed to work directly over OneLake and are authored and managed using Fabric tooling:
  Direct Lake overview - Microsoft Fabric | Microsoft Learn
  https://learn.microsoft.com/en-us/fabric/fundamentals/direct-lake-web-modeling

[jaryszek]

Ok so we misunderstood. 

My question is: if warehouse table using Fabric environment can be used together with OneLake semantic model? 

Assume you have 3 tables on DirectLake over OneLake. This is DirectLake mode.
But now I want to add warehouse table, so I am goinf to fabric, clicking import table, choosing Fabric Warehouse...It will work? 

Or we will have semantic model version issues or something like that?

Best,
Jacek

[v-ssriganesh]

Hello @jaryszek,
Yes, you can add a Fabric Warehouse table to your existing DirectLake semantic model (with the 3 OneLake tables) directly in the Fabric web experience. Here's how it works:

• Go to your semantic model in Fabric > Model view > Get data  from the ribbon > Select Fabric > Warehouse > Choose your table. This automatically creating a composite model (mixing DirectLake + DirectQuery modes).

[jaryszek]

Thank you,

what about compatibility level? I am currently using 1702 for directLake...

It will work or there is no connection between them?

Best,
Jacek

[v-ssriganesh · Accepted solution]

Hi @jaryszek,
As per my understanding, Compatibility level is a property of the semantic model itself, not of individual tables. Your model already runs at compatibility level 1702.

Best regards,
Ganesh Singamshetty.

[Jaywant-Thorat]

Hi @jaryszek 

Short Answer = No, not in the way you’re describing.

Let me elaborate What works vs what doesn’t:
You cannot:

• Take an existing DirectLake semantic model (remote)
• And extend it with DirectQuery tables
• To create a composite model

A remote DirectLake model is read-only for modeling.

You can do this instead:

Option 1 : Fabric way

• Build all tables inside Fabric
  • DirectLake tables (from OneLake)
  • Warehouse tables (SQL)
• Let Fabric automatically handle:
  • DirectLake
  • DirectQuery
• One semantic model → works fine

Option 2 : Classic Composite

• Create a new local semantic model
• Add:
  • DirectLake tables
  • DirectQuery tables (Warehouse / SQL)
• This works only if the model is local, not remote

Key rule to remember:

• You cannot extend a remote DirectLake semantic model.
• Composite models work only when you own the model.

Recommendation: If you’re already in Fabric

• Create everything in Fabric (Warehouse/Lakehouse)
• Let Fabric manage DirectLake + DirectQuery natively

=================================================================
Did I answer your question? Mark my post as a solution! This will help others on the forum!

Appreciate your Kudos!!

Jaywant Thorat | MCT | Data Analytics Coach | Super User
LinkedIn: https://www.linkedin.com/in/jaywantthorat/
Join #MissionPowerBIBharat: https://tinyurl.com/JoinMissionPowerBIBharat
#MissionPowerBIBharat
LIVE with Jaywant Thorat

[jaryszek]

Hello,

thanks. 

Hmm I am not sure what does it mean "No".

I have connected to Fabric Warehouse table using tabular editor: 

I have this as DirectQuery: 

Seems that it is working. 

What I am afraid of if really this table will be compatibile with other tables connected from DirectLake over OneLake? 

Best,
Jacek