Building a Direct Lake Semantic Model in Fabric Workspace with Tabular Editor 3

In previous posts, I demonstrated how to create tables within the Fabric Lakehouse.

In this writing, I will dive deeper into my journey of using Tabular Editor 3 to craft a Direct Lake Semantic Model in the Fabric workspace, leveraging the tables stored in the Lakehouse.

Additionally, I will share how I learned to establish relationships between tables and create new measures within Tabular Editor 3 while building the Direct Lake Semantic Model.

This process is for creating a custom semantic model specifically for Direct Lake mode connection, rather than using the default semantic model of the Lakehouse.

Let's start.

To create a new model, follow the steps described below. In this example, I have named the new model "document_direct_lake".

5 → Workspace connection link

If the following warning appears, navigate to the Admin portal → Capacity settings, and change the XMLA Endpoint option to "Read Write," as shown in the image below.

Now, the blank semantic model is created. It is time to import some tables into the semantic model from the Lakehouse. To import tables from the Lakehouse, you will need the SQL connection string of the Lakehouse SQL endpoint.

Select relevant tables from the Lakehouse.

Now, the new tables are created in the Direct Lake Semantic Model.

Open the new Diagram window, and drag some tables into it. And then create relationships like below.

Save this semantic model.

Navigate to the Fabric Workspace, locate the semantic model, click on it to open the data model. Here, I am able to see what I have created using Tabular Editor 3.

To take it a step further, I attempted to create a new DAX measure in the table named 'measures_expression' within Tabular Editor 3.

In Expression Editor window, I wrote DAX expression like below.

After saving this model in Tabular Editor 3, I could see the newly created DAX measure in the Semantic Model in the Fabric workspace.

Moreover, when hovering over a table, as shown below, I could clearly see that the storage mode is Direct Lake mode.

To summarize, I tried to detail the creation of a Direct Lake Semantic Model using Tabular Editor 3 within the Fabric workspace. Starting with table creation in the Fabric Lakehouse, it progressed to creating a customized semantic model for Direct Lake mode, including establishing relationships and creating DAX measures. In my opinion, learning how to use Tabular Editor 3 empowers Power BI developers with a more efficient way to create Direct Lake Semantic Models.

I hope this helps having fun getting started with creating Direct Lake Semantic Models using Tabular Editor 3.