You can use data from a Fabric Lakehouse in Power BI in two ways: via Direct Lake or via the SQL Endpoint. In this chapter, we explain how Direct Lake works and outline its advantages and disadvantages.
With Direct Lake, Power BI connects directly to the Delta files in the Lakehouse, without requiring refreshes or import processes. This provides a very fast and almost real-time data experience.
Advantages of Direct Lake:
- Power BI reads the data directly from the Lakehouse. Changes in the underlying Delta tables are immediately visible in Power BI without the need for a dataset refresh.
- Because Direct Lake does not use a traditional dataset, there is no need to configure a refresh schedule in Power BI.
- You can build reports entirely within the Power BI service, without using Power BI Desktop. This is especially useful for Mac users, who would otherwise rely on virtual environments or alternative tools.
Disadvantages of Direct Lake:
- Direct Lake only works when Fabric capacity is enabled. If the capacity is turned off, Power BI cannot read the data and the dashboard will not function.
- Viewing or interacting with Power BI reports that use Direct Lake consumes Capacity Units (CUs) in Fabric. This means dashboards incur processing costs not only during data operations but also during interactive use.
Import via Direct Lake (in Power BI Desktop)
With Direct Lake, you can use data directly from a Fabric Lakehouse in Power BI Desktop without importing or refreshing the data. Follow the steps below to load the data (in this example, we use the English version of Power BI Desktop):
- Go to OneLake Catalog in Power BI Desktop and select Lakehouses (alternative: choose Get Data and then Lakehouses).
- You will see a list of all available Lakehouses in your tenant. Select the Lakehouse you want to use; in our example, this is ‘LakehouseExactOnline’.
- Click the Connect button and then select the option ‘Connect to OneLake’.
- A screen will appear where you can create a new Semantic model.
- Give the semantic model a name (in our example, ‘Power BI Exact Online’) and select the tables you want to include in this semantic model. These are usually all tables from the ‘dbo’ schema; you do not need to load the tables from the ‘staging’ schema.
- Click OK to create the semantic model.
The semantic model is now being built and immediately displayed in Power BI Desktop. At the same time, it is also added as an item in the Power BI workspace in the online Power BI Service. Changes you make in Power BI Desktop are instantly synchronized with the Power BI Service, and vice versa.
You can now further shape the data model, for example by:
- Adding relationships to create an optimal star schema
- Defining measures for calculations
- Creating visualizations in the report
All standard Power BI capabilities remain fully available when working with Direct Lake.
Import via Direct Lake (in Power BI Service)
You can perform the same process available in Power BI Desktop entirely within the Power BI Service. This means you no longer need Power BI Desktop. Follow the steps below to create a new semantic model directly in the online environment based on a Fabric Lakehouse:
- Click ‘+ New item’ in the Fabric workspace and select ‘Semantic model’.
- Click ‘OneLake Catalog’.
- You will see a list of all available Fabric items in your tenant. Select the Lakehouse you want to use; in our example, this is ‘LakehouseExactOnline’. (Optionally, you can first filter to show only Lakehouses.)
- Click the Connect button.
- A screen will appear where you can create a new Semantic model.
- Give the semantic model a name (in our example, ‘Power BI Exact Online’) and select the tables you want to include in this semantic model. These are usually all tables from the ‘dbo’ schema; you do not need to load the tables from the ‘staging’ schema.
- Click Confirm to create the semantic model.
The semantic model is now being built and immediately displayed in the Power BI Service. From this point on, you can continue working entirely in the online environment without Power BI Desktop. You can also add a new report as an item in the workspace and select the newly created semantic model as the data source.
You can now further shape the data model, for example by:
- Adding relationships to create an optimal star schema
- Defining measures for calculations
- Creating visualizations in the report
All standard Power BI capabilities remain fully available when working with Direct Lake.