The Pulse
Databricks previews publishing metric views to Power BI
Databricks has made Publish to Power BI available in private preview for Unity Catalog metric views. The feature creates Power BI semantic models from Databricks definitions and can republish updates through Lakeflow Jobs.
Databricks says Publish to Power BI is available in private preview, letting customers publish Unity Catalog metric views to Power BI semantic models. In a November 10 blog post, the company said the feature lets teams keep metric definitions in Databricks while using them in Power BI reports.
Databricks describes the approach as a way to use shared business definitions across reporting and AI tools, including its AI/BI dashboards and Genie. Once enabled, the feature supports publishing on demand from Catalog Explorer or through a Power BI task in a Lakeflow Job.
From metric views to reports
Each published metric view becomes a table in the Power BI semantic model. The dimensions become columns, while measures become DAX measures. The data model can include joins, relationships and other semantic metadata alongside the dimensions and measures.
Power BI queries the metric view itself, with joins defined in the view running in Databricks. Joined dimensions appear as ordinary columns in Power BI, rather than requiring the relationships to be rebuilt there.
In an example from the company, a marketing team defines a view with channel and region dimensions and measures for spending, revenue and return on ad spend. Publishing carries those fields into Power BI, where an analyst can build a report comparing return on ad spend by channel and region.
The feature supports standard aggregations, distinct counts, ratios, filtered measures and measures composed from other measures. It also supports current, offset and trailing window measures. Descriptions and display formats, including currency, decimal and percentage settings, carry over to Power BI field properties.
Keeping definitions aligned
Teams can publish on demand or use a Power BI task in a Lakeflow Job to republish metric views to a target semantic model on a schedule or after upstream tables refresh. Updated or new measures and new dimensions can then flow into Power BI automatically.
Publishing runs only from Unity Catalog to Power BI. An edit to a published measure in Power BI is not written back to the metric view. To change the governed metric, the view must be updated in Databricks and republished, either directly or through a scheduled job.
Access and storage
DirectQuery is the default storage mode and sends live queries to Lakehouse/RT or Databricks SQL; Import mode loads data into Power BI. With DirectQuery, queries follow Unity Catalog access controls and security policies. In Import mode, a copy of the data is loaded into Power BI, where access to that copy follows Power BI permissions.
Databricks also says Power BI will soon be able to benefit from metric view materialization, an optimization that precomputes results. The company says that, when a matching materialization exists, Databricks will serve the query from it; otherwise, the query will run against the metric view.
Preview access
To test the private preview, customers need to work with their account team. Databricks' preparation steps call for adding measure inputs as dimensions and naming every field before publishing.
Users then sign in with their Power BI credentials and select a target workspace in Catalog Explorer. After publishing, users can check the semantic model to confirm that the view appears as a table, its dimensions as columns and its measures as DAX measures.