Transition from ODBC to ADBC drivers in Power BI and Fabric

Microsoft Power BI and Fabric are transitioning supported data source connections to Apache Arrow Database Connectivity (ADBC) drivers. ADBC brings performance improvements for large queries with lower overhead and built-in memory safety.

Use cases

Imagine you need to refresh large semantic models and want to reduce load times and memory overhead when fetching data from sources like Databricks, Snowflake, or Google BigQuery.

• Test ADBC behavior in pilot workspaces.
• Identify dependencies on the legacy driver using the audit notebook.
• Migrate connections individually or at scale via tenant settings.

How to use

Add the implementation parameter to your M connection string to force ADBC usage on specific queries.

Source = Databricks.Catalogs("meu_servidor.cloud.databricks.com", "meu_catalogo", [Implementation="2.0"])

What happens in this code:

1. The Databricks connector initializes targeting the provided host and catalog.
2. The Implementation="2.0" parameter forces the query to use the new ADBC driver instead of the legacy ODBC one.
3. Data is fetched efficiently leveraging the Apache Arrow in-memory format.

Limitations and requirements

• This transition affects specific connectors such as Databricks, Snowflake, Google BigQuery, Impala, and Spark.
• Refreshes routed through an on-premises data gateway continue to use ODBC drivers during the transition phase.
• ADBC becomes the universal cloud default as part of the scheduled 2027 roadmap.