Migrating from embedded ODBC drivers to ADBC in Power BI and Fabric

Connectors and driver changes

 

For the following connectors, Power BI/Fabric will move away from the current embedded ODBC drivers and toward the replacement drivers (primarily ADBC).

• Databricks: Simba Spark ODBC Driver → Databricks ADBC Driver
• Dremio: Simba Drill ODBC Driver → FlightSQL ADBC Driver
• Google BigQuery: Simba Google BigQuery ODBC Driver → BigQuery ADBC Driver
• Hive: Simba Hive ODBC Driver → Deprecated
• Impala: Simba Impala ODBC Driver → HiveServer2 ADBC Driver
• Snowflake: Simba Snowflake ODBC Driver → Snowflake ADBC Driver
• Spark: Simba Spark ODBC Driver → HiveServer2 ADBC Driver

 

Who’s impacted

You are impacted if you use any of the connectors in the previous list and meet one of the following conditions:

• You haven’t explicitly specified an implementation for the connection
• Your organization wants centralized control over whether ADBC or ODBC is used by default.

 

This default behavior is intended to apply anywhere a connection is created (e.g., semantic models, dataflows Gen2, DMTS, and paginated reports).

 

How to migrate

There are three ways to migrate, depending on how much control you want:

• Per-connection opt-in (good for individual testing)
• You can opt in to ADBC by adding Implementation="2.0" to supported connections. This lets you validate at your own pace.
• Tenant setting (good for organization-wide control)

Tenant admins can set the organization’s default using the Admin portal setting:

 “Users can connect to data sources by using Apache Arrow database connectivity (ADBC)”

• Disabled (default): selects the legacy ODBC drivers as the default option
• Enabled: selects the new ADBC drivers as the default option

This setting also controls the “Test Connection” behavior for each connection type inside of the tenant.

 

Workspace override (good for side-by-side validation)

The tenant setting is delegated to workspaces so workspace admins can override the default and compare behaviors without modifying each individual connection. This enables side-by-side validation. For example, you can use one workspace to test ADBC defaults while keeping another workspace on ODBC.

 

How “Implementation” is chosen (use this as the quick mental model to predict which driver is used)

ADBC is used when:

• Implementation="2.0", OR
• Implementation is not specified (null) AND the workspace setting is On

ODBC is used when:

• Implementation="1.0", OR
• Implementation is not specified (null) AND the workspace setting is Off

If the Implementation is explicitly specified in the connection, that value is used regardless of the tenant/workspace defaults.

 

Next steps

1. Start by selecting a pilot workspace and enabling ADBC there first (workspace override) to validate key datasets and refresh scenarios.
2. For any critical connections you want to validate immediately, opt in per-connection with Implementation="2.0".
3. After validation, decide whether to enable ADBC by default at the tenant level.

 

Key dates

• ODBC drivers will be removed from service beginning September 2026. If you require staying on ODBC (Implementation="1.0" or workspace setting Off), you will need to use a Gateway to continue executions using ODBC.
• The ODBC drivers referenced above will no longer ship with Power BI Desktop or Gateway beginning with the March 2027 release. Users who wish to retain ODBC functionality will be required to stay on the February 2027 release or earlier.