The CData MCP Server for Databricks leverages Databricks Thrift to enable bidirectional SQL access to Databricks data. The server supports databases that run Databricks Runtime Version 9.1 and later. It also supports the Pro and Classic Databricks SQL versions. The data model is fully dynamic and automatically reflects the data that is available in your Databricks environment. Any updates to tables or schemas are immediately accessible through SQL.

Discovering Schemas

The CData MCP Server for Databricks dynamically retrieves schema information from Databricks. To ensure that any metadata changes (for example, added or removed columns or modified data types) are reflected in the driver, reconnect to your data source.