JDBC Unity Catalog connections are now generally available

JDBC connections in Unity Catalog allow you to access external databases securely using the Spark Data Source API or the Remote Query SQL API. The JDBC driver is loaded from a Unity Catalog volume to ensure isolation and security.

Use cases

Imagine you need to read data from a legacy database or write pipeline results to a system not supported by query federation. With this feature you can:

• Create a centralized connection with credentials hidden from querying users.
• Use custom JDBC drivers stored in Unity Catalog volumes.
• Perform read and write operations using the Spark API.

How to use

You can create a connection using SQL commands in Databricks SQL or notebooks.

CREATE CONNECTION oracle_connection TYPE JDBC
ENVIRONMENT (
  java_dependencies '["/Volumes/my_catalog/my_schema/my_volume_JARs/ojdbc11.jar"]'
)
OPTIONS (
  url 'jdbc:oracle:thin:@<host>:<port>:<SID>',
  user '<oracle_user>',
  password '<oracle_password>',
  externalOptionsAllowList 'dbtable,query'
);

What happens in this code:

1. The CREATE CONNECTION command registers the object in Unity Catalog.
2. The java_dependencies parameter points to the driver JAR file stored in a volume.
3. The OPTIONS block defines the connection URL, credentials, and allowed query-time options.

Limitations and requirements

• Generally available (GA) on serverless compute, Databricks Runtime 19 and above, and SQL warehouses 2025.40 and above.
• OAuth M2M authentication remains in Beta.
• Requires the CREATE CONNECTION privilege on the metastore and read/write access to the Unity Catalog volume.
• For serverless SQL warehouses, isolated workload networking must be enabled.