Multiple DB connection join in Superset Dataset

[wadooodd]

Trying to create a dataset that is going to join tables present in 2 different DB connections. Was able to create multiple DB connections but unable to join them.
Context: I'm trying to convert a Tableau workbook to Superset. The relevant tableau workbook has a TDS which uses 2 different DB connections lets say DB connection A and DB connection B. Now the connection A has a row level security configs which cannot be moved to connection B. Tableau allows multiple DB connection join. Does Apache Superset support this? If yes, then how?

[dosubot[bot]]

Hi @wadooodd! I'm Dosu and I’m helping the superset team.

Superset does not natively support joining tables from different database connections in a single dataset, unlike Tableau. However, there is an experimental feature called the "Superset meta database" that enables cross-database joins using a special connection (superset://). With this enabled, you can write queries like:

SELECT * FROM "DB_A.schema.table1" JOIN "DB_B.schema.table2" ON ...

To use it, set the ENABLE_SUPERSET_META_DB feature flag to true, then add a new database of type "Superset meta database" with the SQLAlchemy URI superset://. You can restrict which databases are accessible via the allowed_dbs engine parameter. Permissions and row-level security are enforced, but all joins and aggregations happen in memory, so this is only suitable for moderate data sizes and comes with security and performance risks. More details and configuration steps are in the official documentation.

If you need production-grade cross-database joins, consider using a federated query engine like Presto/Trino and connecting Superset to it as a single source.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}