Considerations

When you work with the cross-database query feature in Amazon Redshift, consider the following:

• Amazon Redshift supports cross-database query on the ra3.4xlarge, ra3.16xlarge, and ra3.xlplus node types.

• Amazon Redshift supports joining data from tables or views across one or more databases in the same Amazon Redshift cluster.

• Amazon Redshift Serverless supports the same cross-database capabilities as Amazon Redshift clusters, so you can join data from tables or views across one or more databases in a serverless namespace.

• All queries in a transaction on the connected database read data in the same state of the other database as the data was at the beginning of the transaction. This approach helps to provide query transactional consistency across databases. Amazon Redshift supports transactional consistency for cross-database queries.

• To get metadata across databases, use SVV_ALL* and SVV_REDSHIFT* metadata views. You can't use the three-part notation or external schemas to query cross-database metadata tables or views under information_schema and pg_catalog.

Limitations

When you work with the cross-database query feature in Amazon Redshift, be aware of the limitations following:

• When you query database objects on any other unconnected databases, you have read access only to those database objects.

• You can't query views that are created on other databases that refer to objects of yet another database.

• You can only create late-binding and materialized views on objects of other databases in the cluster. You can't create regular views on objects of other databases in the cluster.

• Amazon Redshift doesn't support tables with column-level privileges for cross-database queries.

• Amazon Redshift doesn't support query catalog objects on AWS Glue or federated databases. To query these objects, first create external schemas that refer to those external data sources in each database.

• Running cross-database queries on tables with interleaved sort keys isn't supported.