An internal catalog keeps table registration, metadata updates, and access coordination inside the catalog service while table data and metadata remain in configured cloud storage. Query engines interact through the Iceberg REST interface, separating storage from compute and supporting shared access across engines. section 7.4.10 section 7.4.4
Operational characteristics include:
- Supports multi-engine access from Spark, Flink, Trino, Snowflake, or other Iceberg REST-compatible engines. section 7.4.10
- Applies centralized RBAC through service principals, principal roles, and catalog roles at catalog, namespace, and table levels. section 7.4.4
- Uses credential vending to provide temporary runtime access without exposing cloud storage credentials directly. section 7.4.10
- Contrasts with external catalogs, whose synchronized tables are read-only from the managing service's perspective. section 7.4.4
