Skip to main content
RisingWave supports two interaction modes with Apache Iceberg: internally managed Iceberg tables (often called RisingWave-managed Iceberg tables) and externally managed Iceberg tables. In internally managed Iceberg tables, RisingWave manages the entire lifecycle of the Iceberg tables, including the catalog, compaction, and ingestion processes. RisingWave provides a PostgreSQL-compatible interface for accessing Iceberg data and ensures interoperability with all major query engines. This mode offers the simplest way to build a Medallion architecture. In externally managed Iceberg tables, RisingWave connects to Iceberg tables managed by external systems. RisingWave can both read from and write to these tables, enabling seamless data exchange across different environments. Rule of thumb: Internal tables → CONNECTION + ENGINE = iceberg. External tables → SOURCE (read) / SINK (write).
ConceptWhen / what (internal vs external)Key SQL
CONNECTIONInternal — save Iceberg catalog + object storage config for ENGINE = iceberg. See Internal Iceberg tables.CREATE CONNECTION + SET iceberg_engine_connection = 'schema.conn'
SINKExternal — write streaming results to an external Iceberg table. See Deliver to external Iceberg.CREATE SINK ... WITH (connector = 'iceberg', ...)
SOURCEExternal — read an external Iceberg table into RisingWave. See Ingest from external Iceberg.CREATE SOURCE ... WITH (connector = 'iceberg', ...)

​
RisingWave in a medallion architecture

RisingWave bridges real-time streaming systems with Iceberg-based data lakes. It functions as the continuous ingestion, transformation, and maintenance service within a Medallion architecture, ensuring that data seamlessly transitions from raw sources to optimized, query-ready Iceberg tables through incremental computation and cascading materialized views. image

​
Next steps