Iceberg-native databases decouple data storage from query engines. That means you can store data once in an open table format and let different engines read and analyze it without rewriting the data for each tool [1]. This setup could make analytics more flexible as workloads evolve and new engines enter the scene.
What iceberg-native means — The main selling point of open table formats is to decouple the data from query engines [1]. Isn't that almost what we got with DuckDB + DuckLake [1]?
Concrete examples — In practice, projects like DuckDB paired with DuckLake illustrate decoupled storage and compute, showing the practicality of engine-agnostic data layouts [1].
Benefits for interoperability and evolving workloads — • Interoperability across engines and tools, since storage is decoupled from engines [1] • Evolving analytics workloads can be supported by swapping compute engines without touching stored data [1] • The design hints at a storage-first, engine-agnostic architecture and tradeoffs in performance and flexibility [1]
Closing thought: as this approach matures, SQL design and optimization may start treating storage as the shared backbone rather than a single-vendor silo [1].
References
The case for an iceberg-native database
Argues for an iceberg-native database; cites open table formats decoupling data from engines; nods to DuckDB and DuckLake.
View source