Announcing DuckLake 1.0 on MotherDuck

MotherDuck announces preview support for DuckLake 1.0, the first major stable release of the open table lakehouse format from DuckDB Labs. DuckLake stores all metadata in a SQL database rather than thousands of JSON files on object storage, delivering 10x+ faster queries and transaction throughput compared to Apache Iceberg and Delta Lake. Version 1.0 introduces a stable backwards-compatible specification, multi-engine support (DuckDB, DataFusion, Trino, Spark), data inlining for updates/deletes (solving the small files problem), data clustering for faster selective reads, bucket partitioning, geometry types with geospatial predicate pushdown, and a VARIANT type with JSON shredding. On MotherDuck, a full managed lakehouse can be created with a single SQL command, with options for fully managed, bring-your-own-bucket, or bring-your-own-compute deployments.