ClickHouse now writes Apache Iceberg tables to Microsoft OneLake

Summary

ClickHouse previously wrote Iceberg tables to object storage without catalog registration, limiting their discoverability. Now it writes Iceberg tables directly to Microsoft OneLake, registering in the catalog via Iceberg REST Table APIs and securing with OneLake Security. Use cases include AI agents monitoring order data and publishing real-time anomaly detection results to OneLake, and applications writing directly via ClickHouse without a separate ingestion layer. The connection uses the DataLakeCatalog engine with catalog_type=‘onelake’, warehouse parameter combining workspace and lakehouse IDs, and Microsoft Entra ID service principal authentication; writing uses standard INSERT INTO syntax with backtick-quoted namespace.table names. OneLake automatically syncs Iceberg and Delta Lake formats, provides governance with lineage tracking and data protection, and its security APIs are opening to third-party engines; this is the first catalog to support writes from ClickHouse, with more catalogs planned.

Takeaways

1. 1

   ClickHouse writes Iceberg tables to OneLake using the DataLakeCatalog engine and INSERT INTO syntax, with tables automatically registered in the OneLake catalog.

2. 2

   OneLake automatically syncs Delta Lake and Iceberg formats, so data written as Iceberg is readable by Delta Lake tools.

3. 3

   OneLake Security provides table, row, and column-level access controls, and Microsoft announced APIs are opening to third-party engines.

FAQ