Part 3 of a Snowflake & Databricks interoperability series covering the 'External-First' Catalog-Linked Database (CLD) pattern. Databricks writes Iceberg tables to cloud object storage managed by Unity Catalog, while Snowflake reads them zero-copy via the Iceberg REST API. The post walks through production setup including service principal creation, catalog integration, CLD creation, and the 6 required grants (highlighting the commonly missed integration grant). It covers performance characteristics, governance via a parallel model, and an advanced pattern using Dynamic Tables with Cortex AI functions for enrichment on top of CLD data — all without moving data between platforms.
Sort: