Skip to content

Snowflake

Engine: Snowflake SQL · Format: native Snowflake tables · Storage: Snowflake database · Status: ✅ Live.

Reference data mesh lakehouse: lakelogic-snowflake-data-mesh-lakehouse — the full runnable RideFlow mesh lives there: the self-deploying bootstrap.sql + setup.sql, the native Snowflake Project (notebooks + a Task DAG), and every table it materializes.

The same contract

No change from the canonical RideFlow contract. Backend choices:

materialization:
  strategy: merge        # native Snowflake tables; Iceberg deferred to Horizon

Special configuration

How the LakeLogic framework adapts to this backend (handled for you):

  • Identifier casing: Snowflake upper-cases unquoted identifiers, so the framework uses UPPERCASE casing and a matching dedup sort_by ordering to keep keys stable.
  • Type casts: TRY_CAST → TO_VARCHAR and related casts are adapted for Snowflake SQL (part of the snowflake-engine-fixes on the reference framework).
  • Shared connection: the engine sets a shared Snowflake connection so notebooks and tasks in the Project reuse one session.
  • Iceberg: native Snowflake tables today; Iceberg tables are deferred to the Horizon path.