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:
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_byordering to keep keys stable. - Type casts:
TRY_CAST → TO_VARCHARand related casts are adapted for Snowflake SQL (part of thesnowflake-engine-fixeson 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.