Snowflake data lineage now tracks integrations and Cortex Search

Snowflake's data lineage feature has been expanded to map new connections among storage, AI, and search objects. You can now track how storage integrations feed stages, how stages feed external tables, and map model versions to their monitors and Cortex Search services.

Use cases

Imagine you need to audit the origin of data in an external table or trace the source dependencies for a search service. With this update, you can:

• Trace dependencies from a storage integration down to a stage and its external table.
• View the relationship between ML model versions and their respective model monitors.
• Identify the source tables and dynamic tables used by Cortex Search services created from a stage.

How to use

To view the updated lineage, open the Catalog in Snowsight and select the Lineage tab on a supported object.

SELECT * FROM TABLE(SNOWFLAKE.CORE.GET_LINEAGE('my_db.my_schema.my_table'));

What happens in this code:

1. The GET_LINEAGE function is called with the fully qualified name of the target Snowflake object.
2. Snowflake returns the programmatic lineage information for the requested object.

Limitations and requirements

• Data lineage requires Snowflake Enterprise Edition or higher.
• Relationships are recorded when the object is created. Objects created before this update must be recreated to display the new lineage.
• Creating a Cortex Search service from a stage is in preview.