Snowpark uses a Spark engine to generate optimized SQL query plans.

Snowpark automatically sets up Spark within Snowflake virtual warehouses.

Snowpark does not require that a separate cluster be running outside of Snowflake.

Snowpark allows users to run existing Spark code on virtual warehouses without the need to reconfigure the code.

Snowpark executes as much work as possible in the source databases for all operations including User-Defined Functions (UDFs).