serverless is able to start and scale faster than non-running SQL endpoints (seconds intead of minutes) Stage’s detail screen and Executor’s log files in the Spark UI are two of the primary indicators that a partition is spilling to disk