Optimizing Apache Spark™ on Databricks