Joining a large and a small Dataset | Apache Spark - Best Practices and Tuning