Run Spark and Hadoop faster with Dataproc