Comment by jerdavis

Just use bigquery. The last thing I want these days is to manage a cluster. Next to last thing I want is to worry about the zillion tuneables on hadoop/spark.