Is Apache Spark really a MapReduce replacement? All the recent technical books I've read look to MapReduce as a thing of the past.