This is exactly what I was thinking as well. It seems that we are slowly moving from the Hadoop ecosystem to the Spark ecosystem.