While there has been a lot of effort in recent years in optimising Big Data systems like Apache Spark and Hadoop, the all-to-all transfer of data between a MapReduce computation step, i.e., the shuffle data mechanism between ...