Hadoop, the open-source implementation of Google’s MapReduce, has become enormously popular for big data analytics, especially among researchers. Due to Hadoop’s popularity, it is natural to ask the question: how well is it working to answer this question, the authors need to move beyond the conventional “Cluster-centric” analysis that models each job as an independent black box and focuses on coarse global metrics such as resource utilization, usually in the context of a single cluster.