Date Added: Mar 2010
The authors present Kahuna, an approach that aims to diagnose performance problems in Map Reduce systems. Central to Kahuna's approach is the insight on peer-similarity, that nodes behave alike in the absence of performance problems, and that a node that behaves differently is the likely culprit of a performance problem. The authors also present empirical evidence of the peer-similarity observations from the 4000-processor Yahoo! M45 Hadoop cluster. In addition, the authors demonstrate Kahuna's effectiveness through experimental evaluation of two algorithms for a number of reported performance problems, on four different workloads in a 100-node Hadoop cluster running on Amazon's EC2 infrastructure.