Evaluation of Failure Detectors: A Decision Theory Approach

This paper proposes a method of analysis for evaluating failure detectors. The method can be applied to compare the performance of failure detectors in a given environment. Through a cost analysis approach, this method creates a cost metric, averagelost, which might tackle with comparing two failure detectors. Through the consequent experiment, we conclude that the value of a failure detector depend on the user’s request, the probability of process crash and the network environment where the failure detector runs.