Fault diagnosis of web service based on trace matching

With the advent of cloud computing, it is a critical challenge to guarantee the reliability of web service. Fault diagnosis is one of the major issues which build the reliable service applications. In this paper, we present an automatic diagnostic framework for simplifying the diagnostic procedure, improving the diagnostic quality and assuring the privacy of service. For locating the service faults and explaining the faulty causes, we propose a trace matching-based diagnosis method. Our method can find a correct execution trace, which has the maximum likelihood with the faulty execution trace. Then our method can identify the faults by finding the differences between these two traces, and explain the faulty causes. We use 2507 real web services to conduct the simulation experiments for evaluating our method. The results of experiments show that our method is effective to web services of various scales with noise.