Log-based service diagnosis method in cloud

Abstract

Cloud computing is becoming an important and popular platform of service computing for supporting on-demand service sharing in the large-scale distributed systems. One of the difficult challenges in cloud computing is how to deliver the reliable web service composition over unreliable web services around the world. To improve the automated diagnosis capability for service composition in cloud, we present diagnosis architecture and a log-based diagnosis method. By decoupling diagnosis service to share diagnosis resources, our diagnosis architecture could select an optimal diagnosis service according to the faulty type. By converting the successful activity execution logs into the diagnosis model, our diagnosis method can identify the faulty activities based on the differences between correct and observation execution traces. Case study shows that our method is effective in diagnosing service faults in cloud.