Dependency Based Approach for Software Analysis Web Services Replacement

The concept of Software as a Service has been recently introduced, allowing software to be accessed, composed into workflows, and executed over the Internet. However, there is no guarantee on the execution reliability of such workflows. In case of component service failure, that particular service has to be replaced quickly to ensure the correct execution of the composite service. While most of the existing replacement approaches require the faulty workflow to be completely re-executed, we advocate that dependencies among component services is vitally important for service replacement. In this paper, we present a dependency-based component service recommendation approach that takes into account the dependencies among the component services, such as input/output, pre and post conditions, and correlation among services. Correlation analysis is a new dimension that we introduce to the service replacement algorithms. To demonstrate our approach, we use an existing software analysis services framework as a case study. Finally, we argue that dependencies among services can affect the accuracy of the replacement.