Survivability-Aware Configuration Management of Service-Oriented System Based on Service Dependency

Service-oriented computing (SOC) emerging as a new computing diagram that supports the rapid and low-cost development of distributed applications by composing platform-independent, coarse-grained and loosely coupled services. With the prevalence of service-oriented system, how to build survivable applications in the immature research field of service-oriented system engineering (SOSE) has become a big challenge. An approach of configuration management based on service dependency is presented in this paper to analyze and optimize the survivability of applications. Three dimensions, i.e. distance, strength and criticality, are defined to describe the survivability of service dependency. A directed acyclic graph called weighted system configuration graph (WSCG) is proposed to model the survivability of an application, where each direct dependency between two services is assigned a weight which is calculated based on the three dimensions. Then we give an algorithm in WSCG to obtain the maximal survivability system configuration from redundant service configurations. The mobile video conference (MVC) scenario is illustrated to show the effectiveness and the soundness of our approach and algorithms. At last, the conclusion is drawn with future work given.