Fault Tolerance for Web Service Based on Component Importance in Service Networks

Service Oriented Architecture (SOA) is a widely used computing model in heterogeneous network environment. In the study of SOA, the reliability of Web Service Composition is a key issue concerned by researchers and many previous studies have focused on the optimization of fault-tolerance strategies. In this paper we make a comparison between service networks and complex networks, and then apply the node ranking algorithms in complex networks to service networks. Two algorithms are proposed: PageRank based Service Component Ranking Algorithm (PSCR) and HITS based Service Component Ranking Algorithm (HSCR). With PSCR and HSCR, service components are ranked and those with higher PageRank or HITS values will be chosen to have stronger fault-tolerance abilities in order to improve the reliability with a low cost. Experimental results on a real-world dataset show that PSCR and HSCR algorithms perform better than other component ranking algorithms.