Based On The Cost-benefit Analysis Method Of Web Service Fault Tolerant Strategy Planning

Web services have been widely used in SOA and Cloud Computing these days, gaining more and more attention by its benefits of loose coupling and platform independence. However, problems like failure and poor reliability of web services caused by the instability and faults of software and hardware has become obstacle of the further development of web services. Fault Tolerance approaches based on the diversity of design and data has been made good use of traditional software. As for fault tolerance in web services, a lot of research and exploration has been done on fault tolerance approaches based on replication strategy. Web service fault tolerance approaches based on various replication strategy has been widely used on the reliability and performance guarantee of critical systems. In response to the problems in web service application, we propose a web service fault tolerance approach supporting multiple quality analysis. In this approach, this paper first raised and gave detailed description of nine calculation methods of cost of web service fault tolerance approaches. What’s more, the calculation formulas are given. These formulas take service price, all compensation and the characteristics of the approach into consideration, making the calculation of cost more accurate. Current web service fault tolerance strategy selection approaches usually take performance related qualities like response time and failure rate into consideration, ignoring the cost of service calling and compensation, leading to the shortcomings of selected strategies in terms of cost effectiveness. Therefore this paper comes up with these cost calculation methods to solve these shortcomings.Secondly, as quality on different fields of web services are not always the same, quality information should be saved as Heterogeneous data. This paper uses the concept of Ontology used in semantic network to organize this heterogeneous web service information, and use SPARQL language to do semantic retrieval. This paper also suggests the use of proxy pattern on the frame of web service fault tolerance, making the process of fault tolerance visible to the user, and maintaining the loose coupling of web services.Thirdly, this paper raised a selection algorithm of replication strategy supporting multiple qualities. This algorithm improves the effectiveness of the fault tolerance strategy with acceptable overhead. In this algorithm, multiple qualities like cost of service, response time, failure rate, compensation cost and so on are taken into consideration, guaranteeing that the final selected replication fault tolerance strategy can satisfy users’ need of performance and at the same time lower the cost of the strategy.Finally, to verify the validity and performance of the algorithm, this paper compares this algorithm with a web service fault tolerance strategy selection algorithm which merely takes one quality into consideration. The result shows that, as the algorithm in this paper involves many qualities, it can better benefit users’need and choose a fault tolerance strategy which has lower cost and acceptable loss of performance. In addition, the result of the experiment shows that the cost of executing the algorithm in this paper will not make apparent influence on the calling process of web services.