An Active and Opportunistic Service Replacement Algorithm Orienting Transactional Composite Service Dynamic Adaptation

Various exceptions leading to the business process no response temporarily or service interruption are common in dynamic service environment,which reduce the reliability of business process greatly.Due to absent of support,most existing replacement mechanisms are of poor adaptabilities.Therefore,the atomicity and the consistency of the transactional service can not be well assured during the system implementation process,and the correctness,the real-time performance and the efficiency of the replacement process can not be guaranteed either.This paper takes the transaction as the core and proposes an active and opportunistic transactional composite service replacement algorithm,which fully takes into account the multiple relationships among Web services and the transactional characteristics of Web services.In this paper,the authors first present a mechanism to obtain the transactional granularity and a method to identify the cascading compensation range,then,propose a transactional replacement cost/benefit QoS model,which completes service replacement with the least cost by integrating transactional compensation service QoS and service replacement QoS.Further,a novel QoS-driven services replacement algorithm with the transactional compensation support is designed.In order to enhance the real-time performance,the authors introduce the early prediction pattern mining to identify the failed or unavailable services as earlier as possible by adopting a real-time monitoring to the running services.Experimental results show that the proposed model not only guarantees the atomicity and the data consistency of the process business in the replacement,but also improves the reliability and the real-time performance of the system.