Summary: | 碩士 === 國立臺中教育大學 === 資訊工程學系 === 107 === Nowadays, cloud computing and the SaaS model have changed the way for people to develop and use software. Composite SaaS applications based on the Service-Oriented Architecture (SOA) are becoming a common and important application software service model. A major concern is efficient execution of composite SaaS applications. This thesis presents our study on service deployment and component service request scheduling for improving the overall performance of executing composite SaaS applications. A waiting-queue based iterative service deployment approach, considering both inter-service communication costs and dynamic execution parallelism, together with different component service request scheduling methods are proposed for batch processing of composite SaaS applications. The proposed approach was evaluated with a series of simulation experiments. Experimental results show that our approach has the potential to outperform previous methods significantly, reducing at most 47% service response time in the experiments. This thesis also points out some promising research directions worth further studies, including the relationship between performance metrics and component service request scheduling methods as well as the converge speed and algorithm efficiency of the iterative approach.
|