A QoS-aware system for service composition and management on overlay network

Abstract

The research work of this thesis is motivated by a desire to build up efficient Quality of Service (QoS) insurance framework on overlay network. With the rapid development of Service Oriented Architecture, there is a growing interest in the design of efficient QoS-aware system. The difficulties of building QoS-aware system include service addressing, service composition, service management among others. Therefore, an integrated framework needs to be proposed to fulfill all these features. Meanwhile, load balancing is also a critical issue for solving scalability problem, when achieving high performance for large scale distributed systems due to the limited computing resource in general. In this thesis, we design and implement a JXTA Message Layer based infrastructure for improving the performance and functionalities of JXTA P2P framework, which can be used for designing better SOA based systems. Furthermore, a design of a QoS-aware hierarchical service composition and management framework is proposed and implemented on top of this infrastructure. A comparison experimental study of our design with commonly used flat-based service composition and management is conducted. The result of our experiments show that our design outperforms flat-based one in term of satisfaction of user requirements. A load balancing scheme is also proposed on this framework aiming to optimize the performance of our proposed framework on large scale distributed systems. We compare our load balancing scheme with non-migration and non-load-balancing approaches, and find that our solution based on a combination of task scheduling and service migration paradigm using Genetic Algorithm is much better than the other two in terms of QoS success rate.