Research On QoS-Assured Service Routing Algorithm In Distributed Service Network

Services in the Internet become more and more for the rapid development of the technology and scale of the Internet. Gradually, Services are changing from a single data transmission to a variety of integrated multi-media information transmission. At the same time, the user's demand for ever-changing. Service composition can fast and flexible compose existed independent service components to a new service.The crux of the problem of service composition is the problem of finding a service path, that is the problem of service routing algorithm. The algorithm considers not only whether the request demand is satisfied but also the QoS demand. The research of QoS-assured service routing algorithm is one of the most important and attractive core studying. And it is also a difficult problem. The current service routing algorithms are prevalent can not meet the QoS demand, difficult to implement, low in the success rate and there is no a standard service network structure.For the above problem, this paper made the following work:First of all, for the problem of current service network structures, we used a improved service network architecture. It consists of a number of service nodes and service composers. Service nodes provide a variety of services, and the service composer handle client service requests and find appropriate service path to meet the demand. Therefore, it has good scalability and service performance and provide system-level load balancing and resource management.Secondly, we obtained the vector length method which used to measure the service path through the analysis of the single optimization problem of service path. The method also can be used in the multi-constraint QoS performance optimization problem. From this we get non-linear function F. And simulation show that, F can achieve a good service success rate, especially in the case of the scarcity of resources, the service success rate is improved 5%.A bottle-neck selection algorithm– BNS is also proposed. The algorithm rules out part of the candidate nodes or links of services which most unlikely to become the"bottleneck"to reduce the scale of the candidate service graph and improve the speed of searching service path.Finally, we present and discuss the loop problem, resources competition problem, load balancing problem, service path maintenance and recovery problem for the practical application of service routing algorithm and the corresponding solutions.