| With the fast development of the Internet, the demands of end-users accessing to network resources have been increasing rapidly. Therefore, the Internet content providers are often faced with the problems of resource over-provisioning, degraded performance, and the decline of QOS (Quality of Service), which not only enhance their operational risks and costs but also restrict the scale and scope of their services. In consider of these shortcomings, today more and more Internet content providers primarily utilize Content Delivery Network(CDN) to deliver contents to end-users, aiming to improve users’ web access experience, reduce content providers’ operational risks and costs, and expand the scale and scope of their services.This thesis introduces working principle to enable cooperation and coordination among multiple replica servers, which is termed as Content Delivery Network (CDN), and also studies the related techniques of CDN. Differentiated from other researches in this field, this thesis concentrates on the CDN utility computing. On the basis of utility computing, this thesis develops the solutions of content delivery performance optimization that are advanced, efficient, scalable, high-performance and cost-effective. Besides, a fact has been proved that those solutions are effective in moving beyond the shortcomings of resource over-provisioning and degraded performance from existing content providers, and improving Web access experience of Internet end-users. The main contributions of this thesis are as follows:1) This thesis identifies the key research challenges and core technical issues for CDN, along with an introduction to the evolution of CDN and a systematic understanding of the CDN space by covering relevant implementation techniques;2) This thesis introduces a service utility model of CDN to measure its content-serving ability by capturing the traffic activities in the system as well as reflecting the collaboration among the replica servers. Additionally, the model provides an introduction to the concept of service benefit and develops quantitative criteria for CDN content-serving ability. The service utility model provides the basis for developing a better solution of content delivery performance optimization;3) This thesis develops the solutions of content delivery performance optimization for CDN, i.e. server selection policies and request-redirection algorithms, to improve system service benefit, alleviate imbalanced load conditions and minimize redirection cost. In this thesis, two request-redirection algorithms (a greedy algorithm and a heuristic algorithm) are proposed as the solutions of content delivery performance optimization. Based on CDN service utility model, the two algorithms both utilize the quantitative criteria of content service utility in the model; and4) This thesis performs simulation experiments for CDN service utility model and the proposed solutions of content delivery performance optimization. Comparisons on the analysis of experimental results demonstrate the overall improved performance of CDN and enhanced web access experience of Internet end-users. |
