Distributed Computing Model and Supporting Technologies for the Dynamic Allocation of Internet Resources

Internet computing becomes more and more popular, such as peer-to-peer, grid and cloud computing. Resource allocation problem is a key problem of all kinds of Internet applications. However, the dynamics of Internet resources makes the resource allocation problem one of the challenging issues on the Internet. Aiming at the issue, this paper presents a computational model of the dynamic Internet resource allocation and some related distributed algorithms. Firstly, the dynamics of the Internet resources is analyzed and the result shows which characteristics are invariable in resource allocation process. Secondly, based on the invariable characteristics during the resource allocation process, an organization model of distributed resources, a computational model of allocating resources and the APIs for use of Internet resources are presented. Thirdly, some distributed resource allocation algorithms on system level, such as publishing resources and requesting resources, are presented to support the models. In addition, the definition of good serving peer and the good serving peer selection algorithm are given. Finally, based on the models, two Internet applications are tested. The experiment results show that the models and the algorithms are effective and that the good serving peer selection algorithm can decrease the ratio of the rejected request drastically.