Achieving Load Balance in a Hierarchical Peer-to-Peer Architecture

With the fast development of networking, the demands made of computers are greater than ever before. Determining how to utilize the resources of networking to reach the objective of cooperative computing has remained an important topic in recent years. In addition, with the great advances in technology of the Internet, Peer-to-Peer (or P2P) computing has gradually become the mainstream of distributed applications; it not only provides enormous resources for complicated computing that a single computer cannot solve, but also integrates resources more effectively. A P2P architecture relies primarily on the computing power and bandwidth of the participants in the network rather than concentrating the work in a relatively limited number of servers. P2P architectures typically are used for connecting nodes via large-scale connections. The topology is useful for many purposes. Furthermore, every joint in a P2P computing system has its own resources. Determining how to take the different characteristics of every joint set into consideration for loading assignments is an important topic. However, in this study, a three-phase scheduling algorithm under P2P architecture is advanced. The proposed scheduling algorithm is composed of BTO (Best Task Order), TOLB (Thresholdbased Opportunistic Load Balancing) and TLBMM (Threshold-based Load Balance Min-Min) scheduling algorithm that can better utilize executing efficiency and maintain the load balancing of system. Key-Words: Distribution system, Peer-to-Peer computing, Scheduling, Load balance