ARCN-Tree:多集群P2P网格系统中计算资源查找框架研究与实现

ARCN-Tree:research and implement of multi-cluster grid P2P system resources lookup framework

为了提高多集群网格系统完成计算密集和数据密集型作业的效率,本文提出了一种新型的基于多集群P2P网格系统的计算资源发现框架.该框架的核心是如何构建分布式的Ascendant Rendezvous Cluster Node Tree(ARCN-Tree).针对集群的异构性,提出一种优势集群选择SCEA算法.通过优势集群选择,在底层chord结构之上构建优化的ARCN-Tree.通过对比实验中看出,资源查找的效率得到有效的提高.

A optimize resource lookup algorithm, namely ascendant rendezvous cluster node tree（ARCN- Tree）, is proposed to improve locate the resources efficiency. The authors designed and implemented the ARCN-Tree algorithm to lookup distributing computing resource. Theory analysis shows with the preference cluster chosen algorithm the average lookup hops length of ARCN-Tree is shorter than the RN- Tree＇s. The simulation results show that compare to other lookup algorithm the SCEA algorithm can decrease the lookup overheads greatly and improve the performance of multi-cluster grid system.