计算机应用 ›› 2011, Vol. 31 ›› Issue (03): 834-838.DOI: 10.3724/SP.J.1087.2011.00834

• 典型应用 • 上一篇    下一篇

因特网上基于节点角色的计算资源共享平台——RB-CRSP

张学锋1,徐胜超2   

  1. 1. 惠州学院 数学系,广东 惠州516007
    2. 广东邮电职业技术学院 计算机系,广州510630
  • 收稿日期:2010-09-13 修回日期:2010-11-03 发布日期:2011-03-03 出版日期:2011-03-01
  • 通讯作者: 徐胜超
  • 作者简介:张学锋(1968-),女,甘肃民勤人,讲师,主要研究方向:计算机网络、网络安全;徐胜超(1980-),男,湖北武汉人,讲师,硕士,CCF会员,主要研究方向:并行与分布式计算、对等计算。

RB-CRSP: Node role-based computing resources sharing platform over Internet

ZHANG Xue-feng1,XU Sheng-chao2   

  1. 1. Department of Mathematics, Huizhou University, Huizhou Guangdong 516007, China
    2. Department of Computer, Guangdong Vocational College of Posts and Telecom, Guangzhou Guangdong 510630, China
  • Received:2010-09-13 Revised:2010-11-03 Online:2011-03-03 Published:2011-03-01
  • Contact: XU Sheng-chao

摘要: 提出了因特网上基于节点角色的计算资源共享平台——RB-CRSP。设计时充分考虑节点的角色性和功能性,把因特网上的网络资源按照角色划分为服务器端节点、协调节点、工作机节点与客户机节点四类实体,通过配合RB-CRSP的应用编程模式,完成并行分布式计算。分析了RB-CRSP中的自适应资源调度策略,该策略考虑了节点的硬件信息与可信誉机制,实现了平台的负载均衡性;在动态的因特网环境下,利用面向工作机的容错方式保证了平台的可靠性。案例程序选择了典型的并行BenchMark程序:N皇后问题,测试结果表明,RB-CRSP可以方便聚集异构环境下的空闲计算资源,平台的性能与机器硬件条件和可靠性密切相关。

关键词: 分布式计算, 因特网计算, 资源共享, 并行处理, 编程模式

Abstract: A role-based computation resource sharing platform called RB-CRSP was proposed in this paper. The node roles and functions were strongly considered in design toplogy of RB-CRSP. All the nodes in RB-CRSP were assigned as server, coordinator, worker, and client. By using the united programming model, it could be in harness with the immense computational resource available in the Internet for parallel and distributed computation efficiently. RB-CRSP platform was also load balancing and fault-tolerant, which was strongly supported by hardware and credit standing based scheduling algorithm and the workers oriented fault tolerance policy. In order to demonstrate the effectiveness of RB-CRSP, a serial of simulation experiments were done. The results obtained from performance analysis show that RB-CRSP is feasible and efficient which can provide a new way for computing resource sharing over the Internet. The hardware conditions and availability of machines are important factors in this test.

Key words: distributed computing, Internet computing, resource sharing, parallel processing, programming model

中图分类号: