[1]陶军.一种基于多Agent的动态负载平衡算法应用研究[J].东南大学学报(自然科学版),2003,33(4):401-405.[doi:10.3969/j.issn.1001-0505.2003.04.006]
 Tao Jun.Application research of a dynamic multi-agent-based load balancing algorithm in heterogeneous distributed system[J].Journal of Southeast University (Natural Science Edition),2003,33(4):401-405.[doi:10.3969/j.issn.1001-0505.2003.04.006]
点击复制

一种基于多Agent的动态负载平衡算法应用研究()
分享到:

《东南大学学报(自然科学版)》[ISSN:1001-0505/CN:32-1178/N]

卷:
33
期数:
2003年第4期
页码:
401-405
栏目:
计算机科学与工程
出版日期:
2003-07-20

文章信息/Info

Title:
Application research of a dynamic multi-agent-based load balancing algorithm in heterogeneous distributed system
作者:
陶军
东南大学教育部计算机网络和信息集成重点实验室, 南京 210096
Author(s):
Tao Jun
Key Laboratory of Computer Networks and Information Integration of Ministry of Education, Southeast University, Nanjing 210096, China
关键词:
工作站网络 动态负载平衡 多Agent 并行计算
Keywords:
network of workstation dynamic load balancing Multi-Agent parallel compute
分类号:
TP393
DOI:
10.3969/j.issn.1001-0505.2003.04.006
摘要:
在介绍动态负载平衡算法基本原理的基础上,提出了一种基于多Agent的负载平衡算法DMAL,定义了节点机可能的4种状态,并讨论了DMAL算法的实现规则; 描述了基于DMAL算法的任务调度模型; 讨论了DMAL算法中任务的分类问题、任务的迁移粒度问题、负载迁移的触发时机问题; 为了增强在负载平衡时处理节点间的通信能力,引入了多Agent技术,并为异构分布式处理系统提出了一种基于DMAL算法的负载平衡框架结构; 最后通过实验仿真,比较了DMAL算法、现有的负载平衡算法以及不作负载平衡的运行结果.实验结果表明,基于DMAL算法的分布式处理系统工作在不同负载和应用问题数据规模的情况下,在任务的处理时间、丢弃率和平均响应比方面都显示出良好的性能,从而验证了该算法方案的可行性和有效性.
Abstract:
Based on the principle of dynamic load balancing algorithm, a dynamic Multi-Agent-based load balancing(DMAL)algorithm is put forward. Four possible states of processor node are defined and the rules of DMAL algorithm are discussed. The task assigning modal based on DMAL algorithm is described and task classifying, size of task transfer granularity and the occasion to trigger load transfer operation are also discussed. To strengthen the node-node communication during load balancing Multi-Agent technology is introduced and a dynamic load balancing framework based on DMAL algorithm for heterogeneous distributed system is proposed. Comparing the execution results of proposed algorithm with other available methods, the experimental simulation shows that under different workload and data scales of applications condition the distributed system based on DMAL algorithm possesses excellent performance on task execution time, discard ratio and average response ratio. The feasibility and effectivity of this scheme is thus verified.

参考文献/References:

[1] Rajagopalan Ashok,Hariri Salim.An agent based dynamic load balancing system[A].In: HiPC’2000[C].Bangalore,India,2000.164-171.
[2] 陈华平,计永昶,陈国良.分布式动态负载平衡调度的一个通用模型[J].软件学报,1998,9(1):25-29.
  Chen Huaping,Ji Yongchang,Chen Guoliang.A universal model of distributed dynamic load balancing [J].Journal of Software,1998,9(1):25-29.(in Chinese)
[3] Jovanovic Zoran,Maric Slavko.A heuristic algorithm for dynamic task scheduling in highly parallel computing systems[J]. Future Generation Computer Systems,2001,17(6):721-732.
[4] Arjen S,Peter M,Martin L,et al.A framework for dynamic load balancing:a case study on explosive containment simulation[J]. Parallel Computing,2000, 26(6):737-751.
[5] Berzins M.A new metric for dynamic load balancing [J]. Applied Mathematical Modelling,2000,25(2):141-151.
[6] 方碧琪,方开泰.聚类分析[M].北京:北京大学出版社.1994.99-260.
[7] Zaki Mohammed Javeed,Li Wei,Parthasarathy Srinivasan.Customized dynamic load balancing for a network of workstations[A].In: HPDC1996[C].New York,1996.282-291.
[8] Balabanovic M,Shoham Y.Fab:content-based,collaborative recommendation [J].Communication of the ACM,1997,40(3):66-72.
[9] Konstan A,Miller N,Maltz D,et al.GroupLens:applying collaborative filtering to usenet news[J]. Communication of the ACM,1997,40(3):77-87.
[10] Terveen L,Hill W,Amento B,et al.Phoaks:a system for sharing recommendations [J]. Communications of the ACM,1997, 40(3):59-62.
[11] Foner Leonard N.Yenta:a multi-agent,referral-based matchmaking system[A].In: Agents’ 97[C].California,USA,1997.301-307.
[12] Lemaitre Christian,Excelente Cora B.Multi-Agent network for cooperative work[J].Expert Systems with Applications,1998,14(2):117-127.
[13] External Interfaces Working Group AROA Knowledge Sharing Initiative.Specification of the KQML Agent — communication Language,working paper[EB/OL].http://www.cs.umbc.edu/kqml/papers/kqmlspec.pdf.1993-6-15/2002-12-16.

备注/Memo

备注/Memo:
基金项目: 江苏省自然科学基金资助项目(BK2001205)、国家自然科学基金重大研究计划资助项目(90104009).
作者简介: 陶军(1975—),男,博士生,juntao@seu.edu.cn.
更新日期/Last Update: 2003-07-20