一种基于超边际分析的分布式计算资源分配方法  

An Inframarginal-Analysis-Based Resource Allocation Method in Distributed Computing Environment

作　　者：王恩泽[1] 乔建忠[1] 林树宽[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110819

出　　处：《东北大学学报（自然科学版）》2011年第2期219-222,231,共5页Journal of Northeastern University(Natural Science)

基　　金：国家自然科学基金资助项目(60873009);辽宁省重点科技攻关项目(R200801036)

摘　　要：针对分布式计算环境,提出一种基于超边际分析的资源分配方法,描述了基于比较优势的资源分配框架,建立了具有交易保护的IRDL模型以及自由交易的IRD模型,通过角点分析得到全局均衡,以达到系统效用最大化.与新古典经济学研究如何提高资源利用率不同,本文基于新兴古典经济学,通过分工理论降低资源稀缺程度,动态调整计算网络以达到整体帕累托最优.仿真结果表明上述两种模型均合理、有效,而IRD模型更适用于分布式计算环境.A resource allocation method based on inframarginal analysis in distributed computing environment is presented to describe the resource allocation framework based on its comparative advantage,which maximizes the system utility by the global balance resulting from corner analysis,with the IRD（inframarginal-analysis-based resource allocation model for distributed computing） model for free transaction and IRDL（IRD limited） model for transaction limitation both developed.Differentiating from the neoclassical economic theory which studies how to improve the resource utilization ratio,the new classical economic theory studies how to decrease the resource scarcity through the division of labor so as to achieve the Pareto Optimality entirely by regulating the network dynamically.Simulation results showed that both the models developed are effective and rational though the IRD model is more suitable for the distributed computing environment.

关 键 词：分布式计算 超边际分析 资源分配 比较优势 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]