虚拟环境下基于市场机制的资源分配方法  

Market-based resource allocation strategy in virtual environment

作　　者：游新冬[1] 徐向华[1] 万健[1] 俞东进[1] 

机构地区：[1]杭州电子科技大学计算机学院,浙江杭州310018

出　　处：《华中科技大学学报（自然科学版）》2010年第6期48-51,共4页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家重点基础研究发展计划资助项目(2007CB310900);国家自然科学基金资助项目(60873023;60973029);浙江省科技厅重大专项资助项目(2008C11099-1)

摘　　要：为了提高资源的利用率和更好地满足不同负载的QoS,提出一种基于市场机制的虚拟资源自适应分配策略:根据负载的特性,构建了体现其对所获资源的满意程度和符合自身经济利润的效用函数,利用资源代理不断调整价格的方式达到提高和合理化资源利用率,同时利用遗传算法最大化每个负载的效益.定义市场模型的均衡状态并证明了均衡状态的最优性.仿真实验验证了资源调价算法的有效性和可行性.In order to use resources effectevely and to provide higher QoS （quality of server） the applications, a marked based resource allocation mechanism was proposed. According to workload characteristic, a QoS-refectitive utility function was constructed. The prices of the resources were dynami- cally adjusted by the corresponding resource agents based on the requirement of all of the workloads. Genetic algorithm was employed to maximize profit of a certain workload. The equilibrium state in our market model was defined and a proof of its optimality is given. The results of simulation experiments validate the proposed resource allocation strategy is effective and practicable.

关 键 词：资源分配 虚拟技术 遗传算法 服务合并 市场机制 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]