异构云计算体系结构及其多资源联合公平分配策略

发布时间：2018-01-03 11:07

本文关键词：异构云计算体系结构及其多资源联合公平分配策略　出处：《计算机研究与发展》2015年06期 　论文类型：期刊论文

【摘要】：资源分配策略是当前云计算研究领域中的一个重要研究热点,异构云计算体系结构下的复杂应用问题研究中,最基本的问题在于如何将总体有限的资源分配给多个租户或应用,以达到效率或收效最大化.但是,在经典的资源分配问题中,任务或者用户往往是"贪婪"的;因此,在总体资源有限的前提下,资源分配的公平性就显得尤为重要.为了满足不同的任务需求,达到多种资源分配的公平性,设计了一个虚拟化的异构云计算体系结构,提出了该体系结构下基于占优资源的多资源联合公平分配算法(maximizing multi-resource fairness based on dominant resource,MDRF),并且证明了算法的帕累托等相关属性;给出了占优资源熵(dominant resource entropy,DRE)和占优资源权重(dominant resource weight,DRW)的定义,占优资源熵更加精确地刻画了用户资源请求与任务所调度到的服务器资源之间的适应程度,使系统的自适应能力更强同时提高了资源利用率.占优资源权重保障了用户优先获取资源的优先次序,协同所采用保障公平性的Max-Min Fairness策略,使资源的分配更加有序.实验表明,我们的策略有更高的系统资源利用率,并且使需求与供给更加匹配,进而使用户的占优资源获取更多,提高了服务质量.
[Abstract]:Resource allocation strategy is an important research focus in the field of cloud computing, and complex applications under heterogeneous cloud computing architecture. The most basic problem is how to allocate the overall limited resources to multiple tenants or applications to maximize efficiency or effectiveness. However, in the classical resource allocation problem. Tasks or users are often "greedy"; Therefore, under the premise of limited overall resources, the fairness of resource allocation is particularly important. In order to meet different task needs, to achieve the fairness of resource allocation. A virtualized heterogeneous cloud computing architecture is designed. In this architecture, a multi-resource joint fair allocation algorithm based on dominant resources is proposed. Maximizing multi-resource fairness based on dominant resource. MDRF, and proved the algorithm of Pareto and other related attributes; The dominant resource entropy is given. Dre) and dominant resource weight DRWs. The dominant resource entropy more accurately describes the adaptability between the user resource request and the server resource scheduled by the task. The adaptive ability of the system is stronger and the resource utilization ratio is improved. The dominant resource weight ensures the priority of the user to obtain the resources. Max-Min Fairness strategy is adopted to ensure fairness, which makes the allocation of resources more orderly. Experiments show that our strategy has a higher utilization rate of system resources. And make the demand and supply more match, and then make the user's dominant resources to obtain more, improve the quality of service.
【作者单位】：武汉大学计算机学院;
【基金】：国家自然科学基金项目(61373040,61173137) 教育部博士点基金项目(20120141110073) 新疆维吾尔自治区自然科学基金项目(2015211B030)
【分类号】：TP303
【正文快照】： 源熵更加精确地刻画了用户资源请求与任务所调度到的服务器资源之间的适应程度,使系统的自适应能力更强同时提高了资源利用率.占优资源权重保障了用户优先获取资源的优先次序,协同所采用保障公平性的Max-Min Fairness策略,使资源的分配更加有序.实验表明,我们的策略有更高的系

【参考文献】