云计算下保障公平性的多资源分配算法  

Enhanced fairness-based multi-resource allocation algorithm for cloud computing

作　　者：卢笛[1] 马建峰[1] 王一川[1] 习宁[1] 张留美[1] 孟宪佳[1] 

机构地区：[1]西安电子科技大学计算机学院,陕西西安710071

出　　处：《西安电子科技大学学报》2014年第3期162-168,共7页Journal of Xidian University

基　　金：国家科技重大专项资助项目(2012ZX03002003);中央高校基本科研业务费资助项目(JY0900120301)

摘　　要：针对云计算平台多资源分配公平性问题,文中在DRF算法基础上,提出了云计算动态资源需求公平分配模型,并提出了基于信誉因子的增强公平性分配算法.算法引入信誉因子,对云中计算节点资源使用情况进行实时评估,对恶意长时间侵占资源行为进行惩罚性分配,刺激节点在任务结束后释放占用资源,确保了平台中其他节点资源配额不受影响.与现有方案相比,基于信誉的增强公平性分配算法在保证分配公平的前提下,增强了对公平性的保障,有效地确保了云计算平台资源调度的公平性、可靠性.To address the issue of faimess in resource allocation under cloud computing , this paper proposes a dynamic-resource-demand oriented model of fair allocation for the cloud platform based on DRF （Dominant Resource Faimess ） . Then , a credit factor based allocation algorithm of enhanced faimess , named cbDRF , is proposed . A credit factor is introduced to cbDRF to evaluate the resource utilization of the computing nodes on the clud platform . Thus , with the credit factor , the nodes which are maliciously occupying resources for a long time will be imposed with Punitive Allocation . Besides , this mechanism can also encourage the node to release its occupied allocations after its task（ Release incentive） to guarantee other nodes＆#39; share not to be influenced . Compared to the existing approaches , cbDRF strengthens the protection for faimess under the premise of ensuring fair allocation , which effectively guarantees the faimess and reliablilty for the resource scheduling of the cloud platform .

关 键 词：云计算 资源分配 公平性 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]