云环境下支持虚拟集群的跨层协同调度机制研究

项目名称： 云环境下支持虚拟集群的跨层协同调度机制研究

项目编号： No.61502165

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 自动化技术、计算机技术

项目作者： 童钊

作者单位： 湖南师范大学

项目金额： 21万元

中文摘要： 本项目对云计算环境下虚拟集群的资源管理及任务调度展开研究，在虚拟化层、存储层和计算层三个层面上对资源的调度进行协同，拟在以下几个方面展开研究：首先，分析不同用户虚机对物理机资源的竞争关系，在此基础上提出虚机动态配置的自适应调节策略，虚机捆绑策略和虚机逻辑簇的性能评估方法。然后，针对虚拟集群动态配置的特性和数据本地性问题，建立物理机+虚机的双层冗余存储机制，包括容量分配、存储方法和替换策略。最后，从计算层任务调度角度，设计满足应用可用性和响应时间约束的双目标调度算法，并能够缓解数据本地性问题，提高虚拟集群的性能。本项目拟通过对云计算环境中支持虚拟集群的调度机制和算法的探索，为云资源的更高效管理和利用提供一定的理论和技术支撑。

中文关键词： 服务调度；服务能力；服务质量；动态协同

英文摘要： This paper is the study on the resource management and task scheduling of virtual cluster in cloud computing environment. Coordination of the resource scheduling is made on virtualization layer, storage layer and calculation layer and breakthroughs are expected to made in the following aspects: First of all, the relation between the virtual machine and physical machine of different users in resource competition is analyzed. Based on this, the self-adaption adjustment tactic of dynamic virtual machine configuration is put forward, so is the assessment method on the performances of binding strategy and logical cluster of the virtual machine. Then, based on the features of dynamic configuration for virtual cluster and the local feature of data, a double-layer redundancy storage mechanism for physical and virtual machines will be established, including capacity distribution, storage method and replacement policy. At last, from the perspective of task scheduling on calculation layer, the design shall satisfy double-objective scheduling algorithm in terms of application availability and response time restriction and shall mitigate the problem in data localization so as to improve the performance of virtual cluster. This paper tries to propose a series of scheduling algorithm in order to provide a broader research approach in optimizing the performance of cloud computing environment which supports virtual cluster.

英文关键词： Service scheduling;Service ability;Quality of service;Colabrate dynamically