云计算环境中面向虚拟资源的能效模型及算法研究

Energy Management and Control is an important issue in Cloud computing. The challenging issue is how to provide a fine-grain and adaptive mechanism to realize optimal energy-consumption as well as energy-efficient management.In this project, we will study on the energy/power model of virtual resources at first. Then, we plan to apply stochastic service theory to describe the working status of virtual resources, in this way, we expect to propose an approach to quantitatively analyzing the relationship between service capability and energy consumption. Based on such an analyzing approach, we will develop an virtual resource supplying and configuration framework, which is aiming to optimize both the energy-consumption and the energy-efficiency in an integrated manner. This project is expected to solve the problem of energy-consumption evaluation in virtual resource environments, so as to push forward the studies on fine-grain and adaptive energy-consumption/energy-efficient management. In addition, the energy-aware framework is expected to be used in practical cloud-based systems and applications, so as to reduce the cost of energy consumption and improve the energy efficiency in high-performance data centers and computational centers.

能耗管理与控制属于云计算领域的基础性研究课题之一，其难点与挑战在于如何在资源虚拟化环境中有效地实现细粒度和自适应的能耗/能效优化管理。本课题以研究虚拟资源的能耗模型为起点，提出采用随机服务理论来建立虚拟资源的抽象服务模型，从而提出一套量化分析虚拟资源"服务能力-能耗"之间关系的方法，最后基于该方法构建一个面向"能耗-能效"双目标优化的虚拟资源供给与配置模型。本课题的研究有望解决虚拟资源在动态负载环境中的能耗评估问题，从而有效推进云计算环境中细粒度和自适应的能耗/能效优化管理研究。此外，课题研究的"能耗-能效"双目标优化模型有望直接应用于实际的云计算系统，从而降低数据/计算中心的能耗成本，并提高云计算系统的总体能效指标。

能耗管理与控制属于云计算领域的基础性研究课题之一，其难点与挑战在于如何在资源虚拟化环境中有效地实现细粒度和自适应的能耗/能效优化管理。本项目以研究虚拟资源的能耗模型为起点，提出了“基于相对性能监控计数器的有限偏差虚拟机功耗测量方法”。进而量化分析了系统负载对虚拟资源能耗的影响。最后为云环境下面向虚拟资源的资源供给与配置提供了一组“能耗-能效”双目标优化策略。经过四年来的深入研究，本项目提出了一套完整的面向虚拟资源的能耗/能效优化管理体系，较好地解决了云计算环境中细粒度和自适应的能耗/能效优化管理的关键问题。.本项目在SMARTCOMP 2014、WISE 2016等高水平国际会议和《Journal of Network and Computer Applications》、《Journal of Computer Science and Technology》、《Computers and Electrical Engineering》、《Frontiers of Computer Science》、《通信学报》、《计算机集成制造系统》等重要期刊发表及录用学术论文30篇，其中SCI检索9篇，EI检索14篇，申请发明专利2项，部分研究成果达到国际一流水平。