Collaborative Research: Autonomic Power and Performance Management in Distributed Computing Systems

该项目的目标是通过(a)对功耗和性能进行在线建模、监测和分析，为分布式计算中心的自主功率和性能管理开发一个理论框架和分层方法；(b)自适应学习和自动识别策略，以尽量减少耗电量，同时为各种工作负荷和应用维持所需的服务质素；(c)根据所选择的优化策略对计算资源、存储资源和网络资源进行动态重新配置。所提出的方法利用新兴的硬件和软件标准来提高处理器和设备的功率效率，如英特尔的quickstart和SpeedStep处理器，磁盘旋转，功率感知页面分配和RDRAM，提供更广泛的低功耗状态并降低转换成本。该项目的影响是显著降低了占美国电力消耗8%的互联网服务和Web服务器的电力消耗。此外，项目成果对环境产生了深远的影响，因为通过减少对能源的需求，发电机每年产生的二氧化碳量大大减少。

The objective of this project is to develop a theoretic framework and a hierarchical methodology for autonomic power and performance management of distributed computing centers through (a) online modeling, monitoring, and analysis of power consumption and performance; (b) adaptive learning and automatically identifying strategies to minimize power consumptions while maintaining the required quality of service requirements for a wide range of workloads and applications; and (c) dynamically reconfigure the computing, storage and network resources according to the selected optimization strategies. The proposed methodology exploits the emerging hardware and software standards to improve power efficiency of processors and devices such as such as Intels quickstart and SpeedStep processors, disk spindown, power aware page allocation and RDRAM that offer a wider range of low-power states and reducing the cost of transitions. The impact of this project is in the significant reduction of power consumptions of Internet services and Web servers that account for 8% of the US electricity consumption. In addition, the project results have a profound impact on the environment because by reducing the demand for energy, the amount of CO2 produced each year by electricity generators is reduced significantly.