Collaborative Research: CSR--AES: Energy Management in Real-time, Multi-tier Internet Services

合作研究：CSR--AES：实时、多层互联网服务中的能源管理

• 批准号: 0615277
• 负责人: Kevin Skadron
• 金额: $ 20.5万
• 依托单位: University of Virginia Main Campus
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2006
• 资助国家: 美国
• 起止时间: 2006-06-15 至 2010-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0615277&HistoricalAwards=false
• 关键词: Collaborative; Research; CSR; AES; Energy

Abstract:This project develops energy management techniques for large-scale,multi-tier data-centers, using formal optimization methods that do notcompromise performance requirements. High-performance Internetservices are often functionally decomposed into a pipeline ofmultiple, semi-independent tiers spanning multiple machines. These inturn might consist of a heterogeneous mix of system generations, dueto phased upgrades. Power delivery and cooling costs constitute asignificant part of total cost of ownership.The expected workload of such a system is aperiodic, with highlyvarying execution demand and possibly significant I/O activity. Thisrepresents an opportunity for major energy savings, but only ifperformance guarantees can be maintained. Scalability also requiresthat local power management decisions must yield globally optimalenergy consumption. Although effective power management policies havebeen devised for simpler architectures, no straightforward extensionsexist to make them applicable to multi-tier clusters while preservingresponse-time guarantees.Results will be disseminated at leading conferences and the softwarewill also be publicly available on the Internet for use by theresearch community in further research, or by the industry as a basisfor possible product development. Moreover, from a global societalperspective, successful energy management helps save naturalresources, while reduced service-provider costs foster economic growthand IT innovation. Finally, this research will help train graduateand undergraduate students.

翻译后摘要：该项目开发的大型，多层数据中心的能源管理技术，使用正式的优化方法，不妥协的性能要求。 高性能的互联网服务通常在功能上被分解为跨越多台机器的多个半独立层的管道。 由于分阶段的升级，这些系统可能又包括不同代的系统。 电力输送和冷却成本构成了总拥有成本的重要组成部分。这样一个系统的预期工作负载是不定期的，具有高度变化的执行需求和可能显著的I/O活动。 这是一个节省大量能源的机会，但只有在性能保证可以保持的情况下。 可扩展性还要求本地电源管理决策必须产生全局最佳能耗。 虽然有效的电源管理政策已被设计为简单的架构，没有直接的扩展存在，使他们适用于多层集群，同时牺牲响应时间的保证。结果将在领先的会议和软件也将公布在互联网上使用的研究社区在进一步的研究，或由业界作为一个基础，为可能的产品开发。 此外，从全球社会的角度来看，成功的能源管理有助于节约自然资源，而降低服务提供商成本则促进经济增长和IT创新。 最后，本研究将有助于培养研究生和本科生。