MRI: Development of a Fully Instrumented Self-Sensing and Self-Regulating Data Center

MRI：开发全仪器化自感知和自调节数据中心

• 批准号: 1040666
• 负责人: Kanad Ghose
• 金额: $ 134.23万
• 依托单位: SUNY at Binghamton
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2010
• 资助国家: 美国
• 起止时间: 2010-10-01 至 2015-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1040666&HistoricalAwards=false
• 关键词: MRI; Development; Fully; Instrumented; Self

Proposal #: 10-40666PI(s): Ghose, Kanad; Gopala, Kardik; Murray, Brice T; Sammakia, Bahgat G; Institution: SUNY - Binghamton Title: MRI/Dev.: Fully Instrumented Self-Sensing and Self-Regulating Data CenterProject Proposed:The project, building an instrument for next generation data centers where energy consumption and management is treated as a first class resource (together with CPU, communications, etc) that needs to be scheduled and managed explicitly to maximize the overall energy-efficiency of the data center, takes a new approach for predicting local and global thermal conditions with new control paradigms. Studying the most energy-efficient use of the cooling resources, the proposed instrument specifically involves:- Experimental, scaled down data center of Linux servers with dynamic cooling facilities, modified kernels. and scheduling components; - Large number of temperature and airflow sensors and power meters, along with software instrumentation; - Floor plenum based chilled air cooling system used to provide nominal cooling and remotely controlled computer room air conditioners (CRACs) that provide a quickly adjustable and directed cooling facility;- Software components for the facility that permit research into the development of a wide variety of techniques, both traditional and innovative, for improving the data center energy efficiency; and - Generation and recording live data on workload levels, power consumption, temperature, and airflow distributions. Broader Impacts: This unique facility primarily provides solutions for energy efficiency in data centers. The project intensively involves minority, women, and undergraduate students in their research. The instrument contributes to prepare all students for rewarding careers in the technology, management, and use of data centers.

提案编号：10- 40666 PI：Ghose，Kanad; Gopala，Kardik; Murray，Brice T; Sammakia，Bahgat G;机构：SUNY - Binghamton标题：MRI/开发：完全仪表化的自感知和自调节数据中心项目建议：该项目为下一代数据中心构建一种仪表，其中能源消耗和管理被视为第一类资源（与CPU，通信等一起），需要明确地进行调度和管理，以最大限度地提高数据中心的整体能源效率，采用新的控制模式预测局部和全局热条件的新方法。研究最节能的冷却资源的使用，建议的仪器具体涉及：-实验，缩小数据中心的Linux服务器与动态冷却设施，修改内核。和调度组件;- 大量的温度和气流传感器和功率计，沿着有软件仪表; -用于提供标称冷却的基于地板通风系统的冷冻空气冷却系统和提供快速可调节和定向冷却设施的远程控制的计算机房空调（CRAC）;- 该设施的软件组件，可用于研究开发各种传统和创新技术，用于提高数据中心的能源效率;以及-生成和记录有关工作负载水平、功耗、温度和气流分布的实时数据。更广泛的影响：这个独特的设施主要提供数据中心能源效率的解决方案。该项目集中涉及少数民族，妇女和本科生在他们的研究。该仪器有助于所有学生在技术，管理和使用数据中心的奖励职业做好准备。