II NEW: Infrastructure to Support Research in Network-Aware Data-Intensive Computing

II 新：支持网络感知数据密集型计算研究的基础设施

• 批准号: 1405767
• 负责人: Amy Apon
• 金额: $ 66.84万
• 依托单位: Clemson University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2014
• 资助国家: 美国
• 起止时间: 2014-09-01 至 2021-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1405767&HistoricalAwards=false
• 关键词: II; NEW; Infrastructure; Support; Research

The rise of data centers, large spaces filled with interconnected computers, have allowed for the rapid, flexible and efficient provisioning of the Internet services on which society depends. However, data centers are power hungry. Collaborative experimental research at the intersection of sophisticated computing infrastructure and data-intensive applications shows promise in increasing the flexibility and power efficiency of data centers. The rapid emergence of new technologies such as Software Defined Networks are enablers of new strategies for placement of computing tasks within a cluster, taking into account factors such as load and power efficency. This computing research infrastructure project provisions computer and software-defined networking (SDN) equipment, metered rack power distribution units, software, and toolkits to establish a programmable, multi-purpose, network and power-aware cluster systems test bed at Clemson University. The infrastructure fills a gap of resources, and supports fundamental and applied research on experimental networking and scalable data-intensive computing system in areas of programmable networks to support data-intensive computing, application and data placement strategies for data-intensive applications, evaluation of software infrastructures to support data-intensive applications, and performance and power measurement, modeling, and analysis. The project provides advanced infrastructure to at least 300 graduate and undergraduate students at Clemson University. The system provides an environment for supporting projects in the highly successful Clemson Research Experiences for Undergraduates Site program in Data-Intensive Computing, which has recruited students from more than 20 states including Puerto Rico. The system also supports collaboration with industrial partners and is an enabling factor in a collaborative relationship between Clemson University and Lawrence Berkeley National Lab.

数据中心的兴起，充满互联计算机的大型空间，允许快速，灵活和高效地提供社会所依赖的互联网服务。 然而，数据中心非常耗电。 在复杂的计算基础设施和数据密集型应用的交叉点进行的合作实验研究显示出提高数据中心灵活性和能效的前景。 软件定义网络等新技术的快速出现，推动了在集群内放置计算任务的新策略，同时考虑了负载和电源效率等因素。 该计算研究基础设施项目提供计算机和软件定义网络（SDN）设备，计量机架配电单元，软件和工具包，以在克莱姆森大学建立可编程，多用途，网络和电源感知集群系统测试床。 该基础设施填补了资源的空白，并支持可编程网络领域的实验网络和可扩展数据密集型计算系统的基础和应用研究，以支持数据密集型计算，数据密集型应用的应用和数据放置策略，支持数据密集型应用的软件基础设施的评估，以及性能和功率测量，建模和分析。该项目为克莱姆森大学至少300名研究生和本科生提供先进的基础设施。 该系统提供了一个支持项目的环境，在非常成功的克莱姆森研究经验的本科生网站程序在数据密集型计算，其中招募了来自20多个州，包括波多黎各的学生。该系统还支持与工业合作伙伴的合作，是克莱姆森大学和劳伦斯伯克利国家实验室之间合作关系的一个有利因素。

项目成果

Measuring Network Latency Variation Impacts to High Performance Computing Application Performance

测量网络延迟变化对高性能计算应用程序性能的影响

• DOI: 10.1145/3184407.3184427
• 发表时间: 2018
• 期刊: ICPE '18 Proceedings of the 2018 ACM/SPEC International Conference on Performance Engineering
• 作者: Underwood, Robert;Anderson, Jason;Apon, Amy
• 通讯作者: Apon, Amy