CC* Compute: Private Campus Cloud for Data Analytics and Machine Learning

CC* 计算：用于数据分析和机器学习的私有园区云

• 批准号: 2018926
• 负责人: Preston Smith
• 金额: $ 39.22万
• 依托单位: Purdue University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-07-01 至 2023-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2018926&HistoricalAwards=false
• 关键词: CC; Compute; Private; Campus; Cloud

New usage patterns of computing for research have emerged that rely on the availability of flexible, elastic, and highly specialized services. Uniform batch computing pools traditionally provided by high performance or high throughput computing environments have difficulty adapting to meet these requirements. A new approach that updates and evolves the research computing ecosystem is needed to respond to these needs. This new model, a “Community Cloud”, provides a cost effective, highly responsive, sustainable, and customizable cloud and container computing solutions for specific applications and domain science communities.This project, through the acquisition of a new compute cluster, knits together central and lab-scale data, instrument, and compute resources into a cloud ecosystem for researchers who need capabilities beyond batch computing, and extends the research computing ecosystem to include cloud capabilities at the campus level. The Community Cloud is designed to: 1) Devise a new approach to establish a community cloud service using virtualization, containers, and infrastructure-as-code (IAC) techniques to create running infrastructure as an artifact; 2) Support diverse science domains via an effective infrastructure that enables new kinds of discovery that cannot be well met through the use of traditional batch computing systems; 3) Develop a reference business model for the evolution of campus “condo” cluster programs to sustainably operate a production community cloud; and 4) Enable scalable and sustainable instructional use of the proposed community cloud for courses, real-world training, and workforce development for the campus and national research computing communities. The new compute cluster includes 8 application nodes (1024 cores), 2 GPU nodes (8 gpus), 6 storage nodes (288 TB), and one bastion node, all interconnected through a 100Gb network, and managed using Kubernetes.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

研究计算的新使用模式已经出现，它依赖于灵活、弹性和高度专业化服务的可用性。传统上由高性能或高吞吐量计算环境提供的统一批处理计算池难以适应这些需求。需要一种更新和发展研究计算生态系统的新方法来响应这些需求。这种新的模式，即“社区云”，为特定的应用程序和领域科学社区提供了一种具有成本效益、高响应性、可持续性和可定制的云和容器计算解决方案。该项目通过收购一个新的计算集群，将中央和实验室规模的数据、仪器和计算资源编织成一个云生态系统，供需要批量计算以外功能的研究人员使用，并将研究计算生态系统扩展到包括校园级的云功能。社区云的目的是：1)设计一种新的方法来建立社区云服务，使用虚拟化、容器和基础设施即代码（IAC）技术来创建运行的基础设施作为工件；2)通过有效的基础设施支持不同的科学领域，使传统的批处理计算系统无法很好地满足新的发现；3)为校园“共管”集群项目的演变制定参考商业模式，以可持续运营生产社区云；4)为校园和国家研究计算社区的课程、实际培训和劳动力发展提供可扩展和可持续的社区云教学使用。新的计算集群包括8个应用节点（1024核）、2个GPU节点（8 GPU）、6个存储节点（288 TB）和1个堡垒节点，全部通过100Gb网络互联，使用Kubernetes进行管理。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Simplifying Scientific Application Access in Kubernetes with Push Button Deployments

通过按钮部署简化 Kubernetes 中的科学应用程序访问

• DOI: 10.1145/3491418.3535164
• 发表时间: 2022
• 期刊: PEARC '22: Practice and Experience in Advanced Research Computing
• 作者: Johnston, Taylor;Shaw, Victor;Werts, Brian;Weekly, Samuel;Gough, Erik;Smith, Preston
• 通讯作者: Smith, Preston

The “Geddes” Composable Platform - An Evolution of Community Clusters for a Composable World

“Geddes” 可组合平台 - 可组合世界的社区集群的演变

• DOI: 10.1109/supercompcloud51944.2020.00011
• 发表时间: 2020
• 期刊: Supercompcloud workshop 2020
• 作者: Smith, Preston M;Gough, Erik;Younts, Alexander;Werts, Brian;Hacker, Thomas J;Neumeister, Norbert;Wisecaver, Jennifer
• 通讯作者: Wisecaver, Jennifer