Integrating Jupyter into Research Computing Ecosystems: Challenges and Successes in Architecting JupyterHub for Collaborative Research Computing Ecosystems
Integrating Jupyter into Research Computing Ecosystems: Challenges and Successes in Architecting JupyterHub for Collaborative Research Computing Ecosystems
复制标题
将 Jupyter 集成到研究计算生态系统中：为协作研究计算生态系统构建 JupyterHub 的挑战和成功
DOI：
10.1145/3311790.3396648
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
M. Dahan
中科院分区：
文献类型：
--
作者：
Joe Stubbs;Julia Looney;Marjo Poindexter;E. Chalhoub;Gregory J. Zynda;Erik S. Ferlanti;M. Vaughn;John M. Fonner;M. Dahan
从2015年底首次发布DesignSafe JupyterHub开始，TACC一直在为不同科学和工程领域的研究小组构建和维护定制的JupyterHub集群。今天，TACC维护着五个生产JupyterHub系统，使用超过半TB的内存和数百个CPU核心，总共支持近1,600个独立用户。在这篇文章中，我们描述了我们在这些网络基础设施项目中利用JupyterHub的方法，以及我们将Jupyter整合到不同研究社区的协作方法。对于两个这样的群体，我们将深入讨论影响和发展我们产品设计的科学用例和技术驱动因素，以及我们为促进采用而进行的外联和参与努力。我们讨论了我们为更好地将JupyterHub集成到网络基础设施平台的其他组件而添加的定制功能的实施，最后描述了我们对未来架构和JupyterHub在TACC的使用的计划。
Beginning with the initial release of the DesignSafe JupyterHub in late 2015, TACC has been building and maintaining custom JupyterHub clusters for research groups across different domains of science and engineering. Today, TACC maintains five production JupyterHub systems utilizing over half a terabyte of memory and hundreds of CPU cores supporting nearly 1,600 unique users combined. In this paper, we describe our approach to utilizing JupyterHub in these cyberinfrastructure projects and our collaborative approach to integrating Jupyter into different research communities. For two such groups, we present an in-depth discussion of the science use cases and technical drivers that informed and evolved the design of our offering, and our outreach and engagement efforts to promote adoption. We discuss the implementation of custom features we have added to better integrate JupyterHub into other components of the cyberinfrastructure platforms, and we conclude with a description of our plans for future architecture and usage of JupyterHub at TACC.