MRI: Track 2 Development of a Platform for Accessible Data-Intensive Science and Engineering

MRI：可访问数据密集型科学与工程平台的轨道 2 开发

• 批准号: 2320600
• 负责人: Joshua Agar
• 金额: $ 399.76万
• 依托单位: Drexel University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-09-15 至 2026-08-31
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2320600&HistoricalAwards=false
• 关键词: MRI; Track; Development; Platform; Accessible

Low-cost, high-performance sensors are producing an explosion of data in every field, from the mysteries of the cosmos to the tiniest particles. Despite this abundance of data, the absence of robust infrastructure and tools impedes the ability to effectively analyze and utilize it. The Development of a Platform for Accessible Data-Intensive Science and Engineering (DISE) will be a system for data management, sharing, and analysis that allows automatic data collection and curation, instant access for computational analysis, and research result sharing for reproducibility. The system will pave the way for new research in data-driven science – and scientific outcomes. A key issue that DISE addresses is the complexity and cost of data movement from storage to analysis sites, especially in cloud-based scenarios. DISE aims for a more cohesive system integrating data storage and analysis. By making research findings more accessible to the public, DISE will provide increased return on research investments to the research community and society at large. Critically, DISE will serve to train the next generation of scientists in data-intensive research, ensuring accessibility for diverse learners, and contributing to an equitable scientific community.DISE aims to advance data-intensive science and engineering and make research more accessible and reproducible, fostering equitable outcomes across all societal sectors. DISE will introduce an innovative platform that facilitates automatic data curation from scientific instruments, detailed metadata querying, intelligent tiering for instant data accessibility for computational workloads, interactive access to Graphics Processing Unit (GPU)-accelerated resources, hosting and sharing of containerized research products and deployment of self-healing micro-services for real-time data analysis. The platform will be equipped with a 2-petabyte scientific data management system, a 500-terabyte all-flash high-performance parallel file system, and GPU-accelerated computation. These features promise not only to streamline data management but also to facilitate input/output operations per second (IOPS) intensive workflows, including data mining, data-driven research, foundations in AI, data-intensive computation for good, and reliable and trustworthy AI. DISE is dedicated to aiding the exploration of new areas of study by utilizing community data repositories while simultaneously fostering the growth of the upcoming generation of scientists skilled in data-intensive research. DISE will also improve access to research data and computation, thus increasing the return on prior research investments. The automated curation feature of DISE is set to revolutionize how research data is organized and shared, making it accessible and interoperable beyond the originating researchers. Additionally, DISE is committed to providing high-performance GPU resources to foster data-intensive science. The platform will connect with researchers across various disciplines and intends to further expand its reach by automating data curation from shared user facilities. In addition, the use of GPU-accelerated computing resources via an easily accessible JupyterHub will aid in educating and training the next generation of researchers. Finally, DISE is committed to promoting diversity in the field by providing equitable access to resources and partnering with institutions such as Morgan State University to train a new generation of diverse data-intensive scientists and engineers.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.

低成本、高性能的传感器正在各个领域产生爆炸式的数据，从宇宙的奥秘到最小的粒子。尽管有如此丰富的数据，但缺乏强大的基础设施和工具阻碍了有效分析和利用数据的能力。可访问数据密集型科学与工程平台（DISE）的开发将是一个数据管理、共享和分析系统，允许自动数据收集和管理、即时访问计算分析和研究结果共享以实现再现性。该系统将为数据驱动的科学和科学成果的新研究铺平道路。DISE解决的一个关键问题是从存储到分析站点的数据移动的复杂性和成本，特别是在基于云的场景中。DISE的目标是集成数据存储和分析的更有凝聚力的系统。通过使公众更容易获得研究成果，DISE将为研究界和整个社会提供更高的研究投资回报。至关重要的是，DISE将有助于培训下一代科学家进行数据密集型研究，确保不同学习者的可及性，并为公平的科学界做出贡献。DISE旨在推进数据密集型科学和工程，使研究更容易获得和可复制，促进所有社会部门的公平成果。DISE将推出一个创新平台，促进科学仪器的自动数据管理、详细的元数据查询、用于计算工作负载的即时数据访问的智能分层、对图形处理单元（GPU）加速资源的交互式访问、托管和共享容器化研究产品以及部署用于实时数据分析的自我修复微服务。该平台将配备2 pb的科学数据管理系统、500 tb的全闪存高性能并行文件系统和gpu加速计算。这些功能不仅可以简化数据管理，还可以促进每秒输入/输出操作（IOPS）密集型工作流程，包括数据挖掘、数据驱动研究、人工智能基础、数据密集型计算以及可靠和值得信赖的人工智能。DISE致力于通过利用社区数据存储库来帮助探索新的研究领域，同时促进下一代数据密集型研究技术科学家的成长。DISE还将改善对研究数据和计算的获取，从而增加先前研究投资的回报。DISE的自动管理功能将彻底改变研究数据的组织和共享方式，使其能够在原始研究人员之外进行访问和互操作。此外，DISE致力于提供高性能GPU资源，以促进数据密集型科学。该平台将与不同学科的研究人员建立联系，并打算通过共享用户设施的自动化数据管理来进一步扩大其覆盖范围。此外，通过易于访问的JupyterHub使用gpu加速计算资源将有助于教育和培训下一代研究人员。最后，DISE致力于促进该领域的多样性，提供公平获取资源的机会，并与摩根州立大学等机构合作，培养新一代多样化的数据密集型科学家和工程师。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。