MRI: Acquisition of CASTOR: A High-Performance Communication and Storage Backbone for Data-Intensive Science and Engineering Computing

MRI：收购 CASTOR：用于数据密集型科学和工程计算的高性能通信和存储骨干

• 批准号: 0421200
• 负责人: Sanjay Ranka
• 金额: $ 60万
• 依托单位: University of Florida
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-09-01 至 2008-08-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0421200&HistoricalAwards=false
• 关键词: MRI; Acquisition; CASTOR; Performance; Communication

This project, adding high performance interconnections and true data-intensive storage facilities, involves a high performance communication and storage backbone, named CASTOR. With the interconnections, this backbone enables achievement of the full benefits of the grid strategy to support data-intensive research in diverse disciplines. Major campus facilities will be linked by a 10Gigabit/s network and supported by 24 Terabyte storage systems. CASTOR's high performance links are well matched to those of Florida Lambda Rail and National Lambda Rail. Serving the following research domains, High Energy Physics, Chemical Physics and Material Science, Coastal and Estuarine Modeling, Medical Physics, and Computational Biology, the new infrastructure will also serve Computer Science and Engineering on data-intensive, high-performance grids. CASTOR is expected to strengthen the collaboration between multidisciplinary teams across the institution and provide a unique resource for the projects in which the researchers participate. The infrastructure provides a unique resource for applications that require storing, processing, and communicating very large-scale data sets at very high speeds. Reflecting real-world scenarios, the work involves the following technologies:* Grid Computing,* Distributed Storage, and* Advanced Networking.The first involves the development innovative middleware techniques that complement and extend virtualization of physical information resources to hide infrastructure heterogeneity and allow applications to execute on grid-enabled resources with user control and interaction from various web-based interfaces. The second provides a trial version of the grid environment to operate queries, providing a proof-of-concept for deployment in larger systems. CASTOR serves as an "ideal sandbox" to develop and test algorithms suitable for a distributed data pipeline. The last enables experiments in a research network operating at ultra-high speeds and multiple domains, hence contributes to enrich projects.

该项目增加了高性能互连和真正的数据密集型存储设施，涉及一个名为CASTOR的高性能通信和存储骨干。通过互连，这个主干能够实现网格战略的全部好处，以支持不同学科的数据密集型研究。主要园区设施将由10千兆位/秒网络连接，并由24 TB存储系统支持。卡斯特的高性能链接很好地匹配那些佛罗里达拉姆达铁路和国家拉姆达铁路。服务于以下研究领域，高能物理，化学物理和材料科学，海岸和河口建模，医学物理和计算生物学，新的基础设施还将服务于数据密集型，高性能网格的计算机科学和工程。预计CASTOR将加强整个机构多学科团队之间的合作，并为研究人员参与的项目提供独特的资源。基础设施为需要以非常高的速度存储、处理和通信非常大规模数据集的应用程序提供了独特的资源。这项工作反映了现实世界的情况，涉及以下技术：* 网格计算，* 分布式存储，* 高级网络。第一个涉及开发创新的中间件技术，补充和扩展物理信息资源的虚拟化，以隐藏基础设施的异构性，并允许应用程序在网格资源上执行，用户可以通过各种基于Web的界面进行控制和交互。第二个版本提供了一个网格环境的试用版来操作查询，为在更大的系统中进行部署提供了概念验证。CASTOR作为一个“理想的沙箱”来开发和测试适合分布式数据管道的算法。最后一个可以在超高速和多域的研究网络中进行实验，从而有助于丰富项目。