MRI: Acquisition of Tera-Scale Data Analytic Platforms for Research in the Combinatorial and Graph Sciences Consortium

提案号：CNS 07-23060PI(s): Gloster, Jr, Clay Boyack, Kevin w；乔治·s·戴维森；拉马纳坦,Murali;机构：霍华德大学华盛顿特区20059-0001职称：MRI/Acq。该项目旨在获取用于组合科学和图科学研究的分析平台，解决一个关键问题，即从大量数据集中提取信息。所要求的仪器是一个大型生产系统（Netezza性能服务器），将位于霍华德和桑迪亚国家实验室，为DNA结合机制、蛋白质结构预测、桑迪亚计算科学的数据挖掘、纽约州立大学布法罗分校的多维可视化和新密歇根州立大学的自然语言处理等科学项目提供服务。这项工作回应了目前淹没科学界的数据洪流/海啸/爆炸/雪崩。数据密集型计算机资源，国家太字节数据分析中心（NTDAC），提供了分析太字节数据集的能力，比目前的替代方案快10到100倍，成本更低。随着时间的推移，NTDAC有望容纳和分析千兆字节的原始数据，具有备份和支持能力，并使用并行关系数据仓库设备技术，该技术速度更快，更易于使用。最初的重点是利用大规模并行关系数据库设备技术，解决计算生物学、化学、物理、文本挖掘、书目耦合和自然语言处理方面的问题。更广泛的影响：基础设施通过更快的数据分析使所有科学领域受益。对海量数据集的分析使从业者和学生能够以更快的速度交付/获得新知识，并创建所需的解决方案。此外，该机构是一所HBCU，少数族裔学生可能会从这种经历中获益。

Proposal #: CNS 07-23060PI(s): Gloster, Jr., Clay Boyack, Kevin W.; Davidson, George S.; Ramanathan, Murali; Zacharski, Ron Institution: Howard University Washington, DC 20059-0001Title: MRI/Acq.: Terascale Data Analytic Platforms for Research in the Combinatorial and Graph Sciences ConsortiumProject Proposed:This project, acquiring analytic platforms for research in combinatorial and graph sciences, attacks a critical problem, that of extracting information from massive sets of data. The requested instrumentation, a large system for production (Netezza Performance Server(s)), are to be housed at Howard and Sandia National Lab and serve scientific projects on DNA binding mechanisms, protein structure prediction, data mining for computational science at Sandia, multi-dimensional visualization at SUNY-Buffalo, and natural language processing at NMSU. The work responds to the data deluge/tsunami/explosion/avalanchethat presently overwhelms the scientific community. The data intensive computer resource, National Terabyte Data Analysis Centers (NTDAC), provides the ability to analyze tera-scale data sets 10 to 100 times faster than current alternatives at a lower cost. Over time, NTDAC is expected to house and analyze peta-bytes of raw data, to have back-up and support capabilities, and to use parallel relational data warehouse appliance technology that is orders of magnitude faster and easier to use. The initial focus, exploitation of massively parallel relational database appliance technology, enables solution problems in computational biology, chemistry, physics, text mining, bibliographic coupling, and natural language processing.Broader Impact: The infrastructure benefits all areas of science through faster data analysis. The interrogation of massive data sets allows practitioners and students to deliver/gain new knowledge at a more rapid pace, and to create needed solutions. Moreover, the institution is an HBCU, where minority students are likely to gain from the experience.