ABSTRACT

Over recent years, there have been two significant changes in the area of high-end computing: 1) an increasing focus on interactive data driven applications and 2) a shift from tightly coupled high-end computing systems to networked clusters of machines within an organization and across geographically dispersed organizations. A particular challenge in using networked clusters for an interactive and data-intensive computing task is that processing of data may not always be possible where data is resident. The principal investigators (PIs) envision that a configuration comprising a networked collection of storage cluster(s), compute cluster(s), memory cluster(s), wired/wireless clients, and facilities for data visualization and collaboration, will be common and popular in modern organizations. Moreover, people within an organization will interact with existing compute and data resources over the Internet, as part of a grid environment.

Intellectual Merit: This project aims to design and develop such systems and their components including communication, networking, I/O, QoS, programming models, compilers, scheduling, middleware, caching, and indexing. The system research will be closely integrated with research in algorithms for visualization and data mining and end data-intensive applications. The overall goal will be to enable collaborative, data-driven, and interactive applications involving multi-terabyte datasets to benefit from end-to-end QoS.

The research testbed will consist of a compute cluster and a memory cluster connected to an existing mass storage cluster with high-speed WAN links (20.0 Gbps and beyond). A wireless testbed with clients and graphics resources (adapters, haptic devices, and headmount displays) will also be connected to the compute cluster to study the end-to-end QoS and interactivity issues.

Broader Impact: This project will benefit 18 investigators, their students (including members of underrepresented groups) and research staff in the Department of Computer Science and Engineering (CSE) and Department of Biomedical Informatics (BMI) to carry out state-of-the research in systems, networking, middleware, data mining and visualization; disseminate the results through research forums; and disseminate the developed software through open-source avenues. Several team members of this project collaborate with faculty members and researchers inside and outside of the campus in designing and developing end-applications in the areas of engineering, science, and medicine. This project will facilitate developing state-of-the-art solutions for these end-applications in an inter-disciplinary manner. A numbers of initiatives are planned for integrating research and education: enhancing several graduate courses in CSE with hands-on projects on networked cluster environment, creating new courses on data mining and data and scientific visualization in CSE, and creating a new interdisciplinary course sequence in the area of application-driven grid computing (jointly to be done by the CSE and BMI). The outreach plan includes participation in K-12 programs, participation in summer training programs for underrepresented groups, and collaboration with minority institutions.

Please report errors in award information by writing to: awardsearch@nsf.gov.