CAREER: Timed Streams: Infrastructure, Applications, and Algorithms

North Carolina State University, Raleigh NC

Investigators

Abstract

One of the primary, and often the most important, goals of any computer system is good performance. So it is for communication networks-throughput is the primary evaluator and impetus for new network research and development. In this proposal, we identify a new paradigm for thinking about distributed algorithms that exploits a shift in the engineering environment, the availability of high-precision globally synchronous clocks with cheap hardware. Many research projects are investigating the limits of precision and uses of precise time for traditional, e.g., operating system, algorithms. We propose to build on this research, focusing on improved performance of application-level algorithms and targeting distributed, real-time databases support for e-commerce applications. The career development plan outlined in this proposal focuses on the principal investigator's efforts to integrate core research and educational objectives. The research project addresses a number of important issues in the discovery and analysis of precise time-based algorithms for distributed, real-time systems. The ultimate goal of the proposed work is and estimate of the benefit of precise globally synchronous time for a wide variety of applications. The objective is to not only develop and analyze new time based network protocols, but to characterize the impact of globally synchronous time on distributed algorithms in general. to achieve this, we introduce the concept of a timed stream and propose to investigate how to best implement and use timed streams to achieve high performance distributed, real-time applications. The educational component is closely related to the research plan. Students at the undergraduate and graduate level, as well as industrial participants, will use the software produced and data gathered during the research activities detailed in this proposal. The tools (e.g., new Web server technology) will serve as a research engine to collect performance data for the purpose of performance-based analysis of distributed algorithms. The education plan also includes development of e-commerce curriculum and courses, integrating the proposed research at all levels, and using the basic scientific principles of the research, as well as the PI's enthusiasm, to encourage young (K-12) scientists.