Panel on smart access to large scientific datasets

Abstract

The enormous technological enhancements in computer chips and processor architectures have provided orders of magnitude improvements to scientific computing in terms of processor(s) performance and network bandwidth. The ultimate result of the technical improvements is multitudes of massive and complex datasets. The analysis of these large scientific datasets is now a problem unto itself. Processor speeds and architectural designs have drastically improved the performance of computer hardware. One also needs to focus on functional improvements which optimize the performance of the application scientist working in a distributed computing environment. This panel proposes to discuss methods of providing smart access to scientific data. Among the areas to be discussed are: required infrastructure for scientific data structures and access; interactive access of scientific data for generic applications; widely distributed access of scientific data from data centers across the nation; the use of metadata as an intermediate level of information; smart access to data in mass storage.