Summary: Implementing Cooperative Prefetching and Caching
in a Globally­Managed Memory System
Geoffrey M. Voelker, Eric J. Anderson, Tracy Kimbrel \Lambda ,
Michael J. Feeley y , Jeffrey S. Chase z , Anna R. Karlin, and Henry M. Levy
Department of Computer Science and Engineering
University of Washington
Abstract
This paper presents cooperative prefetching and caching --- the use
of network­wide global resources (memories, CPUs, and disks) to
support prefetching and caching in the presence of hints of fu­
ture demands. Cooperative prefetching and caching effectively
unites disk­latency reduction techniques from three lines of re­
search: prefetching algorithms, cluster­wide memory management,
and parallel I/O. When used together, these techniques greatly in­
crease the power of prefetching relative to a conventional (non­
global­memory) system. We have designed and implemented
PGMS, a cooperative prefetching and caching system, under the
Digital Unix operating system running on a 1.28 Gb/sec Myrinet­
connected cluster of DEC Alpha workstations. Our measurements
and analysis show that by using available global resources, coop­

Source: Anderson, Richard - Department of Computer Science and Engineering, University of Washington at Seattle
Chase, Jeffrey S. - Department of Computer Science, Duke University
Isaila, Florin - Departamento de Informatica, Universidad Carlos III de Madrid

Collections: Computer Technologies and Information Sciences