University of California, Santa Cruz
Dr. Darrell Long
$18,220 funded 4/98
To provide support for an undergraduate student for one year and a
PC, to improve system performance in particular latency and prefetching.
Improving File System Cache Performance Through Reference Pattern Modeling Abstract
Through simulations it can be seen how the
actions of file systems are related and how these relationships can be
exploited to improve the performance of file system cache. Past
research found several issues, model size, adapting to local activity
and infinite stream length, that needed to be addressed in order for a
reference stream model to be practical and effective in tracking the
events of a file system. In this
project we will implement predictive caching models within a PC UNIX
to provide complete validation of this work.