The MBFS Project

The goal of the MBFS project is to design high-performance file systems via the efficient use of memory. File systems have historically suffered from high access latencies that, in recent years, have become even worse relative to processor speeds. Emerging technologies such as wireless computing, telecommuting environments, and even CD-ROMs exhibit higher latencies than conventional file systems.

The MBFS project tries to make effective use of local and remote memory resources to reduce both read and write latency. MBFS supports a prefetching system that attempts to hide or mask file system latency by automatically moving data into the system at a rate equal but not exceeding the processors demand for data. MBFS also supports a global memory architecture that provides cost-effective reliable memory-based storage that significantly reduces write latency.

Simulation models of both the prefetching and memory based storage system have been implemented and tested on real file traces. In addition, a SunOS 4.1 implementation exists and has been used to evaluate performance in a real system. A Solaris 2.x port is underway.

We gathered our own detail file system traces for use in our simulations. Additional information regarding our trace data can be found by clicking here .

J. Griffioen and R. Appleton , The Design, Implementation, and Evaluation of a Predictive Caching File System , Department of Computer Science Technical Report CS264-96, June, 1996, ( postscript ), ( gzipped postscript )
J. Griffioen , R. Vingralek , T. Anderson , and Y. Breitbart , Derby: A Memory Management System for Distributed Main Memory Databases , In the Proceedings of the 6th International Workshop on Research Issues in Data Engineering (RIDE '96) February, 1996, ( html ), ( postscript ), ( gzipped postscript )
J. Griffioen , T. Anderson , and Y. Breitbart , A Dynamic Migration Algorithm for a Distributed Memory-Based File Management System , In the Proceedings of the 7th International Workshop on Research Issues in Data Engineering (RIDE '97) February, 1997, ( html ), ( postscript ), ( gzipped postscript )
J. Griffioen and R. Appleton , Performance Measurements of Automatic Prefetching, In Proceedings of the ISCA International Conference on Parallel and Distributed Computing Systems, September 1995, ( postscript ), ( gzipped postscript ), A longer more detailed version of this paper appeared as TR CS256-95 (see below)
J. Griffioen and R. Appleton , Performance Measurements of Automatic Prefetching, Department of Computer Science Technical Report CS256-95, May, 1995, ( postscript ), ( gzipped postscript )
J. Griffioen and R. Appleton , Reducing File System Latency using a Predictive Approach , In Proceedings of the 1994 Summer USENIX Conference , USENIX Assoc. , June, 1994, pp 8-12, also released as TR CS247-94, ( postscript ), ( gzipped postscript )
J. Griffioen and R. Appleton , Automatic Prefetching in a WAN, In Proceedings of the IEEE Workshop on Advances in Parallel and Distributed Systems October, 1993, pp 8-12, also released as TR CS243-93, ( postscript ), ( gzipped postscript) )

Page last updated by Todd Anderson ( anderson@dcs.uky.edu ) on 9/12/1997