Second Workshop on Evaluating and Architecting System Dependability (EASY). In conjunction with ASPLOS-X. Sunday, 6 October 2002, San Jose, California, U.S.A.
Michael W. Bigrigg
Institute for Complex Engineered Systems
Carnegie Mellon University
Pittsburgh, PA 15213
File systems make unreasonable attempts to provide data to the point that they will block an application instead of passing the error on to the application to handle. Transient problems such as network congestion or outages and heavily loaded systems or denial of service attacks can lead to failure-like situations. Alternative mechanisms have been developed for the file system to trade performance for robustness in an attempt to always guarantee full availability of data. These mechanisms may not be necessary, as the application programmer may have already accounted for such situations. By hinting to the file system the application’s ability to handle errors it is possible for the file system to make better resource allocation decisions and improve end-to-end dependability.
KEYWORDS: Reliability, Filesystems, Program Analysis.
FULL PAPER: pdf