SUPERB-IT

Summer Undergraduate Program in Engineering Research at Berkeley - Information Technology

June 17 - August 10, 2002

What Happened to My Dial Tone?
A study of FCC service disruption reports

Patricia Enriquez
Mills College
Oakland, CA

Abstract

As a first step in trying to increase the dependability of computer systems, we carry out an analysis of the failure modes of the Public Switched Telephone Network (PSTN), a highly dependable system that shares many similarities with traditional computer systems. In order to identify the primary failure modes of the PSTN, we analyzed six months of PSTN outage reports collected by the FCC. We identified three metrics for characterizing the severity of the outages and use them to evaluate the impact of seven factors on the availability of the PSTN. Our analysis demonstrates that human error is the primary cause of outages, regardless of the metric used. This conclusion has significant implications for the design of dependable computer systems, as human error has traditionally been ignored in fault-tolerant computer design.