Abstract

Modern day computer networks consist of many routers of varying configurations. The complexity of routers begets difficulty in correctly configuring networks. In fact, misconfigurations are so rampant in today’s networks that more than 75% of inter-network routing traffic is the result of misconfigurations, and the result of a single routing fault can cause connectivity issues for up to ten minutes (Mahajan, 2003). Much research is devoted to analyzing the nature of these misconfigurations and attempting to reduce their frequency. As mentioned by Maltz et al. in” Routing Design in Operational Networks”, white-box approaches that derive their knowledge of network topology and configuration from information available to network administrators yield greater insight than black-box approaches that attempt to derive similar knowledge from empirical data obtained without any special privileges.
One challenge facing …

Date

December 13, 2004

Authors

Jay Pujara, Xuejing Chen