The human factor: A challenge for network reliability design

Abstract

Computer and communication networks form part of the critical infrastructure of planetary society, and much work has gone into making the technology for such networks reliable. However, such networks have to be administered and managed by human administrators. The process of such administration, as it becomes increasingly complex, itself poses a challenge to protocols and systems designed to enhance network reliability. Several studies of highly reliable systems have shown that human operator error can account for 20-70% of system failures, and as the system become more reliable, the human factor gains increasing significance. Nevertheless, efforts to design reliability measures have remained largely disjoint from considerations of the human process of network administration. In this position paper, we describe some preliminary effort to examine the question of whether such joint consideration can be useful. We employ the first stages of a Grounded Theory approach to study network administration and management workflows in order to understand their impact in network reliability. Based on our study, we advance the proposition that the future of network reliability engineering must actively address the human process of system administration and management and concommittant misconfigurations.