Journal of Network and Systems Management, Vol. 9, No. 3, September 2001 ( 2001) 1064-7570 / 01 / 0900-0327$19.50 / 0  2001 Plenum Publishing Corporation 327 Using Network Fault Predictions to Enable IP Traffic Management 1 M. Thottan 2,4 and C. Ji 3 IP traffic management is important for the continued growth of the Internet. Several traffic management algorithms exist today. However, to enable these algorithms it is necessary to provide reliable alarms relating to network performance bottlenecks and failures. In this work we propose an algorithm to obtain reliable predictive alarms for network fault conditions. The algorithm is based on modeling network fault behavior. The algorithm has been successfully tested on two production networks. Predictive alarms were obtained for four different types of failures: file server failures, network access problems, protocol implementation errors, and runaway processes. The potential of using this model to do fault classification is also discussed. In addition, it is shown that the proposed algorithm performs better than the majority-vote scheme. KEY WORDS: Network health; change detection; fault model; spatial correlator. 1. INTRODUCTION Efficient management of network traffic is important for the continued growth of the Internet. Research in the area of IP traffic management has primarily focused on developing management algorithms for specific tasks such as admission con- trol [1], policing and shaping of network traffic [20]. These algorithms can be deployed as soon as an alarm is obtained from the system notifying the existence of a performance bottleneck or a network fault. However, the problem of gener- ating these alarms has received little attention. This paper addresses the issue of alarm generation by providing a method of obtaining an online notification of the health of the network. The health indicator is obtained by modeling network fault behavior. 1 Supported by DARPA under contract number F30602-97-C-0274. 2 Lucent Technologies, Bell Laboratories, Department of Network and Service Management, Holmdel, New Jersey 07733. http: // www.bell-labs.com / user / marinat 3 Department of Electrical, Computer, and Systems Engineering, Rensselaer Polytechnic Institute, Troy, New York 12180. E-mail: marinat@lucent.com, chuanyi@ecse.rpi.edu 4 To whom correspondence should be addressed.