Event Correlation using Rule and Object Based Techniques

Abstract

Today's competitive market place has forced the telecommunications industry to improve their service and reliability. One step that telecommunications companies have taken to reduce network failures is the installation of operations centers to collect data from network elements. These centers are staffed by network managers who monitor network activity by correlating alarms across various operational disciplines (switch, facility, traffic) and relating them to a common cause. Accurate analysis is often difficult due to the volume of data and complexity of problems. ECXpert is a product developed recently at AT&T to help network managers monitor and analyze alarms, take corrective actions, and minimize disruptions to the network. Successful implementation of event correlation has increased customer revenue since trouble isolation can be done faster, resulting in quicker restoration of service. The essence ofECXpert is a high level language with which users can specify network events and their correlation with alarms. The system is written in Prolog and C++, a powerful combination which facilitated development to occur on time and in budget. It has been deployed in network management centers throughout the U.S. and is currently being marketed overseas. ECXpert is a success story for Prolog within AT&T.