AutoFocus: Automatically scoping the impact of anomalous service events

Abstract

Networks, and the services they enable, are increasingly diverse and highly utilized. From DSL and fiber-to-the-home access networks, to cellular mobile networks, to contentdelivery networks; all require extensive monitoring in order to meet the increase of user expectations of the availability and quality of those services provided to them. The complexity of these networks and services require better management on the part of providers as the data resulting from service monitoring experiences an increase in dimensionality, making it difficult to fully interpret anomalies in the data. For example, anomaly detection generally says “I found an anomaly with mobile phone A in market Z”. But it is more useful to know what other phones and what other markets are also experiencing the same anomaly.