Date of Original Version
Abstract or Description
Automatic identification of anomalies on network data is a problem of fundamental interest to ISPs to diagnose incipient problems in their networks. ISPs gather diverse data sources from the network for monitoring, diagnostics or provisioning tasks. Finding anomalies in this data is a huge challenge due to the volume of the data collected, the number and diversity of data sources and the diversity of anomalies to be detected.
In this paper we introduce a framework for anomaly detection that allows the construction of a black box anomaly detector. This anomaly detector can be used for automatically finding anomalies with minimal human intervention. Our framework also allows us to deal with the different types of data sources collected from the network. We have developed a prototype of this framework, TrafficComber, and we are in the process of evaluating it using the data in the warehouse of a tier-1 ISP.