Statistical-based anomaly detection for NFV services

Large-scale, carrier-grade Network Functions Virtualisation (NFV) services are expected to involve a significant number of Virtual Network Functions, deployed across multiple Points-of-Presence (PoPs) and possibly in heterogeneous infrastructures. While proper monitoring is crucial for the commercial viability of NFV services, effectively and efficiently monitoring a huge number of VNF instances, promptly detecting any malfunctions or anomalies in order to trigger corrective actions, becomes a real challenge. This paper presents the use of an open-source monitoring system especially tailored for NFV in conjunction with statistical approaches commonly used for anomaly detection, towards the timely detection of anomalies in deployed NFV services.