Improving root cause failure analysis in virtual networks via the interconnected-asset ontology

Abstract

Identifying and locating root cause failures in virtual networks via active probing is a complex and repetitive task. Automation of this procedure requires a machine-processable representation that is not only capable of expressing the composition of physical network topologies but also the virtual networks supported by them. In this paper, we present an extension of the Interconnected-asset Ontology that allows to represent nested virtual networks. This extended representation is used to automatically generate sets of active probes that analyze the state of network paths. A proof-of-concept implementation is applied in the production network of a wide area network service provider. The corresponding evaluation shows that our approach is viable and can streamline the process significantly.