Rethinking multi-layer multi-domain network monitoring

Abstract

Accurate and efficient network fault localisation based on monitoring is obviously one of the vital but also formidable tasks for successful network operations. With proliferations of large-scale network services, e.g. Géant E2E links, monitoring and fault localisation across multiple domains with data obtained from different network layers are becoming unprecedentedly important. Many efforts have been invested to tackle the challenges posed by fault localisation, nevertheless, most approaches only suggest partial solutions to the problematics of multi-domain multi-layer aspects. Most importantly, a comprehensive view and deep understanding of the problem is still missing. In this paper, we intend to systematically discuss challenges and their implications posed by fault localisation with consideration on multi-domain and multi-layer monitoring data. The contributions of this paper are manifold: first we identify key research challenges regarding multiple layer monitoring for fault localisation across domains; then based on the identified multiple layer network patterns, we establish comprehensive problem dimensions in a systematic and structured way which holds key to the solution development; finally we propose a formal definitions on information model which captures essential characteristics of multiple network layers across domains.