A Scalable Architecture for Monitoring and Visualizing Multicast Statistics

Prashant Rajvaidya and Kevin C. Almeroth
Department of Computer Science
University of California, Santa Barbara

k Claffy
Cooperative Association for Internet Data Analysis - CAIDA
San Diego Supercomputer Center,
University of California, San Diego

An understanding of certain network functions is critical for suc- cessful network management. Managers must have insight into net- work topology, protocol performance and fault detection/isolation. The ability to obtain such insight iseven more critical when try- ing to support evolving technologies like multicast. With these technologies, the pace of change is high, modifications to routing mechanisms are frequent, and faults are common. In this paper we introduce Mantra, a tool we have developed to monitor multicast. Mantra collects, analyzes, and visualizes network-layer (routing and topology) data about the global multicast infrastructure. The two most important functions of Mantra are: (1) monitoring multicast networks on a global scale; and (2) presenting results in the form of intuitive visualizations. To achieve accurate monitoring, Mantra collects data from several topologically and geographically diverse networks. For the purpose of presentation, Mantra uses several in- teractive visualization mechanisms to present statistics, topology maps and geographic properties. Another noteworthy feature of Mantra is its exible and scalable architecture. This architecture helps keep our monitoring erts current with the fast pace of mul- ticast developments. It also enables us to expand Mantra's moni- toring scope to more networks and larger data sets.