An Analysis of The Internal Structure of Large Autonomous Systems

This study examines the topological and geographical structure of eight large Autonomous System (AS) topologies. We use a large collection of traceroutes obtained from multiple vantage points, and a set of heuristics designed to infer, from those traceroutes, the topology of individual ASs. We carefully validate some aspects of these topologies using publicly available information. Our analyses uncover several interesting common features of these topologies. We find high variability both in the degree distribution of individual ASs and in the size of AS points-of-presence. These ASs seem to be engineered to match small-world properties: good overall path lengths with significant local clustering. There is an interesting locality in the geographic extent of ASs; there exists a small number of metropolitan areas in which all ASs have points of presence. Finally, the density of city-level connectivity of these ASs suggests some differences in the way these ASs are engineered. Keywords— Network Topology Measurement, Network Topology Analysis, AS Internal Structure