After the phenomenal Internet growth during the 1990s, tracking macroscopic Internet connectivity is a daunting task. CAIDA's Autonomous System (AS) core graph [4] and table [1] provide a ranking of AS outdegree based on massive topology data as measured by CAIDA's macroscopic IP topology monitoring project and observed by RouteViews BGP table snapshots [2,3]. The coverage of these topology probes is unprecedented, dramatically higher than any previous work in this area, and the data yield significant insight into the relative richness of IP connectivity of different ASes.
To extend and improve our methodology for deriving and depicting this connectivity ranking, we would like to extend the project in two ways:
We will provide weekly updates to this table as specifically requested by Cisco (Barry Greene) for support in their critical incident notification process.
We are also experimenting with techniques for animating the AS core graph; a preliminary prototype [4] shows promise to meet this challenge. This first animation displays eighteen months of AS core data from July 2000 to January 2002. This particular data set indicates that the ranking of major ISPs remained essentially unchanged over this period, but infrastructural shifts over the next decade are likely to be exposed using this technique.
We propose to identify and depict exchange and quantify their influence on AS connectivity rankings. Several colleagues in the operational community have expressed interest in having a version of the AS core graph that depicts exchange points in addition to ASes. Finding an appropriate visualization will require experimenting with several different presentation metaphors.
We will rely on CAIDA macroscopic topology data [5], BGP tables from RouteViews [2], and PCH's database of exchange point prefixes [6] to derive a map of ASes and exchange points that integrates topological and geographic information. A knowledge base of this kind will allow pursuit of answers to several infrastructurally relevant questions, e.g.:
Both tasks are aimed at improving our understanding of macroscopic view of Internet connectivity among ASes, and the role that different exchange points play in sustaining this core connectivity. The results will allow for improved incident notification process in the case of learned vulnerabilities (e.g., viruses, protocol weaknesses), as well as increased awareness of potential vulnerabilties deriving from excessive concentration of connectivity provided by a single or few exchange points.
Funding ($100,000) to begin 1 July 2002. (or whenever possible)
Research Milestones
Researchers are available to meet with Cisco staff to discuss methodologies and the implications of analysis results, in particular with respect to Cisco's incident notification process or router instrumentation that might improve the quality of data gathered.