Difficulties Measuring the Internet's AS-Level Ecosystem

Publisher Website
Google Scholar
Abstract
Compared to other large-scale, complex systems, one of the most appealing features of the Internet is that a thorough understanding of its component technologies combined with a unique ability to measure the network means that most of the models and theories proposed for explaining the Internet's observed structure and behavior can be in general unambiguously validated, though perhaps not without substantial efforts. An example that requires such substantial effort is the structure and dynamics of the Internet at the level of autonomous systems (ASs). We show in this paper that the problem of inferring connectivity within the Internet's AS-level ecosystem has been severely hampered by highly ambiguous measurements, with serious but often ignored implications for AS-level topology modeling and model validation. The situation is even worse with respect to inferring inter-AS traffic demands where the basic problem has been a complete lack of any sort of useful and publicly available measurements. We describe recent attempts that try to change this situation and discuss their implications for AS-level traffic matrix modeling and model validation.

This publication has 12 references indexed in Scilit: