Internet topology modeling

In this lecture, Walter Willinger goes into the difficulties of Internet topology modeling (on the IP level).

Accurate models of how the Internet works give insight into where its weakest links are located and how to simulate the network’s behavior under certain circumstances.

Part of the lecture’s data was acquired using traceroute and thorny technical details like detecting aliased IPs and load balancers were skipped.

Some highlights of the talk are:

1. Usage of economic models and current technological constraints to model the network (the resulting topology should be efficient).
2. Pitfalls of paying too much attention to power laws.
3. Differences between the real Internet and scale-free networks.

I believe some of the points the author makes could be applied as well to overlay (P2P) networks.

Further links:

• A First-Principles Approach to Understanding the Internet’s Router-Level Topology
• Towards a Theory of Scale-Free Graphs: Definition, Properties, and Implications (Extended Version)