Abstract
As the Internet grows larger and more complex, it becomes exceedingly difficult to reason about its structure because there exists no scalable method to map the Internet's crucial structural properties such as connectivity, delay and bandwidth. The lack of such structural information in efficiently accessible form leads to poorly functioning and excessively complex network software and protocols. In this talk, I will discuss early work in developing geometric models for the Internet's structural properties. Geometric models are inherently simple and scalable, and thus can enable a wide variety of scalable performance-aware protocols and applications.