Discovering and Securing Shared Resources on the Internet

Abstract

The Internet is a collection of shared resources. Internet users

share bandwidth and processing resources both in the network at

routers and on the network's edge at servers. However, the Internet's

architecture does not prevent nodes from consuming disproportionate

resources. In practice, resource exhaustion does occur due to

inefficiently scaling systems, selfish resource consumption, and

malicious attack. The current Internet architecture has limited

support for both securing and identifying shared Internet resources.

This dissertation has two main contributions. First, I demonstrate

the existence of end-host protocols that protect the availability of

shared Internet resources. I consider resource sharing with respect

to cooperative, selfish, and malicious user models, and for each case

design a protocol that protects resource availability without modifying

the existing Internet infrastructure. Second, I design and validate

measurement techniques for discovering shared Internet resources

including links and routers. Specifically, I improve the completeness

and accuracy of resource maps by combining embedding probes, disjunctive

logic programming, and information from the record route IP option.

We validate and quantify the improvement of our maps by comparison to

publicly available research networks