Year of Publication

2011

Paper Type

Master's Thesis

College

College of Computing, Engineering & Construction

Degree Name

Master of Science in Computer and Information Sciences (MS)

Department

Computing

First Advisor

Sherif Elfayoumy, Ph.D.

Second Advisor

Zornitza Prodanoff, Ph.D.

Third Advisor

Sanjay Ahuja, Ph.D.

Abstract

This master’s thesis involves the definition and development of a policy-based routing scheme for peer-to-peer overlay networks. Many peer-to-peer networks are in existence today and each has various methods for discovering new peers, searching for content, and overcoming connectivity problems. The addition of efficient policy-based routing enhances the ability of peers within overlay networks to make appropriate routing decisions. Policy-based routing provides a means for peers to define the types of network traffic they are willing to route and the conditions under which they will route it. The motivations for these policies are many and are described in upcoming sections. In order to express and enforce policies, a simple policy definition language was developed. This language is sufficient for owners of overlay nodes to choose to route traffic based on their own requirements and gives node owners a means to express these requirements, such that other nodes within the overlay network can learn them. A mechanism is presented that allows these policies to be stored either in a distributed hash table or on a set of directory servers. The effectiveness of policy-based routing was tested using a simulated network. The affect of these routing policies, in terms of both additional network traffic and requirements for client software, was also assessed. Finally, a comparison was made between storing policy information in a distributed hash table, versus on a set of directory servers.

Share

COinS