Stochastic analysis of the interplay between object maintenance and churn
Computer Communications
Resilient and efficient load balancing in distributed hash tables
Journal of Network and Computer Applications
Aurelia: building locality-preserving overlay network over heterogeneous P2P environments
ISPA'05 Proceedings of the 2005 international conference on Parallel and Distributed Processing and Applications
Roogle: supporting efficient high-dimensional range queries in p2p systems
Euro-Par'06 Proceedings of the 12th international conference on Parallel Processing
| Hi-index | 0.00 |
In recent years, the Internet has experienced an enormous increase in the use of specialized content delivery systems, such as peer-to-peer file-sharing systems (e.g., Kazaa, Gnutella, or Napster) and content delivery networks (e.g., Akamai). The sudden popularity of these systems has resulted in a flurry of research activity into novel peer-to-peer system designs. Because these systems: (1) are fully distributed, without any infrastructure that can be directly measured, (2) have novel distributed designs requiring new crawling techniques, and (3) use proprietary protocols, surprisingly little is known about the performance, behavior, and workloads of such systems in practice. Accordingly, much of the research into peer-to-peer networking is uninformed by the realities of deployed systems. This dissertation remedies this situation. We examine content delivery from the point of view of four content delivery systems: HTTP Web traffic, the Akamai content delivery network, and the Kazaa and Gnutella peer-to-peer file sharing networks. Our results (1) quantify the rapidly increasing importance of new content delivery systems, particularly peer-to-peer networks, and (2) characterize peer-to-peer systems both from an infrastructure and workload perspective. Overall, these results provide a new understanding of the behavior of the modern Internet and present a strong basis for the design of newer content delivery systems.