This dissertation, "Scalable Content Distribution in Overlay Networks" by Tin-man, Tony, Kwan, 關天文, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Abstract of thesis entitled Scalable Content Distribution in Overlay Networks submitted by Kwan Tin Man Tony for the degree of Master of Philosophy at the University of Hong Kong in May 2007 Data distribution based on the traditional client-server model is generally inefficient for multicast communications. It suffers from several major problems, including concentrated workload, limited scalability, single point of failure and restricted routing path selection. Recently, overlay networks are shown to be an effective solution for multicast data distribution in the Internet. An overlay network is a peer-to-peer network constructed and maintained at application layer. In this thesis, we focus on designing scalable data distribution protocols in overlay networks. Two original research problems are investigated. We first study the problem of overlay network topology construction and maintenance. A new protocol called Overlay Multicast Tree Protocol (OMTP) is proposed. OMTP follows the direct-tree approach, where a single tree is constructed for dissemination of both data and control messages. Unlike existing schemes, OMTP leverages the IP hierarchical addressing locality for tree formation and maintenance. When a newcomer selects its parent end-host, both bandwidth availability and round-trip-time are considered. An efficient mechanism is also designed to disperse the simultaneous rejoin crowds in the tree repair phase. Simulation results show that the join latency experienced by a newcomer can be reduced by up to 50% as compared with a popular protocol HMTP. We then investigate the bandwidth allocation problem in content distribution networks (CDNs). CDN is an overlay network of CDN servers. It is specialized for bulk data distribution. Aiming at minimizing the content update time, the optimal bandwidth allocation among CDN servers is first analyzed. Based on it, a new bulk data distribution scheme ServerCast is proposed. In ServerCast, the origin server and all the edge servers collaborate and cooperate in data distribution. We show that ServerCast not only outperforms existing schemes, but also yields a content update time within 18% of our analytical lower bound. DOI: 10.5353/th_b3847900 Subjects: Peer-to-peer architecture (Computer networks)Multicasting (Computer networks)