PeerCollaboration

An increasing number of Internet users and growing Internet bandwidth availability and its usage lead to the need for scalable systems, which scale with the increasing number of users and bandwidth. Client/server (C/S) systems have capacity limits and as soon as these limits are reached, the system infrastructure needs to be upgraded or replaced. Peer-to-peer (P2P) systems do not have such capacity limitations, because every participating peer is a client and a server, contributing its resources. Thus, the more users a P2P system has, the more resources are contributed. Compared to C/S systems, P2P systems shift infrastructure expenses from the server to its peers, reducing the expenses at the server-side. However, P2P systems require additional mechanisms to offer the same functionality as a centralized system. Large-scale C/S systems such as Wikipedia, Facebook, or YouTube need to keep extending their server infrastructure due to increasing enduser bandwidth and number of users. To cover theses infrastructure expenses, operators of such C/S systems need to generate sufficient revenues. Lack of revenues leads to reducing or shutting down services. Since P2P systems distribute its infrastructure expenses among its peers, operators could run such C/S systems at a lower cost if changing to a P2P system. The advantages for various P2P systems have been demonstrated by many real-world applications, such as file sharing applications, Voice over Internet Protocol applications, or media streaming applications. However, new types of P2P applications are still challenging as they require new distributed mechanisms and algorithms. The goal of this thesis is to present new distributed mechanisms and algorithms for a scalable, robust, and fault-tolerant, P2P document collaboration system. In a P2P document collaboration systems, users share the resources necessary to host and distribute documents. Supported tasks in such a system are searching, retrieving, creating, changing, and maintaining documents.