Data Replication in P2P Collaborative Systems

Data replication techniques have been extensively used in distributed systems to achieve, among others, due to system nodes failures: (a) high data availability, (b) system's reliability and (c) scalability. Due to the various characteristics of distributed systems as well as system's and application's requirements, a variety of data replication techniques have been proposed in the distributed computing field. One important distributed computing paradigm is that of P2P systems, which distinguish for their large scale and unreliable nature. The study and application of replication techniques becomes a must in such systems. While it is well understood and easy to achieve replication of immutable information (typically files) in P2P systems, it becomes more challenging to implement data replication techniques of dynamic data under highly dynamic nature of large P2P systems. Indeed, replicating documents that could change over time requires addressing the consistency issues. In this paper we study some data replication techniques for P2P collaborative systems. We identify several contexts and use cases where data replication can greatly support collaboration. We then consider as a case study replication techniques for dynamic documents in the context of a peer-group based P2P system of super-peer architecture. P2P collaborative systems arise in many groupware applications, such as collaborative work in online teams, having requirements for high availability and system reliability (e.g. in disaster management scenarios). We propose a replication system for documents structured as XML files to address the dynamics of the documents at peers and use the super-peer to ensure a satisfactory level of document consistency among peers.