Data Replication Strategies in P2P Systems: A Survey

Data replication means storing copies of the same data at multiple peers thus improving availability and scalability. Replication can be seen as a family of techniques. Full documents or just chunks can be replicated. Since the same data can be found at multiple peers, availability is assured in case of peer failure. Consistency is a challenge in replication systems that allow dynamic updates of replicas. Fundamental to any of them is the degree of replication (full or partial), as well as the source of the updates and the way updates are propagated in the system. In this paper we study some data replication strategies and requirements for different P2P applications. We identify several contexts and use cases where data replication can be used. We will also discuss about existing optimistic replication solutions and P2P replication strategies and analyze their advantages and disadvantages.