Peer-to-Peer (P2P) systems are very useful for managing large amounts of widely distributed data. For this purpose Distributed Hash Tables (DHT) offer a highly scalable and self-organizing paradigm for efficient distribution and retrieval of data. Thereby a common assumption of P2P-Systems is, that the participating nodes are unreliable and may fail at any time. Since many of research goes into the design of DHT lookup services, these systems aim to provide a stable global addressing structure. But to storage data reliable in a DHT only few techniques were already developed. However since data has to be stored persistent in the network, it should be retrieved anytime, even if nodes fail. In this work we discuss possibilities to store data fault tolerant in a structured Peer-to-Peer system.

Abstract—Structured Peer-to-Peer (P2P) systems are highly scalable, self-organizing, and support efficient lookups. Furthermore, Distributed Hash Tables (DHTs), due to their features, are used more and more for file-sharing and content distribution applications. However, a major weakness of traditional DHTs is the search for stored content, as data is assigned to nodes based on hash functions, and the very nature of hash tables allows only exact pattern matches. We present Cerco, a solution for the problem of range queries by using the principles of orderpreserving DHTs and a hierarchically structured P2P approach. To guarantee an efficient routing and load balancing, Cerco uses a dynamic hierarchy of DHTs by creating subrings on demand and two explicit load balancing strategies. Our evaluation shows that Cerco is able to achieve the goals of supporting range queries, logarithmic-hop routing, and efficient load balancing.