In this paper, we study the viability of a peer-to-peer backup system on nowadays internet connections. In particular, we show that peer lifetime estimation can be used to reduce the maintenance cost of peer-to-peer backup. Previous studies [5] have shown that lifetimes in a peer-to-peer system follow a Pareto distribution. Consequently, peers can be sorted on their expected lifetimes, depending only on the length of their history in the system. By carefully selecting the peers on which backup data is stored, repairing cost can be highly reduced for long-term backup users, while it is still acceptable for new users. The efficiency of this technique is evaluated through simulations of a state-of-the-art peer-topeer backup system. Therefore, our long term objective is to provide a new alternative backup system, free, efficient, reliable, yet still easyto-use, to help people protect data on their personal computers. Our system would be based on the exchange of free disk space between its members. To lower the cost and to bring a high resilience to faults, it has to be decentralized and to work in a peer-to-peer (P2P) way. In the literature, there are already some P2P backup systems. But most of them are still in prototype phase and has not yet been tested in acceptable conditions. The other ones are too restrictive and do not offer the conditions we want. Before building a complete system, we want to simulate such systems in order to know what their viability is, based on common properties of already designed systems. 1.

Abstract. In this work we tackle the problem of on-line backup with a peer-to-peer approach. In contrast to current peer-to-peer architectures that build upon distributed hash-tables, we investigate whether an uncoordinated approach to data placement would prove effective in providing embedded incentives for users to offer local resources to the system. By modeling peers as selfish entities striving for minimizing their cost in participating to the system, we analyze equilibrium topologies that materialize from the process of peer selection, whereby peers establish bi-lateral links that involve storing data in a symmetric way. System stratification, that is the emergence of clusters gathering peers with similar contribution efforts, is an essential outcome of the peer selection process: peers are lured to improve the “quality ” of local resources they provide to access clusters with lower operational costs. Our results are corroborated by a numerical evaluation of the system that builds upon a polynomial-time best-response algorithm to the selfish neighbor selection game. 1

Abstract—Peer-to-peer (P2P) storage systems are strongly affected by churn —temporal and permanent peer failures. Because of this churn, the main requirement of such systems is to guarantee that stored objects can always be retrieved. This requirement is specially needed in two main situations: when users want to access the stored objects or when data maintenance processes have to repair lost information. To meet this requirement, exiting P2P storage systems introduce large amounts of redundancy that maintain data availability close to 100%. Unfortunately, these large amounts of redundancy increase the storage costs, either by reducing the overall net capacity or by increasing the communication required for data maintenance. In order to minimize storage costs, P2P storage systems can reduce data redundancy. However, less redundancy means lower data availability, which leads to increase object retrieval times. Unfortunately, longer retrieval times could compromise data maintenance processes and could penalize user’s retrieval times. It is crucial then for P2P storage systems to predict the effects of a redundancy reduction. In order to provide this information, we present a novel analytical framework to measure object retrieval times under different redundancy and churn circumstances. Our framework can be directly used by backup applications aiming to maintain durability at the lower cost, or by data sharing applications that seek to reduce costs by penalizing user retrieval times. We validate our framework by simulation using real P2P traces (Skype and eMule’s KAD). I.