Abstract
|
Peer-to-Peer (P2P) networks enable access to shared distributed resources across the Internet. However, the availability of these resources is hindered by the members¡¯ transient participation (i.e., churn) and uncooperative behaviors (e.g., free-riding). Content redundancy using the idle storage space of nodes can be used to improve content availability. Maintain the scalability and self-organization properties of a P2P system, however, requires (i) minimizing the redundancy repair traffic (caused by churn), (ii) self-organizing mechanisms to balance the load and (iii) mechanisms to promote cooperation and enforce fair exchange of resources. In this paper, we take a holistic approach to content availability and propose a framework centered on efficient content redundancy, low-overhead maintenance and repair and incentives to mitigate the impact of churn. To this end, we propose a redundancy scheme that requires reduced repair bandwidth to improve content availability. The scheme is augmented with an efficient redundancy maintenance process to automate repairs. We also introduce a novel incentive-based mechanism to ensure a sustained and fair participation of peers and fair content sharing. The proposed redundancy scheme, referred to as Proactive Repair (PR), is studied analytically. The analysis shows that its repair bandwidth outperforms that of erasure coding and exact-MBR network coding. The proposed algorithms and mechanisms are implemented in an experimental testbed to evaluate their performance. The results indicate that our proposed solution is feasible and that it can improve content availability in P2P networks significantly.
|