最小化同儕式檔案儲存系統之綜合性等待時間

Abstract

Peer-to-peer (P2P) storage systems are cost-effective and scalable platforms that enable users to replicate data on other peers to support variant emerging applications, such as P2P social networks (e.g., PeerSoN [1]) and distributed backup systems (e.g., Wuala [2]). In a P2P storage system, each peer helps backup data of other peers, and all peers form a virtual storage system. Such a system potentially avoids data loss caused by single point failure, and provides a more reliable storage system with high data availability compared to conventional server-based storage systems. However, due to the uneven data popularity and heterogeneous user upload capacity in P2P storage systems, bandwidth-oblivious replica placement might cause extreme long response time for certain data objects. Thus, this paper proposes a new metric called joint response time, which not only considers the waiting time when the requested data is unavailable but also the queuing delay and service time when data become available. We then design a 2-dimensional Markov model to estimate this metric, and propose a bandwidth-aware replica placement algorithm to reduce the joint response time. Our trace-driven evaluation validates that the proposed bandwidth-aware algorithm can efficiently utilize the upload capacity of each peer to provide a shorter joint response time.