结构化对等网中一种副本概率选择算法

Probabilistic replica selection algorithm in peer-to-peer networks

结构化P2P系统常采用为热点文档创建副本的方法,降低存有热点文档节点的负载。从如何高效利用副本的角度出发,提出了一种副本概率选择算法(PRS),采用能者多劳的思想,请求数据包以较大的概率转发到负载轻的副本节点上。让轻负载节点分担更多的负载,充分利用了副本分担负载的能力。模拟实验表明,该算法在创建副本数相同的情况下,大大降低了丢包数,提高了副本的利用率。

Peer-to-peer systems usually used replication techniques to reduce the load of the nodes that host frequently-accessed files. In order to make full use of replicas, this paper proposed a probabilistic replica selection algorithm （ PRS）, in which queries were forwarded to the low-load nodes with high probability. This complies with the principle that the one with more capacity should take more responsibility. In this way, low-load nodes could be used to shed more load, therefore could make full use of the replicas. Simulation result shows that this algorithm can largely decrease dropped queries with the same replica number compared to other algorithm, and increase the utilization of the replica.