一种结构化P2P协议中的缓存计算模型

Caching Model in Structured P2P Protocol

P2P系统是一种分布式系统,降低系统负载对提高P2P系统的扩展性具有重要意义.目前结构化P2P协议主要利用缓存方法解决查询热点问题,这些缓存方法对数据缓存后的收益无法预测,是一种较盲目的缓存,往往导致系统负载加重.本文提出一种结构化P2P协议中的缓存计算模型,该模型以降低系统负载为目标,节点采用一种文件访问统计方法跟踪查询本地文件的消息途经邻居节点的历史次数,根据历史统计记录预测缓存该文件到邻居节点可能减少的查询负载,然后与缓存后可能产生的更新开销对比,进而确定是否缓存文件到相应的邻居节点.实验表明,该缓存计算模型可以有效降低系统负载.

Peer to peer systems are distributed systems, reducing system load is important for improving the sealability of the systems. Caching is always used to achieve load balance in structured P2P systems currently, but none of the current caching algorithms have mechanisms to compute whether a caching is worthwhile, so the current caching algorithms are blindness and always increase system load. This paper presents a caching model in structured peer to peer systems. In order to reduce system load,This caching model uses a passive file requested statistic method which records the current accessing times of each neighbor,and then estimates the reducible query load on the assumption that we cache the file to each neighbor. The reducible query load and the overhead caused by caching determine whether it is worthwhile to cache the file to a neighbor. Simulation results indicate the caching model can reduce system load effectively.