对等网络中基于位置信息和文件流行度的自适应复本管理算法

Adaptive Replication Management Algorithm Based on Location and File Popularity for Peer-to-Peer Network

文件共享服务是对等网络中的一个重要应用,数据传输速率逐渐取代响应延迟成为影响用户体验的首要因素.文中研究了对等网络中的副本管理算法,这对于提高对等网络应用的可靠性,降低带宽消耗具有重要的意义.为了在广域网络存储系统中加速文件共享并降低网络带宽消耗,文中提出了PLAR(Popularity and Locality-based Adaptive Replication)算法.PLAR采用了基于位置信息和流行度的复本管理算法,该算法还同时引入了混合式的服务器选择策略以及远程增强策略.PLAR算法在文中的Granary对等广域网存储系统中得到了实现.实验表明,通过PLAR算法下载速率平均能提高60%以上,有效提高了共享速度并减少带宽消耗.

File sharing is one of the most important applications based on the peer-to-peer net- work structure. In such application, the data transferring speed is more important than the net- work latency. The authors have done some work on the replication algorithms based on the peer- to-peer network. Such algorithms are meaningful for improving the reliability of network applica- tions as well as reducing the network traffic. For improving the performance of file sharing with reduced network traffic, this paper proposes a method called PLAR （Popularity and Locality- based Adaptive Replication）. PLAR is based on the locality and popularity of objects as well as applying the mixed server selection method and remote boosting schemas. PLAR is used in the authors~ Granary system which is data storage system for wide area network. The experiment re- sults show that PLAR can improve the download speed over 60 ~ on average which means that it can improve the sharing as well as reducing the network traffic.