摘要
为解决DHT系统查询本地化的问题以及提高DHT系统查询效率,提出了多拓扑结构的方法。该方法的基本思想是基于一系列准则把DHT系统的逻辑空间划分成一系列的子空间,以经典Chord算法为例,按照两种不同的准则对Chord环拆分以实现多拓扑结构,每一个准则形成一组子环集合;采用两种不同的准则,即按照节点所属ISP关系以及查询量的分布分别对Chord环进行拆分得到多个子环。理论分析与仿真表明,按照ISP所属关系的划分方式能够完美解决查询本地化的问题,按照查询量分布准则的划分方式则能够大大提高系统的查询效率。同时,多拓扑方法易于推广到Pastry、CAN等其他DHT系统。
Aiming to solve the lookup localization problem of DHT systems and improve the lookup efficiency of DHT systems,this paper proposed a multi-topologies method.The basic idea of the multi-topologies method was to divide the logical space of DHT into a set of small blocks.Based on Chord,a classic algorithm of DHT,the multi-topologies method divided the Chord ring into multiple small rings according to two rules.Here,rule 1 formed the nodes which belonged to the same ISP into one group,rule 2 formed the nodes into one group according to the distribution of lookup traffic.The theory analysis and simulation results show that rule 1 can perfectly solve the problem of lookup locality,and rule 2 can greatly improve the lookup efficiency of DHT system.
出处
《计算机应用研究》
CSCD
北大核心
2013年第4期1133-1135,共3页
Application Research of Computers
基金
国家自然科学基金资助项目(61001084)
