一种优化的P2P集群快速测量模型  

An Optimized Agile Measurement Model on P2P Swarms

作　　者：杨哲[1,2] 李领治[1,2] 纪其进[1,2] 朱艳琴[1,2] 

机构地区：[1]苏州大学计算机科学与技术学院,江苏苏州215006 [2]江苏省计算机信息处理技术重点实验室,江苏苏州215006

出　　处：《小型微型计算机系统》2014年第2期296-302,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61070170)资助;江苏省高校自然科学基金项目(11KJB520017)资助;苏州市科技计划项目(SYG201238)资助

摘　　要：对P2P系统进行主动测量,了解其现状和变化趋势,能为其系统建模和仿真提供可靠的测量依据.基于赠券收集问题,提出优化的BT集群快速测量模型.模型通过降低期望、扩大数据源和并发请求等方法,能减少请求实验次数和信息重复率67.12%,且能覆盖集群95%的节点.基于该模型搭建了主动测量平台,对49854个Torrentz集群和98个HDChina区域性PT集群进行长时间的测量与分析.结果表明,绝大多数集群规模不超过10000个节点,对其完成95%覆盖的快照时间不超过22分钟.对于集群中的节点,则分析了其地理分布、使用偏好及传输性能等特征.此外,将Torrentz和HDChina集群对比后发现,后者作为区域性的PT集群,集群规模小、地域分布集中,但是节点的下载速率普遍较快.Active measurement on peer-to-peer system is the best way to understand the current situation and its evolution trends. And it also provides the real measurement basis to support the research on system simulation and optimization. Based on optimized coupon collector＇s problem in random sampling, agile measurement model was proposed to guide the active measurement on BitTorrent swarm. And with the optimization of lowering expectations, expanding the data resource and requesting concurrently, the request rep- etition rate is 67.12% off, while still covering the 95% peers in a swarm. Moreover, the measurement time can be further shortened by interacting with peers by DHT and PEX. Based on the theoretical model, a low cost measurement platform was set up to trace the 49,854 Torrentz swarms and 98 HDChina swarms for a long time. Most of them hosted no more than 10,000 peers, and the average snapshot time on them was less than 22 minutes. Furthermore, the geographical characteristics, user preference and transmit perform- ance of peers were also analyzed. Because HDChina provides regional PT download, so the swarm size was smaller, peers distribu- tion was more centralized, and the download speed was faster than Torrentz.

关 键 词：P2P集群 主动测量 赠券收集问题 地域特性 用户偏好 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]