论文部分内容阅读
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.
The active measurement of P2P system to understand its status and trends can provide a reliable measurement basis for its system modeling and simulation.Based on the coupon collection problem, an optimized BT cluster rapid measurement model is proposed.The model expands the data by reducing the expectation and expanding the data Source and concurrent requests can reduce the number of request experiments and information repetition rate 67.12%, and can cover 95% of the cluster nodes.An active measurement platform is built based on this model, and 49854 Torrentz clusters and 98 HDChina regional PT clusters For a long time measurement and analysis.The results show that the vast majority of the cluster size does not exceed 10000 nodes, 95% of its coverage to complete the snapshot time of not more than 22 minutes.For nodes in the cluster, then analyzed its geographical distribution, use Preference and transmission performance, etc. In addition, comparing Torrentz with HDChina cluster, we found that the latter, as a regional PT cluster, has a small cluster and a concentrated geographical distribution, but the node download speed is generally faster.