论文部分内容阅读
P2P应用在当今的互联网中占有了越来越重要的地位.由于这类系统往往由数千至数百万台计算机组成,没有一个大规模分布式系统模拟器是很难预测其行为的.文中尝试通过建立一个并行模拟器AegeanSim来预测大规模P2P系统的行为,该模拟器提供这类系统在高性能服务器集群上的的并行离散事件模拟.我们将P2P应用程序执行抽象成一个具体的事件模型,然后在一个集群上并行这些事件,从而不但扩展了模拟规模而且大大提高了模拟效率.文中还提出了一种提高性能的单步同步方法,并为研究人员设计了一个事件回调接口,使得在保持模拟器独立性的同时很容易地执行应用程序逻辑.用AegeanSim预测了一个典型的P2P系统:BitTorrent.比较了模拟的BT行为和目前BT相关的研究,证实了它的效率、可扩展性和准确性.文中还尝试预测BT系统受到各种人为攻击后的行为,得到一些合理的结果:(1)因为苛刻的攻击时间要求,Tracker隔离很难起作用;(2)限制BT的带宽可能是控制它的一种有效的方法.
P2P applications are playing a more and more important role in today’s Internet, and as such systems often consist of thousands to millions of computers, it is difficult to predict their behavior without a large-scale distributed system simulator Attempts to predict the behavior of large-scale P2P systems by establishing AegeanSim, a parallel simulator that provides parallel discrete event simulation of such systems on high-performance server clusters, we abstract P2P application execution into a concrete event model , And then parallelize these events on a cluster, which not only extends the simulation scale but also greatly improves the simulation efficiency.This paper also proposes a single-step synchronization method to improve performance and design an event callback interface for researchers, It is easy to implement the application logic while maintaining simulator independence, and AegeanSim predicts a typical P2P system, BitTorrent, that compares simulated BT behavior with current BT related research, confirming its efficiency, scalability, and Accuracy.In this paper, we also try to predict BT system after being subjected to various man-made attacks to get some reasonable Results: (1) severe attack because of time requirements, Tracker isolation difficult work; (2) limiting the bandwidth of BT control it may be an effective method.