基于Chord协议的P2P负载均衡机制研究

发布时间：2018-04-16 09:14

本文选题：对等网络 + 热点资源　；参考：《沈阳航空航天大学》2014年硕士论文

【摘要】：结构化对等（peer-to-peer，简称P2P）网络是一种分布式网络。由于网络中用户对存储资源的需求存在不均衡性，以及各计算机的处理能力的差异，导致了P2P网络中某些节点的访问量会超过该节点的实际处理能力，网络出现负载不均衡现象，造成节点性能下降、请求延迟、网络拥塞等性能问题，影响整个网络的服务质量。因此有效地解决结构化P2P网络的负载均衡问题，对网络性能的提升、服务质量的保证具有积极的意义。对引起P2P网络负载不均衡的热点问题进行分析，研究了热点问题产生的原因及主要的解决方案，在分析已有解决方案的利弊因素之后，选取改进的复制技术作为本课题所采用的负载均衡方法。提出了基于资源访问热点预测解决负载不均衡问题的方法。利用网络节点中文件访问量的历史数据预测节点近期的文件访问量，，对可能出现的访问热点提前进行副本复制，进而缓解网络中的负载不均衡问题。论文比较了一次指数平滑模型、BP神经网络模型及ARMA模型三种模型作为预测模型的预测效果，通过对实验结果的分析，该方法能够对热点资源进行有效预测，减少热点资源所在节点因为访问过量而产生超载问题。论文对近期的热点资源利用复制副本的方法防止集中访问，选择合适的节点放置资源副本。论文在热点资源的形成，预测模型的选取，副本创建的数量及副本缓存节点的选取上都进行了深入的研究。最终的实验结果表明本文的算法在资源利用率和节点响应时间上都得到了明显改进。
[Abstract]:Structured peer-to-peer (P2P) network is a distributed network.Because of the imbalance of the users' demand for storage resources and the difference of the processing power of each computer, some nodes in P2P network will have more visits than the actual processing capacity of the node.Network load imbalance causes performance problems such as node performance degradation, request delay, network congestion and so on, which affects the quality of service of the whole network.Therefore, it is of positive significance to solve the load balancing problem of structured P2P network effectively and to improve the network performance and guarantee the quality of service.This paper analyzes the hot issues that cause the load imbalance in P2P networks, studies the causes of the hot issues and the main solutions. After analyzing the pros and cons of the existing solutions,The improved replication technique is chosen as the load balancing method in this paper.A method to solve the load imbalance problem based on hot spot prediction of resource access is proposed.Using the historical data of file access in network nodes, the paper predicts the recent file visits of nodes, and replicates the possible access hotspots in advance, thus alleviating the problem of load imbalance in the network.In this paper, the prediction results of three kinds of prediction models are compared between the primary exponential smoothing model and the ARMA model. Through the analysis of the experimental results, the method can effectively predict the hot resources.Reduce the problem of overloading due to excessive access to the nodes where the hot resources are located.In this paper, we use replica method to prevent centralized access to hot resources, and select appropriate nodes to place resource replicas.In this paper, the formation of hot resources, the selection of prediction model, the number of replica creation and the selection of replica cache nodes are deeply studied.The experimental results show that the proposed algorithm has been improved obviously in terms of resource utilization and node response time.
【学位授予单位】：沈阳航空航天大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【相似文献】