P2P流媒体点播系统中数据缓存策略的研究

发布时间：2018-03-15 08:57

本文选题：流媒体点播系统　切入点：P2P　出处：《华南理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：随着互联网技术的发展和网络基础建设不断加强，流媒体应用发展迅速，并且仍在迅速增长。一方面P2P技术在流媒体发展过程中发挥着越来越重要的作用；另一方面随着近些年服务器性价比的提高，服务器在流媒体系统中的作用也有一定的加强。本文的研究对象是流媒体点播系统中的缓存策略，包括服务器集群中的副本放置策略和客户端P2P网络中的缓存管理策略。缓存策略对于流媒体点播系统有着重要的影响。集群中优化的副本放置，可以充分利用集群的存储空间和网络带宽，，减少服务拒绝率，提高系统的服务能力。P2P网络中的客户端通过将播放过的资源缓存在本地，不仅能够降低服务器负载，还能够提高资源传输效率，减小主干网络的压力。本文首先分类研究了服务器集群的副本放置策略，包括静态副本放置策略（MMPacking算法）和动态副本放置策略（基于阈值的副本放置策略），分析了它们的优缺点。之后研究了传统P2P流媒体缓存管理机制，进行分类说明。服务器集群中的副本和P2P网络中客户端的缓存一样，也是一种缓存，基于这种考虑，我们提出综合考虑集群副本和P2P流媒体缓存的一个方案。对于不同热度的流媒体资源采用不同的缓存方法。另外基于半结构化的P2P网络架构，针对服务器集群和P2P流媒体网络采用基于阈值的副本放置策略和基于段缓存价值的副本管理策略。之后在实验室项目的基础上给出了它们的实现。最后，通过仿真实验对方案的有效性进行了验证。
[Abstract]:With the development of Internet technology and the strengthening of network infrastructure, streaming media applications are developing rapidly and are still growing rapidly. On the one hand, P2P technology plays an increasingly important role in the development of streaming media. On the other hand, with the improvement of the performance and price ratio of servers in recent years, the role of servers in streaming media system has been strengthened. The research object of this paper is the cache strategy in streaming media on demand system. It includes replica placement policy in server cluster and cache management policy in client P2P network. Cache policy has an important impact on streaming media on demand system. It can make full use of the storage space and network bandwidth of the cluster, reduce the rejection rate of service, improve the service capability of the system. The client in the P2P network can not only reduce the server load by caching the played resources locally, but also reduce the server load. It can also improve the efficiency of resource transmission and reduce the pressure of backbone network. First of all, this paper classifies and studies the replica placement strategy of the server cluster. It includes static replica placement strategy (MMPacking algorithm) and dynamic replica placement strategy (replica placement strategy based on threshold), and analyzes their advantages and disadvantages. Then, the traditional P2P streaming media cache management mechanism is studied and classified. A copy in a server cluster is a cache, just like a client cache in a P2P network. We propose a scheme that considers cluster replica and P2P streaming media cache synthetically. Different caching methods are adopted for streaming media resources with different heat. In addition, based on semi-structured P2P network architecture, For the server cluster and P2P streaming media network, we adopt the replica placement strategy based on threshold and the replica management strategy based on the value of segment cache, and then give their implementation on the basis of the lab project. Finally, the effectiveness of the scheme is verified by simulation experiments.
【学位授予单位】：华南理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN948.64

【参考文献】