论文部分内容阅读
相对于传统网络服务系统的客户端/服务器构架,分布式VoD系统通过系统中节点间的协作能够获得更好的性能.而节点的存储空间和网络带宽是有限的,因此缓存管理的研究对流媒体服务系统的研发具有重要的意义.该文基于一种灵活的热度估计方法对系统中的存储资源进行分层分析,给出一种减小服务器磁盘I/O负载的优化式及其次优解,同时针对客户节点的不稳定特性提出一种以二次规划为基础的数据置换算法.仿真实验结果表明热度的计算能够较好的反映用户行为的变化,而缓存管理策略有利于减小服务器磁盘数据的读写以及增加系统的可利用资源.
Compared with the client / server architecture of traditional network service system, the distributed VoD system can achieve better performance through the cooperation among nodes in the system, while the node storage space and network bandwidth are limited, The development of service system is of great significance.This paper presents a flexible method of heat estimation based on hierarchical analysis of storage resources in the system, and presents an optimized and second-best solution to reduce server disk I / O load, At the same time, a data replacement algorithm based on quadratic programming is proposed for the unstable characteristics of customer nodes.The simulation results show that the calculation of heat can reflect the change of user behavior better, and the cache management strategy is helpful to reduce the server disk data Read and write and increase the system’s available resources.