信息中心网络中内置缓存块调度策略的研究

发布时间：2018-09-08 06:47

【摘要】：伴随着用户海量的内容获取需求,互联网正从主机驱动的点对点的通信模式逐渐转变为接受者驱动的内容获取模式。在这种模式下,用户希望更高效地获取信息,而不关心信息来源于哪个主机终端。面对这种通信模式的变化,信息中心网络被提出,这种以信息为中心的新型网络体系架构摒弃了过去以网络中间件的形式,而是在网络体系架构层面,通过基于内容项命名的路由方式和网络内置缓存技术,为提高用户层面可感知的高效内容获取以及整个网络的资源利用率提供原生支持。然而信息中心网络内置缓存所呈现出的缓存细粒度化、透明化以及泛在化等特征,在没有相应优化技术与方法的情况下,应对大量需要转发的内容时表现出能力相对弱小,并且内容放置缺乏平衡分布等问题。为了解决上述问题,本论文研究信息中心网络内置缓存块调度策略,主要在缓存决策策略和缓存替换算法上对信息中心网络缓存机制的性能进行了优化,并使内容放置实现更平衡的分布。在缓存决策策略的优化设计方面,首先本论文分析以信息中心网络默认缓存策略on-path caching为代表的全局性缓存会导致的问题;接着针对其产生的大量无效缓存替换以及内容放置失衡的问题,提出基于内容流行度和节点中心度匹配的缓存策略。该策略通过对缓存内容和缓存节点进行多参量评估,选择性缓存经过的内容,以降低缓存冗余、提高内容分发沿路节点的空间使用效率。在缓存替换算法的研究方面,本论文首先分析研究了在信息中心网络中体现出的与传统web/流媒体等缓存系统不同的新特征,然后分析了适用于web/流媒体缓存系统的LRU,FIFO等单参量缓存替换算法在移植到信息中心网络中所导致的缓存污染等问题,提出基于生存时间和缓存大小的信息中心网络缓存替换算法。该算法结合了信息中心网络中数据包对生存时间的原生支持以及缓存节点空间大小,建立效用函数,选取当前节点最适合替换的内容项进行缓存替换,让更流行的对象更靠近接近用户的网络边缘缓存节点,并获得更持久贮存,以提高缓存系统整体的缓存命中率。最后本论文通过使用加入了外生请求局部性原理后生成的改进模型,提高算法策略验证的准确性和真实性。结合改进的模型,将缓存决策策略和替换算法运用于搭建的运行CCNx协议的实验床,经试验分析发现,本文提出的缓存决策策略和缓存替换算法实现了对内容更平衡的放置、减少了缓存冗余、同时提高了缓存系统的稳定性。
[Abstract]:With the huge demand of users, the Internet is changing from the host-driven point-to-point communication mode to the receptor-driven mode of content acquisition. In this mode, users want to obtain information more efficiently, regardless of which host terminal the information comes from. In the face of the change of communication mode, information center network is proposed. This new network architecture based on information has abandoned the form of network middleware, but at the level of network architecture. By means of content-based routing and network built-in cache technology, native support is provided for improving efficient content acquisition at the user level and resource utilization of the entire network. However, the information center network has the characteristics of fine-grained cache, transparency and ubiquity. Without the corresponding optimization techniques and methods, it has a relatively weak ability to deal with a large number of contents that need to be forwarded. And the content of the lack of balanced distribution and other issues. In order to solve the above problems, this paper studies the scheduling strategy of the built-in cache block in the information center network, and mainly optimizes the performance of the cache mechanism in the information center network based on the cache decision strategy and cache replacement algorithm. And make the content placement to achieve a more balanced distribution. In the aspect of the optimization design of cache decision policy, this paper firstly analyzes the problems caused by global cache, which is represented by default cache policy on-path caching in information center network. Then a cache policy based on the matching of content popularity and node centrality is proposed to solve the problem of a large number of invalid cache replacements and the imbalance of content placement. By evaluating the cache content and the cache nodes, the strategy can selectively cache the content, reduce the cache redundancy and improve the spatial efficiency of the content distribution nodes along the route. In the aspect of cache replacement algorithm, this paper firstly analyzes and studies the new features which are different from the traditional web/ streaming media cache system in the information center network. Then it analyzes the problems of cache pollution caused by LRU,FIFO and other single parameter cache replacement algorithms which are suitable for web/ streaming media cache system when they are transplanted to the information center network. An information center cache replacement algorithm based on lifetime and cache size is proposed. The algorithm combines the native support of data packets in the information center network and the size of the cache node space to establish the utility function and select the most suitable content items for the current node to cache replace. The more popular objects are closer to the user's network edge cache nodes and can be stored more persistently to improve the cache hit ratio of the whole cache system. Finally, an improved model based on the principle of exogenous request locality is used to improve the accuracy and authenticity of the algorithm strategy verification. Combined with the improved model, the cache decision strategy and the replacement algorithm are applied to the experimental bed of running CCNx protocol. Through the experimental analysis, it is found that the cache decision strategy and the cache replacement algorithm proposed in this paper achieve a more balanced placement of the content. It reduces the cache redundancy and improves the stability of the cache system.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP393.0

【相似文献】