信息中心网络内缓存替换算法性能分析与优化

发布时间：2018-10-11 08:36

【摘要】：信息中心网络(information centric networking,ICN)是一类受到广泛关注的新型互联网体系结构.通过对网络内(in-network)缓存的充分利用,信息中心网络可以极大地增强内容分发效率.网络内缓存的管理机制一直是信息中心网络研究中的热点问题.分析了网络内缓存的基准缓存替换最近最少使用(least recently used,LRU)算法的性能,指出多跳LRU缓存中广泛存在的"缓存退化"问题:在首个缓存节点发生缺失的内容请求也很难被下游的缓存命中.针对这一问题,提出一种基于预过滤的O(1)复杂度的改进算法.在原有缓存前放置一个仅记录内容标识的预过滤缓存,完成对原始内容请求的整形,使得预处理后的请求流量可以更容易被后面几跳缓存命中.基于真实互联网拓扑的实验表明,在信息中心网络典型应用场景下,预过滤LRU的缓存命中率可以达到LRU的2~3倍.
[Abstract]:Information Center Network (information centric networking,ICN) is a new type of Internet architecture. By making full use of the in-network cache, the information center network can greatly enhance the efficiency of content distribution. Cache management mechanism has been a hot issue in the research of information center network. This paper analyzes the performance of the most recently used (least recently used,LRU algorithm for the base cache replacement in the network cache. The problem of "cache degradation" in multi-hop LRU cache is pointed out: it is difficult to miss the content request in the first cache node by the downstream cache. To solve this problem, an improved O (1) complexity algorithm based on prefilter is proposed. A pre-filtered cache which only records the content identification is placed before the original cache, which completes the shaping of the original content request, which makes the request flow after preprocessing easier to be hit by the next few hops of cache. Experiments based on real Internet topology show that the cache hit ratio of pre-filtered LRU can reach 2 times that of LRU in typical application scenarios of information center network.
【作者单位】：中国科学院计算技术研究所;中国电子科技集团公司第五十四研究所通信网信息传输与分发技术重点实验室;
【基金】：国家科技重大专项课题(2012ZX03002016) 国家科技支撑计划项目子课题(2012BAH01B03) 国家"八六三"高技术研究发展计划基金项目(2013AA013501) 国家自然科学基金重点项目(61133015,61272473) 中国科学院重大科研装备项目(YZ201229)
【分类号】：TP393.02;TP333

【参考文献】