【摘要】 为了降低访问时延,提升用户体验,当前网络交互性能改进的主要手段包括缓存技术与预取技术。当前的缓存替换机制主要考虑对象的访问时间与访问频度,然而Web对象本身存在语义性。本文首先对实际Web数据访问情况进行分析,发现访问间隔的变化率对于命中率的影响具有更高的准确性,进而提出一种基于对象属性的缓存替换策略,该策略通过统计近期缓存对象的平均访问间隔,并结合该对象的标签属性作为对象在缓存中的价值。实验结果表明该策略比基于Aging的缓存策略和基于协作式中心化决策缓存策略提升7%-10%的命中率。更多还原

【Abstract】 In order to decrease the access delay and improve the user experience,the current schemes include cache and pre-fetching technology. The current cache replacement schemes only consider the arrive time and frequency. The research finds the access interval change rate that is more valuable in predicting the new objects arrival through analyzing the real network logs. Considering this new metric,the paper proposes a novel cache replacement algorithm based on object property. Using this novel method,the cache can achieve higher byte hit ratio. The experiments result shows that the proposed method improves 7% to 10% hit rate than the ABC( Age-based-Cooperative-Caching) and( APDR) content-Aware Placement Discovery and Replacement schemes.更多还原