摘要
针对Web系统中热点数据短期内访问量高,生存周期短的特点,本文以电商系统作为研究对象,提出了一种针对热点数据预测及其缓存管理方法。首先采用支持向量机(SVM)算法对热点数据进行预测和筛选;其次,针对预测数据设计一个层次模型进行缓存管理;最后在LRU算法基础上改进淘汰策略以实现缓存数据的有效置换。此外,系统通过缩短热点数据生存期的过期清除策略,对缓存中的数据进行定时清除,并利用内存阈值策略来保证缓存空间占用率。通过与传统缓存系统中数据响应时间对比,证明基于热点数据的缓存管理系统具有更快的响应速度。
Aiming at the high-speed data and short life cycle of hotspot data in Web system, this paper takes E-commerce system as the research object, and proposes a method for hotspot data prediction and its cache management. Firstly, the support vector machine(SVM) algorithm is used to predict and filter the hotspot data. Secondly, a hierarchical model is designed for the predictive data for cache management. Finally, the elimination strategy is improved based on the LRU algorithm to achieve effective replacement of the cached data. In addition, the system clears the data in the cache by shortening the expiration-clearing policy of the hotspot data lifetime, and uses the memory threshold policy to ensure the cache space occupancy rate. Compared with the data response time in the traditional cache system, it is proved that the cache management system based on hotspot data has faster response speed.
作者
韩兵
张转霞
方英兰
HAN Bing;ZHANG Zhuan-xia;FANG Ying-lan
出处
《信息技术与信息化》
2019年第12期187-190,共4页
Information Technology and Informatization
