用户访问特征驱动的中间件语义缓存替换策略

An user access feature driven semantic cache replacement policy in middleware

Web应用中大量用户请求频繁与数据库交互会严重影响系统性能,中间件层缓存有助于减少请求与数据库的交互开销、提高服务器的响应速度。研究了在Web应用服务器中引入语义缓存机制,在查询缓存时利用语义信息进行匹配,然后动态追踪和分析用户访问时间性、访问内容量、业务关联等用户访问特征,在此基础上设计一种基于用户访问特征的语义缓存替换算法。原型实验表明其能够提高缓存命中率,并使服务器的响应时间和吞吐量也得到改善。

The performance of Web applications may be restricted because of large numbers of users interacting with database frequently.Cache in middleware can reduce the cost of the interaction between the requests and the database,and improve the response time.A mechanism of semantic cache is introduced in Web application server.It uses semantic information to match for querying cache,and then traces and analyzes the user access features including user visit timing,visiting content and business association.An efficient cache replacement policy based on the above features is presented.The prototypical experiments shows that the policy is effective at promoting the cache hit rate,and the response time and the throughput of application server can also be improved.