针对应用服务器对数据库的大规模并发访问所产生的性能下降问题,设计了面向应用服务器的缓存对象管理组件.集成了分布式对象缓存系统Memcached客户端,并采用基于分级队列思想的缓存替换算法.根据队列中元数据内容将可缓存对象存储在不同级别的分布式缓存池中.模拟实验表明该机制能够有效减少数据库I/O负载,降低应用服务器的访问延迟,提高了系统的并发处理能力.%To the performance degradation problem caused by large-scale concurrent access to database from application server, Cache Object Management Component( COMC) is designed which integrates the client of dis-tributed memory object caching system Memcached. And an improved cache replacement algorithm in COMC based on Grading Queue is proposed, cacheable objects can be stored in different-level cache pool according to the meta-data. Result of simulation experiment proves that this mechanism reduces I/O accessing load of database and appli-cation server's system latency effectively, and improves concurrent processing capacity.
展开▼