一种面向SaaS应用多级定制模型的缓存技术研究  

RESEARCH ON AN SAAS APPLICATION MULTI-LEVEL CUSTOMIZATION MODEL ORIENTED CACHE TECHNOLOGY

作　　者：孙明[1] 孔兰菊[1] 李庆忠[1] 桑成良[1] 郑旭旭[1] 

机构地区：[1]山东大学计算机科学与技术学院,山东济南250101

出　　处：《计算机应用与软件》2011年第11期49-51,95,共4页Computer Applications and Software

基　　金：国家自然科学基金(90818001);国家科技支撑计划(2009BAH44B02);山东省自然科学基金(2009ZRB019YT)

摘　　要：对共享架构共享存储的SaaS应用来说,如何实现SaaS应用数据低代价高速查询一直是其中的焦点问题之一。在元数据驱动的SaaS应用多级定制模型中,定制元数据存在层次关系,查询代价较高,普通的数据查询优化机制不足以实现SaaS应用的性能优势。采用元数据传递闭包策略解决该问题,针对多级定制的SaaS应用的特点,提出面向租户的两级缓存模型及查询代价最小的最近最少使用缓存替换算法(Min-QT-LRU算法),实验结果表明,该策略能支持更高效的元数据访问。For shared database and shared schema SaaS applications, how to complement low-cost and high-speed SaaS application data query has always been a focused issue. In a metadata driven SaaS application muhi-level customization model, as there is hierarchy among customization metadata, querying costs high, so that the ordinary data query optimization mechanism is insufficient to achieve the performance advantages of SaaS applications. The paper uses metadata transitive closure strategy to solve the dilemma. Aiming at the characteristics of muhi-level customization in SaaS applications, the authors propose a tenant oriented two level cache model as well as query cost minimization LRU algorithm called Min-QT-LRU. Experiments show that the strategy can provide more efficient metadata access.

关 键 词：多级定制 元数据 两级缓存 缓存替换 

分 类 号：TP333[自动化与计算机技术—计算机系统结构]