面向工作流管理系统的动态构件预取方法  

Dynamic Component Prefetching Method for Workflow Management System

作　　者：李海波[1] 战德臣[2] 徐晓飞[2] 

机构地区：[1]华侨大学计算机科学与技术学院,福建厦门361021 [2]哈尔滨工业大学企业与服务智能计算研究中心,哈尔滨150001

出　　处：《计算机学报》2012年第5期1038-1045,共8页Chinese Journal of Computers

基　　金：中央高校基本科研业务资助项目(JB-ZR1147);厦门市科技计划项目(3502Z20110013);泉州市科技计划项目(2011G5);福建省高校产学研科技重大项目(2010N5008)资助~~

摘　　要：基于软构件和工作流技术的ESA(Enterprise Software&Application)系统运行效率不仅取决于功能性的因素,还受到非功能性因素的影响,而且常常被忽视.把业务频率这一重要的非功能性因素融入到ESA系统中,同时结合构件的使用频度和构件出现的间隔这两个因素,提出构件预取模型,包括相对依赖强度和绝对依赖强度两个计算模型.前者体现了构件使用频度,后者体现了构件出现的间隔.业务频率则通过工作流日志融入两个模型中,以此提高构件实例化时缓冲的命中率,缩短应用服务器对用户访问的响应时间.最后的实验结果表明,该方法具有工作流的优势,能为构件缓冲提供更加准确的预测.Efficiency of ESA（Enterprise Software and Application）,which based on software component and workflow technologies,depends on not only functional factors,but also nonfunctional ones.These nonfunctional factors are often ignored.Considering the important nonfunctional factor,business frequency in ESA,and combined it with the other two factors,frequency of instantiated components repeatedly and intervals of components defined in workflow or component models,two component prefetching models are proposed for ESA.One is relative dependency model,which reflects frequency,and the other was absolute dependency model,which reflects intervals between components.Business frequency is merged into the two factors by mining workflow log.The model can provide accurate prediction for component cache to increase hit ratio when components are instantiated,shorten the application server＇s response time on user access.The final experiment result shows that our method has advantage of workflow technology and can provide a more accurate prediction.

关 键 词：工作流 软构件 预取模型 业务频率 依赖 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]