服务工作流中服务时序组合的动态预取方法.pdfVIP

服务工作流中服务时序组合的动态预取方法 李海波1) 战德臣2) 徐晓飞2) 1(华侨大学计算机科学与技术学院,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) Dynamic Pre-fetching Method for Web Service Workflow System LI Hai-Bo1), ZHAN De-Chen2), XU Xiao-Fei2) 1)(College of Computer ScienceTechnology,Huaqiao University,Xiamen,361021,China) 2)(Center of Intelligent Computing of Enterprises Service, Harbin Institute of Technology, Harbin 150001, China) 摘 要: 服务工作流是实现分布式环境下服务资源协同工作的一种重要手段，在候选服务少 的情况下，在候选服务少的情况下，很难依靠目前服务质量的一些软件技术特性提高服务工 作流整体效率。把业务频率这一重要的非功能性因素融入到服务工作流中，同时结合服务的 使用频度和服务出现的间隔这两个因素，提出服务预取模型，包括相对依赖强度和绝对依赖 强度两个计算模型，前者体现了服务使用频度，后者体现了服务出现的间隔。业务频率则通 过服务工作流日志融入两个模型中，以此提高服务实例化时缓冲的命中率，缩短用户的访问 延迟。仿真实验结果表明，该方法具有服务工作流的优势，能为服务缓冲提供更加精确地预 测。 关键词: Web服务工作流; 业务频率; 预取模型;依赖 Abstract: Web Service Workflow is an important solution to make service resources work collaboratively in the distributed environment. If candidate services are few, it is difficult to improve the efficiency of whole workflow by QoS(Service of Quality) which depends on software technical features. Considering the important nonfunctional factor, business frequency in the web service workflow system, and combined it with the other two factors, frequency of the instantiated web service repeatedly and intervals of services defined in web service workflow, two web service pre-fetching models are proposed for web service workflow. One is a relative dependency model, which reflects the use frequency of a web service. The other is an absolute dependency model, which reflects intervals between services. Business frequency is merged into the two models by mining logs of web service workflow. In this way, it can improve the hit ratio when a web service instantiated, shorten the user access delay. Simulation results indicate that this approach has the advantage of web service workflow and provides a more accurate prediction for web