如何有效评估服务系统的性能表现是解决服务对动态网络环境适配性的关键.然而,基于互联网的服务系统的规模尺度和复杂程度不断提高,使得服务系统的性能建模和分析越来越困难.通过分析服务系统的执行过程,提出了一种基于排队Petri网的性能建模和分析方法.该方法将服务系统运行过程分为服务准备和服务提供两个阶段.服务准备阶段被建模为多层Web系统,刻画了浏览、选择、组合和注册4种行为的性能表现.服务提供阶段被建模为组合服务,重点讨论了交互关系、节点失效和恢复机制对组合服务执行性能的影响,并在此基础上建立组合服务在集中和分散两种执行方式下的性能模型.最后,采用QPME工具仿真和比较提出的模型在不同配置下的性能表现.结果表明,该方法在定量评估服务系统性能时具有一定的优越性.%How to evaluate the performance of service systems is the key for resolving the adaptation problem between services and dynamic network environment. Nevertheless, the increasing of size and complexity make it more difficult to model and analyze the performance of service systems. In this paper, an analytic model is proposed for evaluating performance of service systems using Queueing Petri Nets. The approach divides the execution process of service systems into two stages, service preparing and services providing. The service preparing stage is modeled as a multi-tier Web system, which can show the performance characteristics of four behaviors including browsing, selection, composition and register. The service providing stage is modeled as a composite service, which focuses on targeting the impact of interaction relations, node failure and node recovery on the system performance. Based on the interaction model and node model, two performance models of composite service with centralized and decentralized execution are given. Finally, QPME tool is used to simulate and compare the parameters under different configures. The simulation results show that this method can quantitatively analyze the performance of service systems.
展开▼
