面向部署方案的SaaS软件性能建模与分析方法  

Deployment-oriented Performance Modeling and Analysis of SaaS Software

作　　者：李琳[1,2] 应时[1,2] 董波[1,2] 

机构地区：[1]武汉大学软件工程国家重点实验室,武汉430072 [2]武汉大学计算机学院,武汉430072

出　　处：《小型微型计算机系统》2016年第12期2660-2666,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61373038)资助;国家"八六三"高技术研究发展计划项目(2012AA011204)资助

摘　　要：部署方案的抉择是影响SaaS软件运行性能的一个关键因素,因此,有必要为SaaS软件构建性能模型,以预测其在不同部署方案下的运行性能.然而,现有的研究大多关注外部环境对软件性能的影响,提出的建模方法不适用于部署方案变化的性能预测情况.为此,本文基于排队Petri网提出一种新的性能建模方法.该方法从三个层次来构建SaaS软件的性能模型,各层之间通过映射关系相关联,使得模型具有更高的灵活性和可读性,从而能够更方便的预测SaaS软件在不同部署方案下的性能.最后,利用该方法对不同的SaaS软件案例进行建模,并利用QPME工具对其进行模拟,以研究不同部署方案以及不同交互关系对软件性能的影响.The decision on deployment is a key factor that influences the running performance of SaaS software. Therefore,it is neces- sary to build a performance model for SaaS software to predict its performance under different deployments. However,most of existing researches mainly focus on the influence of the external environment on the performance of software,and the modeling methods in them are not appropriate for the performance prediction with deployment changing. Therefore, this paper proposes a new performance modeling method based on queuing Petri net. The method build SaaS software performance model in three different levels, which are related by the mapping relationships, making the model more flexible and readable, so that it can be more convenient to predict SaaS software performance under different deployments. Finally, the method is used to construct performance models for different SaaS soft- ware cases, and the QPME tool is used for simulation. From the simulated results, the influence of different deployment plans and dif- ferent interaction relationships on the performance of software is studied.

关 键 词：SaaS软件 排队Petri网 性能建模 性能分析 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]