SaaS云服务弹性特征测试方法的研究与实现

发布时间：2018-05-17 08:47

本文选题：SaaS云服务 + 弹性特征测试　；参考：《内蒙古大学》2016年硕士论文

【摘要】：随着云服务的普及,客户可以根据实际需求向云服务提供商定购所需的应用软件服务,并通过互联网获得云服务提供商提供的服务。云服务采用即付即用(pay-as-you-go model)模型,云服务提供商计算各资源(CPU、内存和公网带宽等)的使用总量,最后按不同资源的使用费用标准计算出总的使用云端资源的费用。弹性是指云服务提供商会根据应用负载的增加或减少实时地增加或移除云资源,这使云端资源得到了合理有效地利用,避免了资源的浪费。云端的资源浪费少了,用户需要支付的SaaS云端资源使用费用也就少了,确保了用户的利益。弹性是随着用户负载的变化云端分配给用户云端资源的自由伸缩性,很多云提供商都声称弹性是他们拥有的优点,但对于用户来说无法考量所使用的云平台的弹性水平。云端资源的使用情况对用户是不透明的,云服务用户为了满足系统的正常运行,通常一直申请用户认为能足够满足用户最大需求的云端资源,这将造成不必要的云端资源的浪费。因此,本论文针对云服务的弹性特征提出一种弹性测试方法,即从云端资源的使用情况角度评测云服务的弹性水平,对比分析不同用户负载下的云端资源使用情况,给出弹性的云端资源分配方案,云服务用户可以根据自身的需求变化申请所需的云资源,避免云资源的浪费。论文主要完成以下三方面的工作：(1)针对云服务弹性特征测试需求的测试例生成：基于LoadRunner编写云用户请求云服务脚本模拟单个用户请求云服务,将运行测试例生成的用户负载加载到Hadoop技术搭建的云端。(2)基于dstat的云端弹性资源的监控：首先在云端部署性能监控工具dstat,然后在模拟的用户负载加载到云端的过程中实时监控云端资源的使用情况,最后将监控到的云端资源的使用情况的数据导出生成折线图,方便对弹性性能测试结果的分析。(3)弹性性能测试的执行及测试结果的分析：通过对比分析不同的测试例负载运行过程中云端弹性资源的使用情况图形,得出弹性云端资源分配方案。用户可以根据自身的需求变化申请所需的云端资源,这样既保证了云服务的质量,也避免了云端资源的浪费。
[Abstract]:With the popularity of cloud services, customers can order the required application software services from cloud service providers according to their actual needs, and obtain the services provided by cloud service providers through the Internet. The cloud service adopts pay-as-you-go model. The cloud service provider calculates the total amount of each resource CPU, memory and public network bandwidth, etc. Finally, the total cost of using cloud resources is calculated according to the cost standard of different resources. Elasticity means that cloud service providers increase or remove cloud resources in real time according to the increase or decrease of application load, which makes cloud resources be used reasonably and effectively, and the waste of resources is avoided. The waste of resources in the cloud is reduced, and the cost of using SaaS cloud resources is reduced, which ensures the interests of users. Elasticity is the free scalability of cloud resources assigned to users with the change of user load. Many cloud providers claim that flexibility is their advantage, but for users, the level of flexibility of the cloud platform can not be considered. The use of cloud resources is opaque to users. In order to meet the normal operation of the system, cloud service users usually apply for cloud resources that the users think can meet the maximum needs of users. This will result in unnecessary waste of cloud resources. Therefore, this paper proposes an elastic testing method for cloud service elasticity, which is to evaluate cloud service elasticity level from the perspective of cloud resource usage, and to compare and analyze cloud resource usage under different user loads. A flexible cloud resource allocation scheme is presented. Cloud service users can apply for cloud resources according to their own needs and avoid the waste of cloud resources. This paper mainly completes the following three aspects of work: 1) Generation of test cases for the test requirements of cloud service elasticity: writing cloud user request cloud service script based on LoadRunner to simulate a single user requesting cloud service, Load the user load generated by the running test case into the cloud built by Hadoop technology) Monitoring of cloud elastic resources based on dstat: first deploy the performance monitoring tool dstat in the cloud, then load the simulated user load into the cloud The program monitors the use of cloud resources in real time. Finally, the data of the monitored use of cloud resources are exported to generate a broken line diagram. It is convenient to analyze the test results of elastic properties. (3) the execution of elastic performance tests and the analysis of test results: by comparing and analyzing the usage of cloud elastic resources in the load running process of different test cases, The flexible cloud resource allocation scheme is obtained. Users can apply for cloud resources according to their needs, which not only guarantees the quality of cloud services, but also avoids the waste of cloud resources.
【学位授予单位】：内蒙古大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP393.09

【相似文献】