Research And Implementationon Of The Test Of The Elastic Characteristics For SaaS Cloud Services

With the popularity of cloud services, customers can order and acquire application software services according to their actual demands from cloud services providers via the Internet. Cloud services adopt the pay-as-you-go model. Cloud service providers compute the total amount of the resources (CPU, memory, and public network bandwidth, etc.), and finally calculate the total cost of cloud resources according to the different charge standards of different resources. Elasticity refers to that cloud services providers can add or remove cloud resources in real time according to the increase or decrease of load. This makes cloud resources used effectively and reasonably, and avoids the waste of resources. The less waste of cloud resources decreases the payment of cloud resources, and this will ensure the interests of users.Elasticity is the flexibility of the cloud resources which are allocated to users according to user requirements. So most of the cloud service providers claim elasticity as the advantage of their services. But users hardly estimate the elasticity level of the cloud platform. Due to the opaqueness of the usage of cloud resources, users of cloud services always apply for sufficient cloud resources, which cloud satisfy the maximum requirements of users. This will lead to the waste of the cloud resources. So, in terms of the elastic characteristics of cloud services, this thesis proposes an elastic test method, which evaluates the elastic level of the cloud services under the perspective of the usage of cloud resources, and analyze the usage of cloud resources used by different users under different loads and finally gives elastic cloud resources allocation scheme. The cloud service users can apply for the required cloud resources according to the change of their demands. This can avoid the waste of the cloud resources.This thesis mainly completes the following three aspects:(1) Requirements of elastic characteristics oriented test case generation for the cloud service. First, we write test scripts to simulate a single user requesting cloud services based on JavaScript development technology of the LoadRunner. Second, we perform the test case loads to the cloud constructed with Hadoop technique.(2) Monitoring the cloud resource usage based on dstat. First, we deploy the performance monitoring tools named dstat. Second, we monitor the cloud resources usage in real time during the process of loading user load to the cloud. Finally, in order to convenient the analysis of the elastic performance test results, the data of the cloud resource usage is exported to the line chart.(3) Test execution towards the elastic characteristics testing and analysis of the elastic test results. We obtain the elastic resource allocation scheme by analyzing the usage of the cloud elastic resources during the process of the operation of different test cases contrastingly. Users can apply for the required cloud resources according to the change of their needs. This not only guarantees the quality of service but also avoids the waste of the cloud resource.