Measurement Of The Cloud Service Performance

In recent years, the cloud computing service has gained a lot of attention anddevelopment. Cloud service providers lease their cloud platform, and in accordancewith demand charges. In the cloud computing data center, they have the separateinfrastructure. Compared with the local traditional computer-based model, in termsof cost control and reliability it has the unprecedented advantages. At the same time,the cloud platform has a charge model of pay-as-you-go, which enables the user topay for what he uses. Moreover, it allows allow for expansion of the instance basedon demand.In the background of large-scale development of cloud computing, more andmore companies have launched their own cloud computing products, such asAmazon and Google. These cloud providers offer a wide variety of cloud platformswhich offers different levels of service. However, due to different hardware devicesand virtualization extent, the performance of the cloud platform is also different.There are too many cloud platforms to choose and the user can’t make a decision.In this thesis, in order to help users choose a suitable cloud, we studied thevirtualization performance measurement technology of SPECjvm. Throughconstructing a cloud service performance measurement platform between the clientand server-side, implementing various services accessing, then upon the cloudplatform, we develop a cloud service quality measurement framework, a systematiccomparator of the performance and cost of cloud providers. The frameworkmeasures the elastic computing, persistent storage, and networking services offeredby a cloud along metrics that directly reflect their impact on the performance ofcustomer applications. We use three metrics to measure the performance of elastic,and the storage is all the same. And the optimal is used on the cloud network latency.We used the framework to perform a comprehensive measurement study over fourpopular and representative cloud providers: Amazon AWS, Microsoft Azure, GoogleAppEngine and Rackspace, whose together account for most of the cloud customers today. In the experiment, we find that their offered services vary widely inperformance and costs, underscoring the need for thoughtful provider selection. Atthe same time, we can also help the cloud provider to find out its poor performanceof the service, and then put forward the corresponding optimization scheme for theaspects, improving its quality of the cloud services.