Research And Implementationon On Cloud Resource Elastic Scheduling Method

With the increased popularity of cloud computing,more and more enterprises migrate their software system to cloud platform,and the elasticity services of it can automatically deal with the change of workload,which increasing the usage of cloud resource.Elasticity is that cloud platform can expand or shrink cloud resources automatically according to the change of workload.A suitable purchase scheme is significant to the application on cloud,including the selection of elasticity levels and the decision that how much cloud resources should be bought.Although most cloud services providers say that elasticity is their advantage of cloud platform,however,it is impossible for cloud resource consumers to know the specific usage of the cloud resources,and they also cannot consider that whether the used elasticity level is suitable for their products or not.Therefore,we expect to find out a feasible testing method which can help users to acquire the best elasticity level and specific usage of the cloud resources when application run on the cloud platform.For the cloud resource consumers,the best elasticity level means the lowest usage of resource.And this result of testing can offer the best purchase scheme which helps users to get the required performance with the lowest cost;For cloud services providers,best elasticity level represent the best elasticity service.The result can provide the optimal scheme of resource allocation,which the cloud resource can be used sufficiently.This thesis,from the perspective of test,proposes and implements an available elastic scheduling method of cloud resources.The main works of this thesis is as follows:(1)Constructing the elasticity cloud platform which integrate OpenStack with Docker:OpenStack and Docker are complentary.The elasticity ruleset can be set free by the integration of OpenStack and Docker,which simulate a variety of elasticity levels for application test.(2)Designing the test scenario based on CloudSuite:According to the user's application behavior,application test scenario is divided into three types:benchmark application test scenario,horizontal application development test scenario and vertical application development test scenarios.The applications are deployed to the platform by writing hot template code according to these test scenarios,which simulate the user's cloud application and its workload.(3)Generating the test case based on elasticity ruleset:The rule set consists of monitoring interval,upper threshold,lower threshold,upper breach duration,lower breach duration and cool down period.And different elasticity rulesets are designed by equivalent division and boundary value method.Then,according to different application test scenarios,the test cases are designed based on rulesets and virtual machine system performance.(4)Analyzing the execution and result of elasticity scheduling test:Ceilometer module in OpenStack is used to monitor the execution and collect the test data,which writes to log.The best purchase scheme of elasticity cloud resources under various test scenarios will be obtained by analyzing the resource scheduling of each test case.