Design And Implementation Of Elastic Cloud Resource Scheduling System Based On Service Quality Evaluation

With the rapid development of cloud computing technology,major governments,enterprises,and engineering fields have gradually replaced traditional software with cloud applications in a service-oriented architecture(SOA)model.As a cloud application developer,by renting a large amount of cloud resources to improve the service quality of own cloud applications,in order to solve the huge number of online applications,increasing data volume,fluctuating loads and so on to provide users with convenient and stable services.At the same time,cloud resource providers have also continued to expand the number and scale of cloud data centers.The elastic cloud products provided for commercialization are also varied and difficult for developers to choose.Therefore,how to make the cloud application developers dynamically choose the right types and quantity of elastic cloud products according to the requirements of the QoS at different stages,so as to minimize the cost,increase the efficiency and ensure the long-term stability of the system is an urgent problem needed to be solved.For this reason,this paper will design and implement the scheduling system of elastic cloud resources based on service quality evaluation,and provide different stages of services for cloud application developers.Before the cloud application runs,according to the relationship between the optimization of the cost and quality of service,the suitable cloud products and quantities are allocated to provide a reasonable deployment plan.When the cloud application runs,combined with load prediction results and real-time monitoring status,a dynamic scheduling scheme for elastic cloud instances is provided.The main work includes the following aspects:Firstly,in order to fully understand the requirements of the elastic cloud resource scheduling system,this paper analyzes the functional requirements,non functional requirements and business processes,clarifies the relationship between the roles in the system and the main functions of the roles,and elaborates the key business processes in detail.Secondly,for evaluated the quality of service of cloud applications,this paper proposes a method based on queuing model.A cloud resource consumption model is established to deploy single service and multi service cloud instances respectively.Then according to the cloud resource consumption model,the single-server and multi-server queuing models are used to analyze the QoS metrics in single node and cluster environment,and the evaluation results are generated.Besides the evaluation method is verified by experiments.Thirdly,in order to solve the problem of optimal allocation and dynamic scheduling of elastic cloud resources,this paper establishes a problem model for the optimal allocation problem and use the improved genetic algorithm based on variable neighborhood to solve it.Then based on the historical load time sequence of cloud applications,the BP neural network is used to predict the load of the future time points.A dynamic scheduling strategy is generated by combining prediction results with real-time monitoring.After that,design experiments and comparative analysis of optimal allocation algorithm and dynamic scheduling strategy.Finally,based on the above analysis and design,this paper designs the architecture,function module and database of the elastic cloud resource scheduling system.An elastic cloud resource scheduling system based on QoS evaluation is implemented,and complete the testing of the function and performance.