Research And Implementation Of Dynamic Resource Scheduling Based On Kubernetes

Cloud computing is a new business model that provides computing resources to users to meet users' complex dynamic resource requirements,thereby reducing the cost of infrastructure and hardware maintenance.How to dynamically supply the cloud platform 's physical resources,such as CPU,memory,disk and network bandwidth,to the cloud in the most efficient way to users,it need to be addressed.At present,the management and scheduling of cloud computing resources are mostly focused on the IaaS layer.When the PaaS platform applies from the IaaS layer to the resource,how to maximize the resource utilization is very important.For the container cloud platform,the focus of the study is how to manage and schedule the application container resources.Based on the container cloud platform,this dissertation studies the dynamic scheduling of container resources,examples of automatic scaling,load balancing and so on.This dissertation first focuses on the representative open source container cloud platform Kubernetes,and details of the architecture and design of Kubernetes.Then,a variety of forecasting models for forecasting application resource consumption are researched.The ARIMA model and neural network model are studied emphatically.Based on these two models,a combined model is proposed.Based on the forecasting results of the combined model,a resource dynamic scheduling algorithm is proposed which can dynamically schedule resource allocation for the nodes in the Kubemetes cluster.And then design a load-based example of automatic expansion module,the application of resource-sensitive degree of load balancing module.Finally,a series of experiments are designal to verify the dynamic scheduling,automatic scaling and load balancing of the extended Kubernetes.The experimental results show that the extended Kubernetes can effectively improve the resource utilization and the service quality of the application.