Research On Container Cloud Resource Scheduling Strategy Based On Docker

Container technology has brought great changes to the field of cloud computing,which has greatly improved cloud platform resource scheduling in terms of delay,granularity and resource utilization.Container technology is based on the operating system kernel,As a result,the original resource management mode based on virtual machine is not fully applicable to the container cloud based on Docker container technology.At the same time,considering that the running environment on which the container depends can be physical or virtual,the resource scheduling of the container becomes more complex.In view of the above problems,this thesis focuses on studying and optimizing the resource scheduling and container migration strategies of the container cloud platform,so as to reduce the SLA default rate while improving resource utilization.The specific work is as follows:1.The service openness of the container cloud platform is the reason why it is widely deployed.In order to support various kinds of services,container resource scheduling will be greatly differentiated,and the application deployment with the container as the resource allocation unit will inevitably lead to the resource fragmentation problem of the container.To solve this problem,a QoS-EACO container scheduling strategy was proposed,which evaluated the load state of nodes by combining node attribute information.The candidate node set of container scheduling is selected by constructing evaluation model.Furthermore,the container QoS model is constructed,and the container scheduling is guided by the improved ant colony algorithm,and the optimal solution of container scheduling is obtained.Compared with the current mainstream container scheduling strategies,the experimental results show that the QoS-EACO container scheduling strategy can reduce the container resource fragmentation,improve the resource utilization rate while ensuring the container service quality.2.The load and node resources of container services will change dynamically with the operation of the container cloud platform,and the initial scheduling strategy cannot always meet the resource requirements of container services.Therefore,a LB-ECMS container migration strategy was proposed.In this strategy,the three-threshold method was used as the critieria of load balancing,and the node load difference matrix was constructed.The migration time and migration frequency were considered in container selection,and the load balancing degree was defined to guide the selection of target nodes.The experimental comparison with the current mainstream migration strategy shows that LB-ECMS container migration strategy can improve the node resource balance and effectively reduce the SLA default rate of container cloud platform.The QoS-EACO container scheduling strategy and LB-ECMS container migration strategy are proposed in this thesis.These methods can effectively improve the resource utilization of container cloud platform and reduce the SLA default rate,which has good practical significance.