Research And Implementation Of D-open Stack Multi-cluster Collaborative Architecture And Virtual Machine Migration Scheduling Strategy

Cloud computing service is a new type of computing service in recent years.It can configure and manage various resources,and use a unified resource pool to provide users with various computing resources and improve the utilization rate of these resources.With the rapid development of cloud computing and the continuous expansion of the size of data centers,the following problems are prone to occur.First,a data center often contains multiple clusters,and the resource usage of these clusters is not the same.Some clusters are nearly full,while some clusters have a large amount of available resources.Because clusters are independent from each other,it is impossible to cooperate and schedule multiple clusters in a unified way,so it is easy to cause the overall load imbalance of the data center.Second,most of the industry uses virtual machine migration technology to adjust the load of the cluster.However,most cloud platforms do not provide virtual machine migration technology between clusters,such as Open Stack and Cloud Stack.When a cluster has a large load and needs to migrate virtual machines,it can only be migrated manually,that is to make a snapshot of virtual machines,then import the snapshot to another cloud platform and restart the virtual machine.Third,with the expansion of the data center,the energy consumption and network cost of the data center also need to be focused on.Too much energy consumption will bring a lot of economic losses to enterprises,while too much network cost will reduce the quality of service.To solve the appeal problem,this paper has done the following three research:(1)First,we studied the Open Stack cloud platform,and proposed a multi-cluster collaborative architecture,d-Open Stack,which is based on Open Stack.This architecture realizes the multi cluster collaborative cooperation,and can manage multiple clusters and schedule virtual machines in a unified way.(2)Based on d-openstack,cross cluster migration technology is implemented,including static migration and dynamic migration.Cross cluster migration technology provides the basis for multi cluster virtual machine migration scheduling.At the same time,cross cluster migration technology has good disaster recovery ability.When a cluster fails and cannot provide services,it can quickly migrate virtual machines to available clusters.(3)This paper proposes a scheduling strategy for virtual machine migration based on load and network awareness.This strategy schedules the migration of virtual machines from the perspective of the entire data center.First,the load prediction and dynamic threshold is used to select the host with unbalanced load,and then according to the load correlation,select the virtual machines that need to be migrated from the host,and finally select the appropriate target host for the virtual machine to be migrated from the entire data center according to the network-aware way.Through experimental verification,the scheme in this paper can reduce the energy consumption and network overhead of the entire data center.