Research And Implementation Of OpenStack-based Cloud Platform Resource Scheduling Strategies

Cloud computing,which subverted the management and usage patterns of the traditional IT resources,has became a new focus during the development of IT industry.With the gradual popularization of cloud computing,many companies began to establish th'eir own public or private cloud computing platform one after another.OpenStack,as the mainstream of open source IaaS in the cloud computing field,is a good choice for enterprise,government agencies and research institutions to deploy cloud computing platform.It provides a solution of cloud platform deployment for users by encapsulating the underlying huge hardware resource into an abstract cloud resource pool.However,OpenStack is not perfect.There are still some deficiencies in its resource scheduling.For example,in the initial allocation phase of virtual machines,OpenStack fails to comprehensively take the resources conditions of physical hosts into account,leading to a low utilization rate of cloud platform resources;in the running phase of virtual machines,OpenStack cannot adjust itself timely according to the real-time load of physical hosts,resulting an imbalanced physical hosts' load of the cloud platform and a low cloud service quality.To solve the above problems,in this thesis,the mainstream cloud platform resource scheduling strategies in the domestic and foreign research are studied first and classified based on the research directions.A typical resource scheduling strategy in each direction is then introduced.Whereafter,the OpenStack resource scheduling and virtual machine migration mechanism are deeply studied and analyzed.Based on this,a dynamic resource scheduling strategy is thus proposed.The dynamic resource scheduling strategy proposed in this thesis optimizes the OpenStack initial allocation strategy of virtual machine by combining the resource needs of the virtual machines to select the target physical host.It makes the initial allocation of virtual machines reasonable.In the running phase of virtual machines,this strategy can guarantee a balance physical hosts' load by making use of the online migration technology of virtual machine and adjusting virtual machines in real time based on the loading conditions of physical hosts.Finally,a cloud computing simulation platform,CloudSim,is exploited to test the proposed strategy.The experimental results show that the propsed dynamic resource scheduling strategy can effectively realize the load balance of physical hosts in the cloud platform,ensure the service quality and improve the resource utilization ratio of the cloud platform.