Research Of Resource Scheduling Method Based On Openstack Cloud Computing Platform For Virtual Machines

Cloud computing has become a reality through the advance in the Internet and virtualization technologies. It integrates a large number of resources into a huge resource pool. Users can access computing, storage, networking and other services as demand. Cloud computing adds the need for extreme flexibility and geographic dispersion for any time, everywhere availability. The virtual resource scheduling can stretch elasticity based on the user requests, and thus it can increase the resource utilization rate, and reduce the costs of system operation and maintenance.With the popularity of cloud computing technology, the scale of cloud computing platform becomes more and more immense. The diversity of cloud services and the heterogeneity of resources bring challenges to virtual resource management. Cloud computing virtual resources mainly are organized and distributed in the form of virtual machines (VMs). With the rapid development of cloud computing and virtualization, VMs scheduling has become a particularly important topic, and VMs scheduling algorithm is also growing in importance. In recent years, as a good open source cloud computing system, OpenStack is developing rapidly, which has become a mainstream architecture of cloud computing platform. Therefore, it has a very important practical significance to study the management of VMs scheduling in OpenStack.This paper studied and discussed the optimization of VMs scheduling methods. The main research contents and innovations are as follows:First, the integrated load model of physical machines is optimized. The model integrates CPU, memory, network bandwidth, storage, and all kinds of the resources of physical machines. The utilization rate of multi-dimensional physical machine resources is converted into the evaluation value of the actual load of physical machines by weighted-sum approach.Second, the resource demand model of VMs is optimized. The model is based on cloud computing resource pool. It simplifies the method to measure the resources requested by the virtual machine, and improves the accuracy.Third, we design a forecasting load model of VMs. The model reflects the scarcity of all kinds of resources in physical machines, and it is able to forecast the balancing degree of the remaining resources after virtual machines are deployed to different physical machines.Furthermore, based on the optimum algorithm (BFD), we design an optimization VMs scheduling algorithm. The algorithm is able to improve the resource utilization on the cloud computing platform, reduce the energy consumption of data centers, and improve the load balancing degree of resources in physical machines.Finally, according to the virtual machine scheduling method proposed above, we optimize the scheduling strategy of OpenStack VMs. In order to test the effectiveness of the virtual machine scheduling strategy, we have several comparison experiments of the distributed resource scheduling algorithms by CloudSim, which is a mainstream resource scheduling simulation system. Experimental results show that the proposed virtual machine scheduling strategies have good effects in reducing data center's energy consumption, improving the utilization rate of resources and balancing resources of physical machines.