Design And Implementation Of Virtual Machine Scheduling Subsystem Based On OpenLava

With the development of high-performance computing, cluster system with its high cost performance, good scalability and high availability gradually becoming the theme of today’s high performance computer architecture. In the cluster system, the cluster job scheduling system integrates the distributed computing resources, and rational allocates of resources based on the resource requirements of the job, and then improve resource utilization of the system. Rely on the existing cluster system about the management of Open Lava job and combined with the actual needs of users. The paper is solved the problem that the Windows application job is lack of the operating resource in the Linux-based compute nodes of the cluster environment. It is designed and realized the virtual machine scheduling subsystem. The subsystem could provide appropriate resources for different operating system applications job, and improve cluster efficiency in the use of existing hardware resources.Firstly, This paper is analyzed the current research about the cluster job scheduling strategy and virtualization technology at home and abroad, and determined the research direction of the paper according to user needs. Secondly, it is studied the overall framework of OpenLava systems, background daemons, job scheduling strategies, and proposed the design scheme of virtual machine scheduling subsystem on Open Lava systems. The subsystem is divided into three modules by function, namely the virtual machine configuration modules, the virtual machine clone module and the virtual machine scheduling module. The paper will design and implementation of the functions of each module. Finally, Thesis focuses on the description of functional testing result on the virtual machine scheduling subsystem. It is realized, when the cluster environment is not Windows compute nodes. The user submits the job required Windows operating environment to the cluster, the virtual machine scheduling subsystem can be dynamically added Windows virtual machines to high-performance computing clusters, and the job is distributed to the virtual machine execution, when the job is finished, Windows virtual machine automatically turns off the cluster without taking the cluster resources. It is implemented to provide appropriate computing nodes resources for different system applications, solve the problem that the cluster is lack of Windows computing cluster node resource, save hardware resources and improve the use efficiency of hardware resources by virtualization technology.