The Design And Implementation Of A Virtual Machine Management Platform

Virtualization technology has been widely used in cloud computing, especially in IaaS(Infrastructure as a Service) mode. It has become a pressing problem to manage the cluster with increasing numbers of virtual machines in cloud computing. So does the resources allocation. Besides charging according to the usage amount of public cloud management platform, the greater challenge is how to make the virtual machine management of private cloud good considering the cost control and efficiency.This paper focuses on the management and integration of virtual machine resource in IaaS cloud management platform. Each KVM (Kernel-based Virtual Machine) and QEMU(Quick Emulator)on physical machine can be managed by using libvirt API. Menu-style model is developed to simplify the configuration of virtual machines, so as to achieve the purpose of saving time and reducing errors. According to the usage of both CPU and memory on physical machines and virtual machines which is monitored by SNMP(Simple Network Management Protocol), the workloads of physical machines and virtual machines can be adjusted. Depending on the state of virtual machine, the memory configuration can be automatically adjusted, so as to achieve the goal of dynamic load distribution in addition with the dynamic migration of virtual machine mechanism. Permission control and simple virtual machine deployment have been realized by the management platform. The current state of the virtual machines and physical machines in the platform were obtained by monitoring them. To implement the memory load balancing, the platform will automatically execute memory increasing or dynamic memory migration when the virtual machine is out of memory. On the opposite, the memory will be recovered automatically. The operation results show that, the data center is the one that improves resource utilization and simplifies the management.The author participates in completing following works:(1) Designing the architecture, database and function modules of the system based on the business requirements of virtual machine management platform;(2) According to the system architecture design, accomplish the development of virtual machine management platform by using KVM and libvirt, which consist of several function modules, such as user management module, control module and controlled module. Robustness of the system and friendliness of user interface are both taken into consideration;(3) Based on the virtual machine management platform, the dynamic memory adjustment strategy can be designed and implemented in a higher utilization ratio of system resources, so as to achieve memory load balancing.