基于云计算虚拟化平台的内存管理研究

发布时间：2018-12-27 20:33

【摘要】：云计算技术能够整合网络、计算、存储等各种计算机资源,通过网络灵活的为用户提供各种优质的计算服务。虚拟化技术是云计算的基础,能够实现计算机资源的高效管理和使用。内存的虚拟化既是虚拟化技术中实现最复杂的部分,并且也是提高虚拟化效率的关键。虚拟化环境中随着不同应用的运行对内存的需求通常是变化的,而传统的内存虚拟化方案无法根据虚拟机的内存使用情况来高效的调整虚拟机内存,这种情况下往往会造成虚拟化平台内存资源的浪费。本文基于KVM虚拟化技术,设计了一个高效的内存管理系统。本系统主要有虚拟机内存监视模块,虚拟机内存平衡模块和多宿主机内存平衡模块三个部分组成。首先本文设计了一种实时和准确的虚拟机内存感知技术,与当前其它感知技术相比本文设计的该技术对宿主机和客户机造成开销都很小。通过实时获取的虚拟机内存使用情况,本文结合蚁群算法设计一种高效的虚拟机内存调整策略,能够合理分配虚拟机内存。并通过结合使用虚拟机内存气球技术和虚拟机内存热添加技术,实现两种技术的优势互补高效的调整虚拟机内存。与其它内存管理技术只能调整单个宿主机的下的内存使用不同,本系统还可以通过虚拟机在线迁移技术实现多宿主机间的内存平衡。最后对本系统进行了相关实验分析,结果表明本内存管理系统不仅能够高效的调整虚拟机内存,还能够实现多宿主机下的内存平衡。最后的综合性能测试表明本系统能够实现宿主机120%左右的内存超配使用,极大的提高计算机内存资源的利用率。
[Abstract]:Cloud computing technology can integrate network, computing, storage and other computer resources, through the network flexible to provide users with a variety of high-quality computing services. Virtualization technology is the foundation of cloud computing, which can realize the efficient management and use of computer resources. Memory virtualization is not only the most complex part of virtualization technology, but also the key to improve the efficiency of virtualization. In the virtualization environment, the memory requirement changes with the running of different applications, but the traditional memory virtualization scheme can not adjust the virtual machine memory efficiently according to the memory usage of the virtual machine. This kind of circumstance often can cause the waste of memory resource of virtualization platform. This paper designs an efficient memory management system based on KVM virtualization technology. The system consists of three parts: virtual machine memory monitor module, virtual machine memory balance module and multi-host memory balance module. Firstly, this paper designs a real-time and accurate memory awareness technology, which is less expensive for host and client than other technologies. Based on the real-time memory usage of virtual machine, this paper designs an efficient strategy of virtual machine memory adjustment combined with ant colony algorithm, which can allocate virtual machine memory reasonably. By combining virtual machine memory balloon technology and virtual machine memory hot addition technology, the two technologies can adjust virtual machine memory efficiently and mutually. Different from other memory management techniques which can only adjust the memory usage under a single host the system can also achieve memory balance between multiple hosts through virtual machine online migration technology. Finally, the experimental results show that the memory management system can not only adjust virtual machine memory efficiently, but also achieve memory balance under multiple hosts. Finally, the comprehensive performance test shows that the system can achieve about 120% of the host computer memory overmatch, greatly improve the utilization of computer memory resources.
【学位授予单位】：杭州电子科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP302;TP315

【参考文献】