Research On Memory Management In Virtualization Cloud Platforms

Cloud computing consolidates network,storage and computing resource,and it can flexibly provide users with a variety of high-quality computing services by the Internet.Virtualization technology is the foundation of cloud computing,and it can achieve the efficient management and use of computer resources.Memory virtualization is the most complex part of virtualization technology,and it is the key to improving the efficiency of virtual machine.While running different applications,virtual machines have a different requirements of memory.The traditional memory virtualization management is not able to efficiently allocate memory dynamically for virtual machines according to the memory usage of the virtual machine.This situation will often result in waste of memory resources.In this thesis we propose an efficient system of memory virtualization management,which is based on KVM.The system is mainly composed of three parts: virtual machine memory monitor module,virtual machine memory balancing module and multi-host memory balancing module.First we present an efficient way to detect the memory demand of virtual machine in real time different with other ways.With knowledge of virtual machine memory demand we propose a reasonable strategy to allocate virtual machine memory,which is based on ant colony algorithm.We use virtual machine memory balloon technology and virtual machine memory hotplug technology to effectively adjust the virtual machine memory according to the virtual machine memory usage.Traditional memory virtualization management can only adjust virtual machine memory in single.However,this system can achieve multi-host memory balancing by virtual machine online migration technology.Finally,our experimental results show that The memory management system not only can effectively adjust the virtual machine memory,but also achieve multi-host memory balancing.This system can improve the utilization of computer memory resources,and it will significantly increase the performance of overall system about 20%.