Virtual machine startup acceleration method and apparatus

Abstract

The invention discloses a virtual machine startup acceleration method and apparatus. The method comprises the following steps that a system allocates a cloned volume for each virtual machine to serveas a system startup disk, and creates a file directory used for caching main volume data for the virtual machines cloned by a same main volume; when the virtual machine is started, the system controlsa corresponding IO process to load a corresponding cloned volume; when the IO process of the virtual machine reads IO data, if the data is located in the main volume, whether a cache file corresponding to the main volume data exists in a cache directory or not is queried; and if the cache file exists, a reading lock of the file is obtained firstly and then the corresponding file is opened to readthe data. According to the virtual machine startup acceleration method and apparatus provided by the invention, in combination with the reading / writing lock of the file and a mapping table of the cloned data, the main volume data of the virtual machine is cached by utilizing a local file system, so that the storage pressure during concurrent startup of the virtual machines is reduced and the timedelay of virtual machine startup is shortened; and the virtual machine startup acceleration method and apparatus is widely applied to the field of the virtual machines.

Description

technical field [0001] The invention relates to the field of virtual machines, in particular to a method and device for accelerating virtual machine startup. Background technique [0002] KVM: Kernel-based virtual machine is a common way of linux virtualization. [0003] Ceph: Open source distributed storage software. [0004] Qemu: IO simulation program for KVM virtualization scenarios. [0005] When solving the problem of virtual machine startup in the prior art, it is necessary to first copy the parent volume of the virtual machine to the memory disk, and then start the virtual machine, which increases the time for copying the virtual machine image. And if there are many different virtual machine templates, more memory needs to be consumed. [0006] In the configuration process of the virtualization solution using KVM+Ceph, a parent volume is usually created first, and the operating system and some necessary application software are installed. After installing the rel...

AI Technical Summary

Problems solved by technology

Since each virtual machine has an independent qemu and librbd process, although most of the data read by each virtual machine is the same, it cannot be shared; although the client side cache developed by Intel can cache the parent volume data locally, but When many virtual machines start at the same time, the local ssd has no parent volume data at this time, so the IO process of the virtual machine still reads the corresponding data from the back-end ceph storage cluster; when multiple virtual machines on the server start at the same time, because the data cannot be shared , causing a lot of IO data to be read repeatedly, which greatly increases the CPU pressure of the server and also lengthens the startup time of the virtual machine

Images