Active sharing memory excessive allocation method in multi-virtual machine system

Abstract

The invention relates to an active sharing memory excessive allocation method in a virtual system. A traditional virtual system adopts an excessive allocation strategy for a memory in order to meet the peak value performance of a client virtual machine, and the waste of memory resources under low load is always caused. The method comprises three parts of content: a client virtual machine memory page marking method based on a tile record method and a residual quantity linked list record method, an active memory sharing method based on a virtual channel and a multi-virtual machine memory excessive allocation method based on the above two methods. According to the active sharing excessive allocation method of the virtual system, a competition degree of system resources can be effectively alleviated under a condition that the system memory is highly competitive, and the performance and the service quality of the client virtual machine system under the whole multi-tenant cloud computing environment can be obviously improved.

Description

technical field [0001] The present invention relates to a method for implementing memory resource management and allocation in a virtualized computer system, especially an elastic cloud computing system with large-scale deployment of virtualization technology, such as the management and allocation of memory resources between virtual machines in Internet data centers and multi-tenant cloud computing environments method. Background technique [0002] In a multi-virtual machine (Virtual Machine, VM) system, multiple virtual machines share hardware resources of the same physical machine (Physical Machine, PM), such as processors, memory, disks and network devices. As the number of virtual machines increases, hardware resources can become over-committed overall. In a multi-tenant cloud computing environment, since the load peak times of different virtual machines may be different, in order to carry the largest customer virtual machine, the over-allocation of physical hardware re...

AI Technical Summary

Problems solved by technology

[0004] However, because the memory management method in the existing virtualization platform cannot distinguish the specific state of the internal memory pages of the guest on the physical machine, the memory pages of the guest are only accessed by the physical machine in the form of the frequency of access and the repetition of content. Perceived, there may be some unreasonable situations, various existing memory overuse strategies also have certain limitations and defects, there are problems such as waste of memory resources, untimely recycling, etc., affecting the performance of the entire virtualization system

Images