Virtual processor scheduling method based on NUMA high-performance network processor loads

Abstract

The invention discloses a virtual processor scheduling method based on Non-Uniform Memory Access Architecture (NUMA) high-performance network processor loads. The method comprises the steps that when a virtual machine starts a virtual function network card distributed by a Single-Root I / O Virtualization (SR-IOV) high-performance network card, the affinity situation of a virtual machine network card buffer and an NUMA node is obtained by monitoring the running of the virtual machine network card; then, the VCPU running load situation is obtained for the VM and a DOM0, and the PCPU running loads on the NUMA node are obtained through the mapping relation between the VCPU and the PCPU; finally, according to the affinity of the NUMA node and the buffer and real-time load information of a processor of the NUMA node, the final scheduling result is obtained. The VCPU processing network data packets can have the advantages of the affinity and processor resources, and therefore the network performance of a high-performance network server under the virtualization environment can be obviously improved.

Description

technical field [0001] The invention relates to the field of computer system virtualization, in particular to a virtual processor scheduling method based on NUMA high-performance network processor load. Background technique [0002] Virtualization technology integrates and redistributes the underlying hardware resources, and virtualizes one physical server into multiple computers, thereby improving the utilization rate of hardware devices. A virtual machine monitor (Virtual Machine Management, VMM) is a software management layer that exists between hardware and traditional operating systems. It manages real physical resources (CPU, memory and I / O devices), and by abstracting physical resources and assigning them to the upper operating system, a server can run multiple operating systems at the same time. NUMA (Non-Uniform Memory Access Architecture, NUMA) architecture is a relatively mature multi-core processor architecture solution, it divides server hardware resources into...

AI Technical Summary

Problems solved by technology

[0007] In view of the above-mentioned defects in the prior art, the technical problem to be solved by the present invention is to provide a method based on the affinity between the current virtual machine network card cache and the NUMA node, and analyze the physical processor (Physical CPU, PCPU) load situation for predictive scheduling. The method, which overcomes the deficiency that the virtual machine VCPU cannot obtain enough processor resources to process network data packets, and effectively improves the processing efficiency of network data packets

Images