Forwarding task scheduling method and system for guaranteeing hierarchical time delay in virtual switch

Abstract

The invention provides a forwarding task scheduling method and system for guaranteeing hierarchical time delay in a virtual switch, and the method comprises the steps: 1, distributing the priority of forwarding task scheduling to each virtual machine in the same server, setting a threshold value for a speed limiting token bucket of each virtual machine according to the batch processing capability of the virtual switch in the server, and constructing a priority queue corresponding to each priority; 2, taking out a data set taking a virtual machine as a unit from the priority queue according to the priority, forwarding a batch of data packets by the virtual switch, judging whether the number of tokens in the data set is greater than the threshold value or not, if so, writing the data set back to the corresponding priority queue, and otherwise, putting the data set into a waiting queue; and 3, monitoring the number of tokens in each data set in the waiting queue, and migrating the data sets of which the number of tokens in the waiting queue is greater than the threshold value to the corresponding priority queues.

Description

technical field [0001] The invention relates to a task scheduling method, and provides a scheduling method and system for guaranteeing hierarchical time delay in a virtual switch. Background technique [0002] Cloud computing has become an indispensable part of the computer industry due to its high cost performance and flexibility. Thanks to virtualization technology, more and more enterprises deploy their own applications and services in the form of virtual machines (VMs) on cloud platforms. Many VMs on the same server implement network virtualization through a virtual switch (vSwitch process). A vSwitch process provides traffic forwarding and external network connection functions for so many VMs at the same time, which makes the vSwitch process a highly centralized virtualization component with serious resource competition, and puts forward higher requirements for improving the network delay of VMs and providing differentiated services. requirements. [0003] On a typic...

AI Technical Summary

Problems solved by technology

Therefore, in the software vSwitch, the main delay source of the VM is waiting for the PMD thread to complete the batch processing process of other VMs. If the traditional packet queuing method is simply used on the network card port, the delay cannot be guaranteed, and what is worse is , as the number of VM deployments increases, all VMs will face a sharp increase in latency without distinction, which requires a scheduling method suitable for the vSwitch operating mode to ensure differentiated latency for VMs

Images