Cloud computing network flow robust forwarding method based on a gateway cluster

Abstract

The invention provides a cloud computing network flow robust forwarding method based on a gateway cluster. The method comprises a gateway cluster initialization stage, a flow forwarding and table item issuing stage and a multi-IP migration scheduling stage. According to the method, in order to improve the robustness of cloud network flow forwarding, for burst flow, a gateway cluster is introduced to jointly bear the flow, and a mapping algorithm of a virtual private cloud and the gateway cluster is designed to achieve load balancing of the cluster. For an abnormal event, the method provides a multi-IP migration scheme to realize rapid fault recovery. The gateway provided by the invention is realized by adopting a fast data path, so that the forwarding performance of the gateway can be effectively improved, and the transmission delay is reduced. The method disclosed by the invention can also be used for deploying network middleware/network functions so as to improve the robustness of the cloud computing network, and has a very wide application prospect.

Description

technical field [0001] The invention belongs to the field of cloud computing networks, in particular to a gateway cluster-based robust forwarding method for cloud computing network traffic. Background technique [0002] As more and more distributed applications are deployed on the cloud, the network communication between virtual machines (Virtual Machine, VM) has become the main load of the cloud computing network. Therefore, cloud networking must support high-speed, reliable inter-VM communication. In the current mainstream cloud platform, network traffic is first forwarded to the default gateway, and the controller only pre-installs a default entry pointing to the gateway on the virtual switch (Virtual Switch, vSwitch) of each computing node. Specifically, when two VMs deployed on different compute nodes communicate for the first time, the vSwitch on the source compute node forwards the initial traffic to the specified default gateway. The gateway then forwards the recei...

AI Technical Summary

Problems solved by technology

Since the gateway model only assigns one primary gateway to each computing node domain, when the computing node domain encounters burst traffic, the corresponding primary gateway is easily overloaded, resulting in severe network congestion in the area

(2) Abnormal events

When the main gateway encounters a failure event, the recovery delay of switching the backup gateway is usually long, which will cause a long-term network interruption, seriously reduce the network quality and affect the experience of tenants

Images