A protocol-independent forwarding network fast fault detection and traffic protection switching method

Abstract

The invention discloses a protocol-independent forwarding network fast fault detection and flow protection switching method. The method is as follows: 1) the controller notifies each pair of switches S1 and S2 to start a timer respectively; 2) when the sending timer on S1 expires, a connectivity check message is generated and sent to S2; S2 sends the first-class metadata Unit FM_CC is set to r; 3) When the check timer on S2 expires, check the FM_CC value, if it is r, set it to b; if FM_CC is b, then generate a fault notification message and send it to the The switch that responds to the event notifies S1 of the fault at the same time; 4) S1 sends the fault notification to the switch that needs to respond to the fault event; 5) When the source node sends data to the destination node through the traffic switching node, the traffic switching The node checks the availability of the primary path, and if available, transmits through the primary path, otherwise selects a backup path for transmission.

Description

technical field [0001] The invention relates to the technical field of the Internet, in particular to a fast fault detection and flow protection switching method in a protocol-independent forwarding (POF)-based software-defined network (SDN). Background technique [0002] Protocol Independent Forwarding (POF) is an extension to the software-defined networking (SDN) southbound interface protocol OpenFlow. POF provides the ability for the SDN controller to directly specify the offset and length of the rule matching field and the instruction operation field, and defines field modification, insertion, deletion, mathematical and logical operations, checksum calculation, instruction jump and other instructions for Various operations are performed on the message, and new features such as Flow Metadata and instruction modification are added. In the POF network, the data plane does not need to have the knowledge of specific network protocols, and only needs to follow the rules issue...

AI Technical Summary

Problems solved by technology

[0007] ·It can only support existing protocols such as Ethernet, IP, MPLS, etc., and is limited by OpenFlow's support range of protocol types, and cannot flexibly support new protocols;

[0008] Need to implement a hardware message generation engine, which increases the complexity and cost of the switch, and can only generate messages with a relatively fixed format, which has poor flexibility;

[0009] It is necessary to establish a BFD session between the entry node and the exit node of the flow. In a large-scale network, the number of flows is large, the entry node and the exit node are different, the number of BFD sessions is large, and the network bandwidth is wasted;

[0010] This mechanism needs to consume OpenFlow group resources

In hardware switches, group resources are limited and generally need to be shared among groups of different types (Fast Failover, All, Select, Indirect). Using too many group resources for traffic switching will affect multicast and load sharing and other network functions

Images