Multi-core network device message ordering method and system by interface

Abstract

The invention discloses an interface order-preserving method and system of messages of a multi-core network device. The interface order-preserving method comprises the following steps of: S1, after receiving current messages by CPUs of a multi-core CPU, obtaining import interface information of the current messages, carrying out import interface judgment on the current messages and grouping import interfaces of the messages; S2, sequencing the current messages by corresponding threads in corresponding import interface groups, processing the current messages; S3, after the current messages are processed, judging whether the former message of the same import interface is forwarded, if yes, forwarding the processed current messages, otherwise, and after waiting the former message is forwarded, forwarding the processed current messages. In addition, the invention also discloses an interface order-preserving system for realizing the interface order-preserving method. Through judging whether the former message of the same import interface is forwarded, the order preservation of the messages of the same import interface is realized, and the disorder when the messages are sent is prevented.

Description

technical field [0001] The invention relates to the technical field of network communication, in particular to a method and system for maintaining the sequence of multi-core network device messages according to interfaces. Background technique [0002] In a program, these independently running program fragments are called "threads" (Thread), and the concept of programming using it is called "multithreading". A common example of multithreading is user interfaces. With threads, the user can press a button and the program responds immediately, rather than having the user wait for the program to complete its current task before responding. Most of the current high-end network devices use a multi-core CPU hardware architecture to improve message processing and forwarding performance, and create corresponding threads for the multi-core CPUs to perform one-to-one binding with the multi-core CPUs. [0003] The application number is: 201010289667.X, and the title of the invention i...

AI Technical Summary

Problems solved by technology

[0003] The application number is: 201010289667.X, and the title of the invention is: A Chinese patent for a message forwarding method and system based on a multi-core architecture, which records a method that associates a hash queue (that is, the area divided by the hash table in the text) with threads The binding method realizes the lock-free design of shared resources, but when there are several packets of data streams on the network device, after hash calculation, they all fall into a hash queue, which will lead to the corresponding hash queue The thread of the thread is very busy, and the thread corresponding to other hash queues is in an idle state, so the usage efficiency of the multi-core CPU is low, resulting in a waste of resources

[0004] In order to improve the usage efficiency of the multi-core CPU in the network equipment, the RMI chip is used in the prior art to evenly distribute the received packets to each CPU. Since the packets are evenly distributed to each CPU, the packets have different sizes, so , the speed at which each CPU processes messages will be somewhat different. After the message processing is completed, there will be out-of-order phenomenon when the message is sent (there is a three-thousandth chance of out-of-order in the test), which is enough for general needs. , but for IPSEC packets and video packets, if out-of-order occurs, packet loss or video mosaic will occur, which will seriously hinder network communication

Images