A data packet classification method

Abstract

The invention discloses a data packet classification method, which includes CPU preprocessing rule set and GPU execution data packet classification. run. Preprocessing operations are done on the CPU. The packet classification operation is the core part of the algorithm, which classifies the input data packets and outputs the rule number with the highest priority matched by the data packets. Packet classification operations are done on the GPU. The invention makes full use of the hardware characteristics and advantages of the GPU, and has greater performance advantages than conventional data packet classification algorithms.

Description

technical field

[0001] The invention relates to a method for classifying data packets. Background technique

[0002] In modern routers, switches and firewalls, in order to achieve priority routing or data packet filtering, data packets transmitted on the network need to be classified according to certain rules according to the header information, and different operations are performed. Such a problem is known as a packet classification problem.

[0003] In the network layered model, the data to be transmitted is encapsulated sequentially by the packet headers of each layer protocol, and the data packet header of each layer contains several fields, which respectively represent the characteristic data of the layer protocol. A rule set contains N rules, and each rule consists of three parts: (1) An expression related to the domain, the expression is generally a d-tuple (f 1 ,f 2 ,..., f d ), where f i It may be a prefix (the IP address in the data packet needs to be matche...

Images