Parallel route searching method and system for eliminating longest prefix match search

Abstract

The invention belonging to technical area for searching Internet IP address in high speed possesses characters: based on searching structure of dichotomy tree, prefix partition is carried out for prefix of all routes according to principle that node heights of current nodes are equal; then, searching for next jumping out port to forward information is carried out concurrently by using search framework in high commonality for some nonoverlapping prefix sets divided from the above said step. The searching system includes distributor, bank unit set, selector and mapper. The said distributor and selector are composed of field programmable gate array (FPGA). The mapper includes static random access memory (SRAM). The said bank unit set is composed of cascaded FPGA and SRAM. The invention also discloses a method for balancing prefix set. Experimental verifies that the invention increases searching speed.

Description

technical field [0001] The parallel routing search method and system for eliminating the longest prefix matching search belong to the technical field of high-speed IP address search in the Internet. Background technique [0002] Routers are responsible for routing and forwarding packets in the network. When forwarding a packet, the router needs to perform a route lookup in the routing table based on the destination IP address of each packet header, and find the correct next-hop exit of the packet. In high-performance routers, the interface rate is getting higher and higher. OC48 (2.5Gbps), OC192 (10Gbps), and even higher rate interfaces have been applied in high-performance routers. Under the OC192 interface rate, the router needs to forward 30 million packets per second (the length of the packet is calculated as 40 bytes), that is, it needs to perform 30 million route lookups per second. Fast route lookup technology becomes the key to improve router forwarding performanc...

AI Technical Summary

Problems solved by technology

[0003] The class-based address structure was adopted in the early days of the Internet. With the development of the Internet, the class-based address structure has two problems: 1. A large number of addresses are wasted

2. The scale of Internet global routing table grows exponentially

TCAM also has disadvantages: high cost, high power consumption, and complex update

The leaf pushing technology can reduce the complexity of routing lookup, but there are two problems: 1. It leads to a large amount of repeated routing information

The routing information of the intermediate node is spread to multiple leaf nodes. Initially, one route of the intermediate node is expanded into multiple routes, which causes a waste of storage space.

2. Complicated updates

Since it is assumed that routers are not connected to shared links, this division method is obviously not universal

Images