Real-time dynamic decomposition storage of routing tables

Abstract

The decomposition storage of routing tables can effectively alleviate the storage problem caused by the Internet routing expansion. However, the existing researches based on certain routing tables cannot apply to unknown ones or support dynamic changes of routing tables. This paper presents a real-time dynamic decomposition storage model, named RDDS, which takes the IP prefix with 8 bits length mask as a pocket prefix and distributes the routing entries (REs) to different line cards (LCs) according to different pocket prefixes. In the light of the mapping relationship between the destination IP addresses of packets and the pocket prefixes, the forwarding engine determines the host LC of one IP packet and complete forwarding in the LC. RDDS needs a mapping table to locate the host LC in the process of packets forwarding, which only introduces very little logical processing. Therefore, RDDS hardly affect the overall performance of packets forwarding in routers. Experimental results show that RDDS achieves real-time part-storage and the load balancing of REs in LCs. Considering the frequent update of routing tables, RDDS is more significant than other storage models in real environments.