BGP parallel computing model based on the iteration tree

Abstract

Scalable architecture has become the main developing direction of core router in future Internet, and the major technical challenge faced by scalable router is the parallel computing technology in its control plane. This paper conducts an in-depth study of the parallel computing model of border gateway protocol (BGP) in scalable router. Based on hierarchical iteration tree, this paper gives a reasonable improvement of arborous BGP division scheme, which has the advantages of balanced task division and low communication cost. Detailed performance analysis is done and is compared with neighbor-based division scheme considering balance in the initial partition. At last, three implementations of the two schemes are compared and evaluated by using actual routing table data gained from Route Views. The conclusion is that, the developed division scheme based on iterative tree has a comprehensive advantage, which can provide workable models and methods to the expansion of control plane in the future scalable router system.