BGP-based inter-domain traffic engineering in BGP/MPLS VPNs

In BGP/MPLS VPNs, traffic engineering (TE) achieved by using multiprotocol label switching (MPLS) was predominantly limited to ultra-domain and single administrative domain. However with the rapid expansion of enterprise scale, a virtual private network (VPN) has spanned a large number of administrative domains, so effective management of inter-domain traffic is urgently demanded. This paper presents the simplest step called as BGP-based inter-domain TE. The method is on the basis of various border gateway protocol (BGP) attributes, BGP policy and autonomous system (AS) relationship. To implement TE, the method applies diverse routing policies by configured BGP attributes to control the outgoing and incoming traffic in an AS. We ensure that the best route in an AS depends on the routes coming from its neighboring ASes that apply export policies, as well as the import policies of the AS. Simulation shows that our approach can effectively implement inter-domain TE in BGP/MPLS VPNs.