Implementing Inter-domain Traffic Engineering with BGP in BGP/MPLS VPNs
在BGP／MPLS VPNs中用BGP实现域间流量工程

In BGP/MPLS VPNs, traffic engineering achieved by using MPLS was predominantly limited to intra-domain and single administrative domain. However with the rapid expansion of enterprise scale, a VPN has spanned a large number of administrative domains. Thus effective management of inter-domain traffic is urgently demanded. Presented the simplest method is on the basis of various BGP attributes, BGP import or export routing policy and AS relationship constructed according to bilateral economical agreements. To implement inter-domain traffic engineering, the best route in an AS depends on the routes coming from its neighboring ASes that apply export routing policies, as well as the import routing policies of the AS. The control of the outgoing traffic is often a requirement for providers that wish to optimize the distribution of their content. For this, they can rely on the LOCAL_-PREF attribute to control the routes that will be chosen for the packets that leave each BGP router of the provider. A customer AS serving a large number of individual users or small corporate networks will typically have a very asymmetric inter-domain traffic pattern with several times more incoming than outgoing traffic. These ASes typically need to optimize their incoming traffic only. In order to balancing incoming traffic, ASes are allowed announce their prefixes to a selected subset of providers instead of all providers. According to BGP export policies, an AS cannot act as a transitive AS for its two providers or peers. Simulation shows that our approach can effectively implement inter-domain traffic engineering in BGP/MPLS VPNs.