BGP based shortcut virtual channels for transit IP traffic over ATM networks

Abstract

This paper suggests an efficient mechanism to establish and utilize shortcut virtual channels for transit traffic over an ATM based IP network which uses BGP as an inter-domain routing protocol. Although in this case, "MPOA" or "CLIP with NHRP" might be alternative solutions, they incur not only inevitable hop-by-hop routing for the first packet, but also multiple communications between NHC and NHSs. What is worse, the NHRP based schemes can not efficiently cope with inter-domain routing instability that can result in temporal inconsistency between existing shortcut VCs and authoritative paths. BGP, de facto inter-domain routing protocol in the Internet, can directly identify an egress router for transit traffic crossing the network. Thus, when ingress routers know the ATM address of egress routers, virtual channels for transit traffic can be immediately and very easily set up. Fortunately, by the virtue of the extensible design of BGP we can extend it to transport egress router's ATM address without any distortion of the basic mechanism of the protocol. Moreover because the resulting scheme inherits several merits of BCP and ATM shortcut VCs, it can easily enforce administrative policies to the network and dynamically adapt itself to the routing instabilities, even render the existing routers and network architecture.