BGP/MPLS VPN Extension for IPv4/IPv6 Hybrid Network

[Michael Dell <mike.dell at dataconnection.com>]

> Defeng
> 
> I've read through your draft 'BGP/MPLS VPN Extension for IPv4/IPv6 Hybrid
> Network' and I had a question relating to BGP next hops.
> 
> In the case were the CE site is running IPv4 and the provider network is
> running an IPv6 backbone, there is obviously an issue relating to what BGP
> next hop the PE router should advertise.  The route will be advertised as
> a VPNv4 route, and MP-BGP requires that the next-hop be of the same
> address family, but the PE router doesn't have an IPv4 address.  Your
> draft suggests the only solution is to use NAT-PT on the PE routers so
> that they are assigned an IPv4 address reachable in the IPv6 network.
> This seems rather constrictive, as there are many possible reasons why the
> provider may choose not to run NAT-PT.  Having given this some thought, it
> would seem that there are two other possible solutions (although there may
> well be others):
> *	Require the PE router to have at least 1 IPv6 mapped IPv4 address
> which is reachable in the provider network.  It could then use the pure
> IPv4 address as the BGP next hop.  This would require the receiving PE to
> perform the reverse mapping from IPv4 to IPv6, however, which could cause
> interoperability issues.
> *	Perform direct address mapping of the CE routes from IPv4 to IPv6 so
> that the routes advertised in the provider network are VPNv6 rather than
> VPNv4.  This seems rather wasteful of resources on the routers, however.
> Had you considered and discounted these solutions, or is your intention in
> the draft merely to suggest one possible solution?
> 
> Regards
> 
> Mike Dell
> Networking Protocols Group
> Data Connection Ltd
> Tel: +44 20 8366 1177
> Fax: +44 20 8367 1039
> E-mail: mailto:mike.dell at dataconnection.com
> Web: http://www.dataconnection.com
>