On Mar 29, 2011, at 10:59 AM, Lee, Yiu wrote: > Hi Mark, > > I fully agree with your comment about the L2 encap. However, gi-ds-lite > somehow allows gre and mpls-vpn label for the NAT binding. I guess the WG > should decide what to include and what not in this draft. For example: Jim > suggested to include the plain mpls label.
Ultimately, you can use anything that correlates directly back to the other side of the tunnel where addresses are being assigned. TE Tunnel makes perfect sense, as does a VPN label, VLAN, or anything else at your disposal. This begs the question though, if you aren't using IPv4 and IPv6 at the same time, where is the "dual-stack" part of "dual-stack-lite" in it? The heart of the "lite" part of "dual-stack-lite" is capitalizing on traffic separation that the NAT is cognizant of, allowing overlapping of the private IPv4 space and potentially eliminating the NAT on the other side of the tunnel. That doesn't require IPv6, and is perfectly fine to define, but if we're going to do that I think we need to consider it more generally and it may end up as part of Behave as much as it would Softwires. - Mark > > /Yiu > > From: Mark Townsley <[email protected]> > Date: Tue, 29 Mar 2011 10:52:55 +0200 > To: Microsoft Office User <[email protected]> > Cc: Frank Brockners <[email protected]>, "[email protected]" > <[email protected]> > Subject: Re: [Softwires] draft-ietf-softwire-gateway-init-ds-list-02 > > So, the "L2-Aware" NAT, or "ds-extra-lite" in the documents below is a CGN > which uses a VLAN, MPLS Label, PPP session, Ethernet MAC address, etc... I > think we should continue to describe this generally rather than try and list > every single possible L2 construct than can carry IPv4 and use it to plug > into a NAT binding.
_______________________________________________ Softwires mailing list [email protected] https://www.ietf.org/mailman/listinfo/softwires
