Hi Toni, I can't condense what I wrote without making a mess of it, but here is what my original message concerns.
This is to let you know whether you might be interested in spending 20 minutes or so reading and thinking about the message itself: http://www.ietf.org/mail-archive/web/rrg/current/msg07587.html Please don't debate the following text, since it is inadequate on its own. I haven't yet read Wes' response. Wes George wrote to challenge my assertion that there's no tearing hurry regarding scalable routing. He wrote that he was concerned about his company - Sprint - having to pay $$$ and go to much trouble before long to upgrade their routers to cope with the growth in the "DFZ routing table" - shorthand for the number of routes advertised in the interdomain routing system. The problem growth is with IPv4, but the same thing could in principle happen with IPv6. He also wrote that he saw no option but to use IPv6 for future large mobile phone networks. He also wrote (in my interpretation) that he was hopeful or expectant that instead of ISPs having to keep spending money to keep the Internet running, in part due to scaling difficulties, that application and stack programmers would work to make hosts operate in new, more scalable, ways. This would alleviate the huge expenses ISPs face for new routers or whatever every few years and so reduce what they need to charge customers. I stated why I believe this hope to be completely unrealistic. I replied by narrowly qualifying the scope of my "no tearing hurry" assertion and agreeing with him about there being an urgent need (in the context of the next 5 to 10 years) for scalable routing solutions, in particular regarding mobility. Subject to an assumption, which I am not at all sure is correct, I agreed with him that IPv6 would soon be widely used in 3G mobile "phone" (and whatever else these things do) networks. The assumption is that operators will find it easier or better to sell a service with native IPv6 connectivity, presumably with a tunnel to an IPv4 NAT box, to do dual stack but with the IPv4 connection behind NAT, than to provide a purely behind NAT IPv4 service. I am agreeing with his statement that there's not enough IPv4 space to support the continuing expansion of 3G etc. mobile networks in a manner where each device gets it own global unicast IP address. That said, I wonder whether 3G networks provide global unicast addresses in general. Are many or most of them providing NATed IPv4 addresses at present? I discussed my understanding of the mobility options for IPv6 - including those which are proposed but yet to be developed: 1 - MIPv6 - which Wes mentioned he did not favor 2 - Loc/ID Separation, such as ILNP 3 - LISP-mn 4 - TTR Mobility (my proposed architecture) with Ivip or LISP I discussed why I believe 2 and 3 are completely inadequate and tried to interest him in 4. I also discussed the potential growth in the number of IPv6 DFZ routes, and queried whether the limit for routers was the total number of routes for IPv6 separately from IPv4, or whether the sum of the two was more important. - Robin _______________________________________________ rrg mailing list [email protected] http://www.irtf.org/mailman/listinfo/rrg
