Re: IPv6 /64 links (was Re: ipv6 book recommendations?)
Owen DeLong wrote privately to me, but as I think I need public responses, I'm Ccing to nanog fairly quoting part of his response:
Moreover, it is easy to have a transport protocol with 32bit or 48bit port numbers with the end to end fashion only by modifying end part of the Internet.
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
That is a fairy tale once believed by so many infants who thought dual stack were enough. They still believed the fairy tale even when they designed automated tunneling. But, as most of them have grown up a little not to believe fairly tales, they are trying other possibilities. However, so far, they are not so successful. Masataka Ohta PS Rest of his response is omitted, because I think it is not worth quoting.
On Fri, Jun 22, 2012 at 8:46 AM, Masataka Ohta < mohta@necom830.hpcl.titech.ac.jp> wrote: <MAJOR SNIP>
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
That is a fairy tale once believed by so many infants who thought dual stack were enough.
Just injecting a real-world comment/observation here: I am sitting at home, on my natively IPv6 connected, Comcast-provided residential service. My phone is sitting next to me, connected to VZW's IPv6-capable LTE network. When I go to one of my client sites, they get IPv6 through a HE.net tunnel. Another client site uses AT&T and/or CenturyLink for IPv6 connectivity. *... the list goes on ...* In all cases, IPv6 is alive and well for me. More importantly (even though the last-mile is not ubiquitously IPv6-enabled in all service regions) those five providers have backbones that are 100% up and running, native IPv6 all over the place. So what is the fairy tale?? Am I saying we are all done, and that IPv6 is fully deployed? Of course not, lots of work to do in the enterprise and last-mile areas ... but progress has been noticeable and is accelerating. /TJ
TJ wrote:
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
Am I saying we are all done, and that IPv6 is fully deployed? Of course not, lots of work to do in the enterprise and last-mile areas ... but progress has been noticeable and is accelerating.
Owen tried to deny my point that: : Moreover, it is easy to have a transport protocol with : 32bit or 48bit port numbers with the end to end fashion : only by modifying end part of the Internet. As "the enterprise and last-mile areas" do not need to be modified to accommodate a new transport protocol, they belong to the center part. Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems. Masataka Ohta
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Masataka Ohta
Those problems are getting solved more and more every day. The rate of IPv6 deployment is rapidly accelerating at this point. QED, your statement does not stand up to current events. Owen
Owen DeLong wrote:
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Those problems are getting solved more and more every day.
The rate of IPv6 deployment is rapidly accelerating at this point.
Remember that you wrote:
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
What do you mean something 99% complete is rapidly accelerating? Is it a theory for time traveling? Masataka Ohta
On Jun 22, 2012, at 6:15 PM, Masataka Ohta wrote:
Owen DeLong wrote:
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Those problems are getting solved more and more every day.
The rate of IPv6 deployment is rapidly accelerating at this point.
Remember that you wrote:
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
What do you mean something 99% complete is rapidly accelerating?
Is it a theory for time traveling?
You redefined center. My definition of center when I claimed 99% was the major backbones and core routers. That is the CENTER of the internet. Different definition of center (yours) where the center includes everything except the edge-most hosts, different metrics for completion and challenges. Owen
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point.
What do you mean something 99% complete is rapidly accelerating?
Is it a theory for time traveling?
Rate of deployment is more inclusive than just the 'center', that would be my guess. Are we really taking this already nearly-pointless conversation to an all new low and arguing semantics? Clearly some of us disagree with each other, perhaps we just hold our tongues (& fingers) and let the real world decide?? /TJ
On 06/22/2012 08:35 PM, TJ wrote:
The center part of the internet is the easiest part of modification for IPv6 and is probably somewhere near 99% complete at this point. What do you mean something 99% complete is rapidly accelerating?
Is it a theory for time traveling? Rate of deployment is more inclusive than just the 'center', that would be my guess.
Are we really taking this already nearly-pointless conversation to an all new low and arguing semantics?
Clearly some of us disagree with each other, perhaps we just hold our tongues (& fingers) and let the real world decide??
/TJ There might be good money in a PPV "cagematch"-style event.
(2012/06/23 10:35), TJ wrote:
Rate of deployment is more inclusive than just the 'center', that would be my guess.
But, the context, as you can see, is this: :> Even though it may be easy to make end systems and local :> LANs v6 capable, rest, the center part, of the Internet :> keep causing problems. :> :> Masataka Ohta : : Those problems are getting solved more and more every day. that the problems are caused by the center part. Masataka Ohta
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Really? My impression is that it's very much the edge that's hard - CE routers, and in particular cheap, nasty, residential DSL and cable CE routers. Lots of existing kit out there that can't do v6, and the business case for a fork-lift upgrade just doesn't stack up. It's a cost issue, though, not a technology one - it's perfectly possible to deliver v6 over these technologies. Tunnelling, while not ideal, is certainly a workable stop-gap, and I'm *very* happy to have real, globally uniquely addressed end-to-end Internet in my house again as a result. Systems can be a problem too - both in convincing IS people to change things, in getting the budget for changes, and in finding all the dark places hidden in the organisation where v4 assumptions are made. But in the Internet core? I don't see any huge obstacles at $ISP_DAYJOB, with any of the people I know in the industry, or with the ISPs I do business with. For co-lo, VPS, leased lines, real Ethernet tails, v6 connectivity is being delivered and working fine today. Regards, Tim.
(2012/06/25 18:00), Tim Franklin wrote:
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Really? My impression is that it's very much the edge that's hard - CE routers, and in particular cheap, nasty, residential DSL and cable CE routers.
Are you saying they are "end systems and local LANs"? Masataka Ohta
On Jun 25, 2012, at 7:09 AM, Masataka Ohta wrote:
(2012/06/25 18:00), Tim Franklin wrote:
Even though it may be easy to make end systems and local LANs v6 capable, rest, the center part, of the Internet keep causing problems.
Really? My impression is that it's very much the edge that's hard - CE routers, and in particular cheap, nasty, residential DSL and cable CE routers.
Are you saying they are "end systems and local LANs"?
Masataka Ohta
Yes... Most people think of everything off the ISP network as "Edge". So from the CPE outward, is the Edge to most people's thinking. Your definition of center vs. edge is misplaced compared to everyone else. This is what created the confusion between us when I first said 99% of the center was already IPv6 -- If you don't count CPE outwards as part of the center, then that is a valid statement. If you count the CPE, etc. as center and only count the very end host as edge, then, my other statement (that IPv6 deployment in this area is accelerating) is true. Owen
participants (5)
-
Astrodog -
Masataka Ohta -
Owen DeLong -
Tim Franklin -
TJ