[arin-ppml] Draft Policy ARIN-2019-19: Require IPv6 Before Receiving Section 8 IPv4 Transfers

Owen DeLong owen at delong.com
Mon Nov 18 04:25:39 EST 2019



> On Nov 17, 2019, at 19:00 , Michel Py <michel at arneill-py.sacramento.ca.us> wrote:
> 
>> Owen DeLong wrote :
>> Well… It could also connect you to the growing fraction of the internet and
>> provide better performance, address transparency, and a few other benefits.
> 
> I'm shocked you still are trying to play this with me. I was on the 6bone, I had IPv6 on a Cisco 2500.
> There are ZERO benefits for me deploying today, and at 2% growth rate per year and slowing down, I have every incentive to stay with the 90% of my ecosystem that does not do anything.

All of the above stated benefits are true and well documented. The fact that you don’t consider them sufficiently important to make it worth deploying IPv6 in your environment is a perfectly valid decision and I’m not challenging you on it.

I’m not playing anything.

I don’t quite buy your idea of 2% per year, nor do I agree that the growth rate is slowing. I think you are using some pretty skewed statistics, or, you’re looking only at the enterprise rates. Even in the enterprise only statistics, I think that the current slow down isn’t a trend so much as a blip.

I’m confused by your last statement. If you dual-stack, you haven’t departed the 90% of your ecosystem that does nothing, but you have gained the 10% that works better.

> I don't have to do anything this year. I don't have anything to do for the next 5 years. IPv6 is not on my agenda, and it is not on any of my competitor's either.

Unless your environment is really really small, it’d be pretty hard to do anything this year anyway at this point.

I think expecting to get away without deploying IPv6 for 5 years is optimistic on your part.

> I do not care connecting the carrier-controlled cell phones. Heck, if Youtube and Facebook were to go IPv6-only, that would be a blessing. This is not what we pay employees to do at the office.

What if _ALL_ of Google went IPv6 only?

> As of the getting rid of NAT, you track record is not good. Look at slide 57. IPv6 is NAT-free ?
> https://www.slideshare.net/RNIDS/ipv6-transition-and-coexistance-jordi-palet

Slide 57 is about transition and coexistence. Once we can finally start taking down IPv4 on the backbone, that entire slide mostly goes away.

Unfortunately, people with your attitude are now the major block to progressing in that direction, so now you’re more likely to attract negative attention with that attitude than in the past.

> Not anyone's slides eihter, Jordi's own. I missed the other benefits; instead of NAT44, now I have NINE different competing schemes to deal with ?

Nope… If you have enough IPv4 to do what you do how you do it now, then you have no need for anything on that slide… Dual stack and move on.

That slide is all about mechanisms designed to help you cope with an IPv6 deployment that can’t get enough IPv4 addresses to do dual-stack and NAT44 in a conventional manner.

It’s kind of a shame Jordi didn’t put a Dual Stack column on that slide as it would have been pretty much green for the entire column:
Tunnel/Translation: N/A
Dual-stack LAN: YES
IPv4 Multicast: YES
Access Network: YES
Overhead: -
Impact on IPv6 addressing plan: NO
CPE Update: —It Depends— Not for any CPE with IPv6 support
NAT44/NAPT: CPE
46/64 Translation: -
Translation at ISP: -
Scalability: High
Performance: High
ALGs: NO
Any Protocol: YES
Sharing IPv4 Ports: optional
IPv6 Aggregation: optional
IPv4 Mesh: YES
IPv6 Mesh: YES
Impacts on logging: NO
HA simplicity: HIGH
DPI simplicity: HIGH
Cellular Support: YES (had this on VZW for a while)
Support in CPEs: YES

> 
>> Deploying it sooner rather than later also provides the benefit of being
>> able to do it in a controlled manner on a more relaxed timeline rather than
>> in a rush when somebody above you finally figures out that they need it yesterday.
> 
> Fortunately, I am high enough in the food chain that I get to make this decision.

I never denied this…

> Zero of my customers have IPv6.

Are you sure?

> Zero of my suppliers have IPv6.

Again… Are you sure?

> Zero of my upstreams have IPv6.

That’s kind of surprising today.

> Zero of my competitors have IPv6.

Since I don’t know which business you’re in, I have to take your word on this.

> Zero of my C-level execs have ever heard of IPv6.

That’s not hard to fix.

> Zero of my board members have ever heard of IPv6.

Also an easy fix.

> Zero of all of these have IPv6 on their agenda.

That might be slightly harder to fix, but not insurmountable.

> I run the show.

Yeah, that’s tragic, but I’m not in a position to fix it. :p
(hey, you kind of asked for this one).

> The failure of IPv6 was to ignore the notion of ecosystem. The IPv6 ecosystem, after 20 years, is still a niche market compared to the IPv4 ecosystem.

I’m not sure what you think could have been done differently here. Most of the time people say this, it devolves into a variation of them thinking there’s some practical way to get a host that only has 32-bit address fields to talk to hosts with 128-bit addresses without modifying the software on the 32-bit only host.

Once they realize that doesn’t work, it’s usually an “oh yeah” moment.

In your case, I’m pretty sure you’re smarter than that, so seriously, what do you think could have somehow better addressed the ecosystem?

>> As much as I am in favor of IPv6 deployment and use IPv6 every day and look forward eagerly to the
>> day when IPv4 is constrained to unimportant islands of in the dark corners of the internet, the
>> reality is that this policy does nothing at all to move us closer to that day. Instead, it helps
>> some of the people promoting IPv6 feel like they’ve accomplished something while only serving to
>> annoy those that are resistant to IPv6 deployment. I’m pro v6 and I oppose the policy as written.
> 
> I am glad you see that. There is nothing wrong being pro-v6, but the line you should not cross (nor should ARIN) is getting anti-v4.

Oh, I am anti-v4. As I said, I look forward to the day we can stop routing IPv4 on more and more of the backbone (to the extent that term even retains meaning).

I’m not anti-your-ability-to-run-IPv4-as-you-become-an-island. I’m all for your freedom to choose the wrong path and suffer the consequences.

I agree that ARIN shouldn’t be developing punitive policies towards IPv4 (I’ve agreed with that from the beginning of this thread).

> The FUD that the IPv4 Internet will stop because of shortages has come, and gone. Any subsequent efforts to sunset IPv4 will be matched in efforts to disable IPv6.

Interesting and all to clever hackery has kept IPv4 on a form of life support for admittedly far longer than any rational person thought possible. I give the IPv4 crew tremendous credit for their clever procrastinations.

Nonetheless, there are 7 billion people on Earth (and growing) and 3.2 billion IPv4 unicast addresses. Shortage has already come. People have chosen to live in a degraded internet that is becoming ever more expensive to support in order to avoid the (comparatively trivial) effort to deploy IPv6. Largely because of FUD in a lot of cases.

Eventually, providers will simply start charging extra for customers that want them to carry IPv4.

Owen



More information about the ARIN-PPML mailing list