[Witarea] Re: [v6ops] How to make an elegant IPv4 outage

> On Jun 14, 2026, at 09:35, Phillip Hallam-Baker <[email protected]> wrote:
> 
> 
> 
> On Fri, Jun 12, 2026 at 9:53 PM Brian E Carpenter <[email protected] <mailto:[email protected]>> wrote:
>> Erik,
>> 
>> On 13-Jun-26 02:19, Erik Nygren wrote:
>> > + v6ops
>> > 
>> > On the topic of the original draft (https://datatracker.ietf.org/doc/html/draft-martin-retry-over-ipv6-01 <https://datatracker.ietf.org/doc/html/draft-martin-retry-over-ipv6-01>):
>> > 
>> > 1) I wonder if this is pointing towards a desire to either start sunset4 back up again, or to charter work within some existing WG (v6ops, 6man, others?) to pick back up on where sunset4 left off, specifically looking for technology solutions and operational recommendations to provide a path towards phasing out IPv4 in various environments.  
>> 
>> I really hope not. I think we have learnt that attacking currently working solutions is self-defeating, which is ultimately why sunset4 failed.
> 
> Exactly. The problem I have seen all along is that what the Internet users need is a mechanism that allows them to interact in a larger Internet without any impact on their user experience whatsover and this has been interpreted inside the IETF as 'force everyone to transition to IPv6 for everything, everywhere'.
> 
> 
> My experience deploying a high end prosumer network in the house tells me that IPv4 is never going to go away and it is futile to expect otherwise.
> 
> The border gateways have a capacity of ~1000 client devices, I am currently using 72 IP addresses after a purge but don't anticipate using more than 250 in the near future.
> 
> All the interior routing of addresses uses private addresses. I could wire a static IP address direct to an internal host but since I have to pay $20/mo for each block of 4 static IPv4 addresses, that would be a waste and limit internal administration. In 1995, a DNS server was a specific host, now it is one virtual machine floating around a cluster of Proxmox hosts.
> 
> So instead, I have rules that map split the ports out to the hosts that service them. And here we get to the part that makes transitioning to IPv6 on the Internal network highly unlikely: 10.x.0.y is so much easier to remember than anything in IPv6 space.
> 
> My internal network is segmented into a series of VLANs so that the IoT devices cannot touch the production network. Each VLAN has a separate prefix 10.x.*.*. If a device has IPv6 service, the IPv6 address will be in a /24 with the lower 32 bits being its net 10 suffix.
> 
> And this isn't just some scheme PHB thunk up, it is pretty much the way people deploy the hardware just as named.config.local is the place most people list out the zone files for their local domains.

And the security of those devices is questionable at best. Like you I don’t see it changing soon. When you are in the shop, there is no way to figure out if any device has IPv6 support, but I know they will want me to open a cloud account, to manage this device. Nothing can work locally (because they sell time on those device for scrapping the Internet for our AI overlords?).

> 
> The mental model I think appropriate is IPv4 is high level language and IPv6 if machine code.
> 
> Yes we are going to be using IPv6 in the Internet to come. But the reason we haven't run out of IPv4 addresses so far is that most of the devices are mobile clients and those work just fine on an IPv6 connection with a carrier grade NAT to talk to legacy IPv4 only sites.
> 
> I will be making my services available over IPv6 as soon as I can find an ISP that will give me the necessary static IP/64. But my expectation is that I won't ever need to interact with them directly as an administrator on the internal network, it will be purely limited to the network gateway interface.
> 
> The net is that while IPv6 will gain deployment, IPv4 will retain its mindshare and I do not expect that to change for at least another century.

Interesting,

Reading a few more comments on that subject.

It feels to me anyone should be able to request an IPv6 allocation to the local RIR AND then ask to have it routed by its provider. Until then, the provider will try to lock the customer, as they always have done.

Market forces at play here.

When I was in Fiji, like 20 years ago, I requested an IPv6 allocation, setup 2 servers in the USA with BGP access, and setup tunnels from 3 providers to these servers. I had IPv6 and BGP over tunnels, something none of the local providers wanted to give me.

-- 
Witarea mailing list -- [email protected]
To unsubscribe send an email to [email protected]