[v6ops] Re: 464XLAT-only networks

[Lorenzo Colitti <[e-mail]>]

I didn't say it would break. I said it would add extra latency (potentially
double the latency) to DNS lookups.

I appreciate the "what can we do to drive adoption" sentiment, but I think
deploying IPv6 without global connectivity is generally*worse* than staying
IPv4-only. This is because it requires apps to code workarounds such as NAT
traversal that are not needed on the vast majority of IPv6 networks. It's
better to deploy IPv4 only and let those hacks die with IPv4.


On Sat, 26 Oct 2024, 08:23 Soni "It/Its" L., <[e-mail]> wrote:

> if issuing IPv6 queries breaks your network, you have bigger issues...
>
> anyway, we have a new development related to this thread, that we kinda
> almost forgot about but got reminded earlier:
>
> we've been working on IPv6-only server software that we plan to tell the
> customer to use a SIIT-DC for. some of those customers may well be
> IPv4-only (if their ISP/hosting provider/etc doesn't support IPv6, which
> is, sadly, still common). since a SIIT-DC is more or less a NAT64 with a
> different configuration, this effectively means they'll be running a
> NAT64-only network. or, well, NAT64+native IPv4, which is a weird edge-case
> from running the software on the NAT64 box itself.
>
> so, how is that for an use-case for a "464XLAT-only" network? (is it still
> a network if there's only one host?)
>
> (we're gonna do this either way. the entire point of making our software
> IPv6-only is to push our customers to adopt IPv6 or use an appropriate
> workaround. what are y'all doing to drive IPv6 adoption? ^^)
>
> On 2024-10-25 19:53, Lorenzo Colitti wrote:
>
> getaddrinfo will still issue IPv6 queries because it decides whether to do
> so based on addresses configured, not routes configured.
>
> On Fri, 25 Oct 2024, 19:57 Soni "It/Its" L., <[e-mail]>
> wrote:
>
>> with an explicit route for the nat64 prefix?
>>
>> On 2024-10-25 07:42, Lorenzo Colitti wrote:
>>
>> If it doesn't have a default route, how is it going to reach the nat64
>> prefix?
>>
>> On Fri, 25 Oct 2024, 19:30 Soni "It/Its" L., <[e-mail]>
>> wrote:
>>
>>> if it doesn't have GUAs or a default route (is a default route required
>>> in SLAAC? can't remember), why would it do any of that?
>>>
>>> On 2024-10-24 22:27, Lorenzo Colitti wrote:
>>>
>>> The problem with such a network is that the host doesn't know that the
>>> IPv6 internet is reachable. Because IPv6 stacks generally prefer IPv6 over
>>> IPv6, that typically results in high latency and failures. Here are the
>>> first three problems I can think of:
>>>
>>>    - Client code that cannot fall back from IPv6 to IPv4 won't work
>>>    - getaddrinfo will always issue AAAA lookups as well as A lookups,
>>>    slowing down DNS lookups
>>>    - HE implementations that bias towards IPv6 (which I think is
>>>    suggested or required by the RFC) will encounter latency penalties.
>>>
>>> There are probably more.
>>>
>>> On Thu, Oct 24, 2024 at 7:33 PM Soni L. <[e-mail]> wrote:
>>>
>>>> we would like to propose a bastardization of IPv6-only networks and
>>>> 464XLAT: the 464XLAT-only network.
>>>>
>>>> it's a network that doesn't provide access to the IPv6 internet, but
>>>> does provide access to the IPv4 internet via 464XLAT.
>>>>
>>>> thoughts?
>>>> _______________________________________________
>>>> v6ops mailing list -- [e-mail]
>>>> To unsubscribe send an email to [e-mail]
>>>>
>>>
>>> --
>>> plural system (tend to say 'we'), it/she/they, it instead of you
>>>
>>>
>> --
>> plural system (tend to say 'we'), it/she/they, it instead of you
>>
>>
> --
> plural system (tend to say 'we'), it/she/they, it instead of you
>
>