[v6ops] Re: Dynamic addresses

[Daryll Swer <[e-mail]>]

Brian Candler

> If a customer pays for static IPv4 (and the ISP offers it), they get
static IPv4. If they pay for static IPv6 (and the ISP offers it), they get
static IPv6.

Not quite. MOST ISPs have a fantastic AAA/Billing (and manual/automation
system) for IPv4-only use-case. When a customer asks for FREE ia_pd /56 on
residential broadband that's static or even ask to pay for it, the answer
is usually "We don't support static IPv6", and even if they did support it,
they only give a single /64 because apparently "We are exhausting our IPv6
space if we give a /56 to each user" is a thing.

And most of them, have horrible IPv6 configuration/deployments internally,
that, moving from one BNG to another = IPv6 dies along with the first BNG,
but IPv4 works fine for dynamic/static.

> What you *can* do is document what the problems and consequences are of
dynamic IPv6 allocation to end users, and document the solutions or
workarounds - one of which is that the problems go away if the ISP provides
a static IPv6 allocation.

Isn't that precisely what BCOP-690 has done?

+1 to David's points. 100% agree.

Dynamic IPv6 MUST be discouraged.


*--*
Best Regards
Daryll Swer
Website: daryllswer.com
<https://mailtrack.io/l/fef49b73bf48e2c048a73e7d9543e5edbd452cf8?url=https%3A%2F%2Fwww.daryllswer.com&u=2153471&signature=a84d97ca666c92a8>


On Mon, 12 Aug 2024 at 12:51, David Farmer <[e-mail]> wrote:

>
> On Mon, Aug 12, 2024 at 01:40 Brian Candler <[e-mail]> wrote:
>
>> On 12/08/2024 10:06, Daryll Swer wrote:
>>
>> IPv4 doesn't have this issue. When a customer pays for a static *IPv4*,
>> that IP is permanently static even across BNGs groups (automation in their
>> backend would move the IPs around if required, god forbid, it's manually
>> done).
>>
>> But come to IPv6 and most (exception always exists) ISPs globally, refuse
>> to give a /56 ia_pd static free or paid. I'm personally really tired of
>> this attitude from ISPs, i.e. the IPv4-centric attitude.
>>
>> If a customer pays for static IPv4 (and the ISP offers it), they get
>> static IPv4. If they pay for static IPv6 (and the ISP offers it), they get
>> static IPv6.
>>
>> I still don't see anything specific to IPv6 here. It's just about product
>> offerings in different markets.
>>
>> In some marketplaces or countries or price brackets, ISPs don't offer
>> static IPv4; and in some they don't offer static IPv6. I don't believe you
>> can mandate a particular service to be available in a particular country by
>> RFC.  If you try to, your RFC will be ignored, and hence irrelevant.
>>
>> What you *can* do is document what the problems and consequences are of
>> dynamic IPv6 allocation to end users, and document the solutions or
>> workarounds - one of which is that the problems go away if the ISP provides
>> a static IPv6 allocation.
>>
>
> No, this problem isn’t unique to IPv6 and exists to some extent in IPv4 as
> well. However, most deployments of IPv4 today, whether static or dynamic,
> utilize NAT. Accordingly, the impact of address changes in most cases for
> IPv4 is quite different from IPv6. NAT isolates individual hosts from much
> of the consequence of address changes with IPv4.
>
> Conversely, with IPv6, those changes are propagated to individual hosts
> through SLAAC or DHCPv6. Therefore, the impact on individual hosts is quite
> different between IPv4 and IPv6, even if the underlying behavior of
> changing addresses from the ISP’s perspective is identical.
>
> It is the difference in architecture between IPv4 and IPv6 that
> exacerbates the issue of changing addresses, and makes it a much bigger
> issue for IPv6 than IPv4.
>
> Thanks
>
>
>
>
>