Re: [94attendees] [homenet] IPv6 Prefix delegation on IETF network, please ?

[Alexandre Petrescu <[e-mail]>]

Hi,

Le 04/11/2015 06:16, Mark Andrews a écrit :
>
> In message <[e-mail]>,
>   Gabriel Kerneis writes:
>> On Tue, Nov 3, 2015 at 10:45 AM, Michael Richardson <[e-mail]>
>> wrote:
>>
>>> Juliusz Chroboczek <[e-mail]> wrote:
>>>> However, a number of ISPs (including the IETF, it seems) do not allow
>>>> prefix delegation.
>>>
>>> No, that's not correct.
>>
>>
>> The "including the IETF" aside might not be correct, but the sentence as a
>> whole is correct.
>>
>>
>>>> Do we ignore the problem
>>>
>>> It's not a problem.
>>>
>>
>> Having dealt with several (French and British) ISPs providing modems with
>> crappy firmwares, let me assure you that this is a very real problem.
>>
>> Gabriel
>
> Different problems.
>
> Getting the IETF's DHCPv6 servers and routers configured to serve
> and support PD requests is a very different problem to getting CPE
> devices to properly support PD.
>
> Getting DHCPv6 servers to respond to PD requests is easy.  Integrating
> that with the routers is harder.  This is having the IETF simulate
> being a ISP with wired home customers (assuming a wired connection)
> compared to being a hotspot provider which it currently is.  If we
> want to do it over wireless as well then the IETF needs to simulate
> being a WISP.
>
> Now doing a demo network w/o integrating the DHCPv6 servers and the
> routers is also possible but requires a different planning.

This sounds like a sensible option.

I wonder whether the existing DHCPv4 deployment at IETF uses DHCPv4 
Relays or not?

I offer as a tester for next IETF.

Testing is very easy.  Download some DHCPv6-PD capable client and run 
it.  It will not install anything on the computer, just show whether or 
not it receives a prefix.

Alex

>
> Providing CTMS and similar is yet again a different problem.  We
> have the vendors in the IETF that can supply all the gear with
> suitable planning.  They have brought this sort of gear for
> Bit-and-Bites in the past.
>
> So what is the scope of the request and what needs to be simulated?
>
> Mark
>