Re: the race to the bottom problem

[Alexandre Petrescu <[e-mail]>]

Le 08/11/2020 à 02:56, Brian E Carpenter a écrit :
> On 08-Nov-20 14:35, Lorenzo Colitti wrote:
>> On Sun, 8 Nov 2020, 08:24 Nick Hilliard, <[e-mail]
>> <mailto:[e-mail]>> wrote:
>> 
>> this is the problem: I did read it, just like I've read through all
>> the other threads on this topic over the last several years. There
>> was a repeat of the usual speculation, generalisation and
>> echo-chamber mechanics which have characterised this discussion at
>> the ietf since more-or-less forever.
>> 
>> You said: "we’ve already seen clear pressure to race to the
>> bottom".
>> 
>> So I politely ask you again: please provide citations to data.
>> Then we can have a discussion.
>> 
>> 
>> Ok, I'll bite. Once. :-)
>> 
>> Here's some data:
>> 
>> https://www.ietf.org/proceedings/96/slides/slides-96-v6ops-4.pdf
>> 
>> says that 29% of survey responses says the LAN prefix size is /64.
>> This is despite pretty much every IETF and RIR document saying that
>> ISPs should assign more.
> 
> There's some 2010-vintage data at
> https://tools.ietf.org/html/rfc6036#section-2.5 suggesting that even
> then, there was a variety of practice including some /64-only
> operators. And: "Mobile operators offer /64 in accordance with 3GPP,
> but at least one would like to be allowed to offer /128 or /126."

I think that could be a good supposition to imagine that that (human) at 
operator might race to a bottom.

But there could be other reasons than racing to a bottom.

Initial IPv6 deployments at mobile operators where with ppp protocol, 
which needed only two addresses on a link, hence just a /127, or so. 
Then SLAAC on 3GPP links arrived, proposing that /64.  But some 
operators wished to _continue_ to use their /127 and not migrate up to 
/64, for backwards compatibility.

In the end, after some changes in the network, _some_ operators migrated 
all their connections to /64 and got rid of /127.  There are still some 
places that are compatible with /127 ppp IPv6 mobile connections, which 
is good, because it is simpler.

But moving uupper from /64 to /56, for example, does not happen.

Because, there is so SLAAC with a /56 either.  It is one reason.

And because there is no DHCPv6-PD on 3GPP modems either.  It is another 
reason.

Alex

> 
> Brian
> 
> -------------------------------------------------------------------- 
> IETF IPv6 working group mailing list [e-mail] Administrative
> Requests: https://www.ietf.org/mailman/listinfo/ipv6 
> --------------------------------------------------------------------
>