Re: draft-bourbaki-6man-classless-ipv6-00

[Alexandre Petrescu <[e-mail]>]

Le 03/06/2017 à 03:40, Fernando Gont a écrit :
> Hi, Mark,
>
> On 06/03/2017 03:35 AM, Mark Andrews wrote:
>>
>> In message <[e-mail]>, Brian Hab
>> erman writes:
>>> Hi Bert,
>>>
>>> On 6/2/17 4:03 PM, Manfredi, Albert E wrote:
>>>> -----Original Message-----
>>>> From: Templin, Fred L
>>>>
>>>>> My meaning was for the ISP to give the cell phone or home gateway
>>>>> a /64, then let the cell phone/ home gateway subnet the /64 to
>>>>> the IoT devices within the subnetwork it provides as it sees fit.
>>>>
>>>> Sorry, I have to make an important correction:
>>>>
>>>> Presumably, using some sort of internal address format, also /64, such
>>> as privacy addresses? Yes, true, but ...
>>>
>>> I interpreted Fred's proposal as:
>>>
>>> 1. ISP gives the phone a /64
>>
>> The ISP could give each phone a /48.  There is NOTHING stopping the
>> ISP giving a /48 today.  IPv6 is sized to allow this.  When you
>> stop trying to hand out the minimum and start handing out reasonable
>> quantities of subnets the so called problems go away.
>>
>>> 2. The phone delegates longer prefixes to the devices behind it
>>> from the /64
>>
>> The phone then hands out /64's to devices behind it on demand.
>
> And such devices, if they feel like sharing, hand a...?

'64share' is INFORMATIONAL, should not be used on a large scale.

Alex

>>> 3. The ISP router has a single /64 route that points to the phone
>>
>> The ISP's router has a single /48 that points to the phone.
>
> At which point we probably should start thinking about ipng-bis :-).
> That's 16-bits away from IPv4, which has (or used to) /32 pointing to hosts.
>
> Cheers,
>