Re: [v6ops] WGLC: draft-ietf-v6ops-unique-ipv6-prefix-per-host-02 - multiple prefixes per device

[Alexandre Petrescu <[e-mail]>]

Le 16/03/2017 à 16:03, [e-mail] a écrit :
>>> I'd like to understand your meaning, if you don't mind.
>>>
>>> A router, per RFC 2460, is a system that receives a message, does
>>> no processing on its content, but repeats it to another system. A
>>> host is one that receives messages and processes their
>>> application layer content.
>>>
>>> I think it's fair to say that an iPhone etc has more than one
>>> interface; WiFi and LTE, perhaps. A laptop often has several
>>> interfaces such as physical and wireless Ethernet, USB,
>>> Bluetooth, and other things. Having multiple interfaces, however,
>>> does not qualify it as a router; what qualifies it as a router is
>>> when it receives a message and repeats it, usually on another
>>> interface.
>>>
>>> Yes, an iPhone acting as a hot spot can route between its WiFi
>>> and LTE ports. If it is not configured as a hot spot, however, it
>>> generally uses its many interfaces as ways to receive messages,
>>> pass them to applications, and respond to them. Absent a specific
>>> (and unusual) configuration, an iPhone therefore is a host, not a
>>> router.
>>>
>>> Are you and I using the words in the same way? I get the feeling
>>> that you see multiple interfaces and infer "router".
>>
>> As others have implied, multiple interfaces means *potential*
>> router, and that's the point, isn't it: if you give a *potential*
>> router only a /64, it can support a single subnet but it can't
>> support multiple subnets and subsidiary routers. And Alexandre is
>> challenging that limitation. But we have said elsewhere that ISPs
>> should provide more than a /64 per subscriber; it seems like a
>> different discussion.
>
> Take the simple case of a laptop with VMs. It has to handle the
> smallest common denominator. As long as networks treat everything
> that connects to it as hosts. Then either the host with VMs (aka
> router) has to pretend to be a host and do one of bridging or NAT. Or
> with this proposal it can subnet the /64.

I am not sure I understand, but it is hard to subnet a /64 further to 
interfaces that want their IIDs to be 64.  Or you want these interfaces 
to not run SLAAC?  Do you have other means to automatically configure 
their addresses?

Alex

>
> Cheers, Ole
>
>
>
> _______________________________________________ v6ops mailing list
> [e-mail] https://www.ietf.org/mailman/listinfo/v6ops
>