I wasn't exactly thinking of prefix delegation. I meant what I said
literally: assign N addresses to a subscriber, where N doesn't have
to be a power of 2 and the addresses don't have to be contiguous.
The CPE can either hand them out to LAN hosts, or statelessly
NAT66 them. Either way, we don't mess with the /64 boundary,
and the SOHO can have a large handful of addresses.
Brian
On 2009-04-13 20:34, Rémi Després wrote:
> Brian E Carpenter - le (m/j/a) 4/10/09 10:41 PM:
>> On 2009-04-10 21:12, Rémi Després wrote:
>> ...
>>> Many personal computers can establish LANs behind them, and/or can run
>>> several internal virtual machines (what I call "host-rooted subnets").
>>>
>>> Let's assume such a host is on a /64 link, with a /128 address.
>>
>> What a depressing assumption.
>>
>> Why not assume it's on a /64 link but has been assigned 16 /128
>> addresses?
>
> The idea is that a typical default behavior consists in taking as IID
> the modified EUI-64 that is derived from the unique MAC address
> available on the link.
>
> But, yes, other assumptions are possible. The CPE, and intermediate
> routers if there are some, can indeed use prefix-delegation. This is
> however, IMHO, not as typical and simple.
>
> Prefix delegation, which has its own applicability scope, seems to me
> somewhat opposed to that of SLAAC. More intermediates solutions seem
> possible. SAM is thought to be a starting point for this, but I am
> conscious that this still needs to be clarified.
>
> RD
>
>
>
>
_______________________________________________
nat66 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nat66
