A homenet implementation may indeed decide to self-destruct in a puff of smoke 
before daring to step foot into the classful lowlands of the Interface ID, but 
I actually appreciate that the algorithm itself is more general; for robustness 
sake, in case it is ever used in a different context, in the event that why64 
ends up being wrong, if ISPs don't listen to us and /64 becomes common and we 
have to do *something* to accommodate, etc. 

Don't get me wrong, I've fought long and hard in many forums and venues to keep 
ISPs from thinking /64 is ever a good idea. I have no problem with the 
strongest language we can provide to try and enforce that. At the same time, 
hedge your bets. 

- Mark

On Oct 8, 2014, at 4:21 PM, Tim Chown <[email protected]> wrote:

> 
> On 8 Oct 2014, at 14:14, Pierre Pfister <[email protected]> wrote:
>> 
>> Why should we mandate homenet implementations to *brake* in situations where 
>> they could work fine ? Why should we voluntarily prevent a link from being 
>> configured if we actually can configure it ?
>> 
>> If MUSTs are the solution, then I would rather see a ‘ISP MUST provide a /56 
>> to customers’ than ‘Homenet MUST brake when the provided prefix is not big 
>> enough’.
> 
> But this is what the homenet arch text says in Section 3.4.1:
> http://tools.ietf.org/html/draft-ietf-homenet-arch-17#section-3.4.1
> 
> i.e. don’t go longer than /64, and ISPs should provide enough prefixes.
> 
> The why64 text is very relevant here.
> 
> Tim
> _______________________________________________
> homenet mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/homenet

_______________________________________________
homenet mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/homenet

Reply via email to