Re: [v6ops] RFC7084

[Alexandru Petrescu <[e-mail]>]

Le 12/12/2013 11:02, Ole Troan a écrit :
>>> I'm not sure I understand how's that materially different/better
>>> than a node just trying to request a PD if it wants one, and
>>> coping with the response (whatever it may be), like it does
>>> today.
>>
>> The usual concern is that a bazillion devices all requesting PDs
>> every so often adds up to a lot of traffic.   But since the
>> existing spec doesn't forbid this, we're stuck with it.
>
> then you're understanding of "usual" is different from mine. :-) as
> far as I know this can only happen in very specific circumstances
> during transition.
>
> we should be more concerned about this idea of using one protocol to
> provision another. in this case ND to configure DHCP. is that a good
> design principle to follow?

PRobably not.  Stated as such it would be too limiting (ND to help
running _just_ DHCP).

Maybe Radius as well.  The RA would just say that some means is
available to offer a delegated prefix, w/o saying which, without saying 
whether it is stateful or stateless, secure or insecure, etc.

Alex

>
> cheers, Ole
>