[c-nsp] IPV6 in general was Re: Large networks

TJ trejrco at gmail.com
Fri Aug 28 20:10:19 EDT 2009


>> >Gert Doering <gert at greenie.muc.de> writes:
>> >
>> >> Yes.  IETF really botched that "everything is automatic but we
>> >> don't tell you how to discover DNS" part.
>>
>> In every design there are tradeoffs that are made ...
>> and with the benefit of hindsight it is easy to point at the wrong
>decisions.
>
>It's very hard to find a reasonable argument why "not considering DNS when
>designing autoconfiguring networks" could be called a *tradeoff*.
>
>Tradeoff against what?  "Mostly unusable result" vs. "invest some brains"?
>
>gert

I believe the thinking was that the RAs are only (pre-RFC5006) providing
information about that which they directly know about.
Tacking DNS information in makes perfect sense from the hosts' perspective,
but the router is now providing 'second hand' information.

And it depends on the deployment scenario.  
A 'cloud' of simple sensors won't be surfing the net, so the need for name
resolution is probably lessened.


In the end, I agree - my host needs information (DNS) along with addressing,
so Just Make It Work(tm) :).


/TJ



More information about the cisco-nsp mailing list