RE: alpha draft requirement

[Harald Tveit Alvestrand <Harald@Alvestrand.no>]

At 03:43 18.01.00 -0800, Andrew Draper wrote:

>"An IDN capable resolver SHOULD NOT generate any more traffic for each
>request (for both IDN
>and pure ASCII names) than a non IDN capable resolver.  This also applies to
>the resolver part of a DNS server providing recursion."
>
>I'm sure this loss was accidental.  If not then here is my justification:
>
>People are always threatening DNS meltdown.  If we specify a protocol which
>increases the amount of DNS traffic we won't be thanked.

actually the amount of traffic due to DNS is minuscule at the moment 
compared to other stuff. (Pictures!!!!)
However, in any random request, the TIME spent doing the DNS lookup may be 
significant - because of overloaded, far-away or misconfigured DNS servers.
(Many many times i've seen DNS lookup failures, and found that the i**** 
configuring the site has placed both master and slave on the same network 
segment - or, in one case, on the SAME SERVER.....the lameness.....)

>Of course, if it is not possible to specify a protocol which uses the same
>amount of traffic then some increase is inevitable, however I suspect that
>some ISPs (for example) in english (only) speaking countries will be
>deterred from upgrading to a new version of their DNS software if it
>produces no benefits for them (since they use only ASCII), but generates
>twice as much traffic.

If it generates twice as much load on the core servers, that's a problem 
for the core (and the client).
If it generates twice as many network packets, I think nobody cares about it.
If it generates twice as many round trips, the users will hate it.

My IMHO.

                         Harald

--
Harald Tveit Alvestrand, EDB Maxware, Norway
Harald.Alvestrand@edb.maxware.no