To:
Robert Elz <kre@munnari.OZ.AU>
Cc:
Randy Bush <randy@psg.com>, Rob Austein <sra@hactrn.net>, dnsop@cafax.se
From:
Johan Ihren <johani@autonomica.se>
Date:
28 Feb 2002 12:34:52 +0100
In-Reply-To:
<2325.1014887913@brandenburg.cs.mu.OZ.AU>
Sender:
owner-dnsop@cafax.se
User-Agent:
Gnus/5.0808 (Gnus v5.8.8) Emacs/20.3
Subject:
Re: Minneapolis - agenda items please.
Robert Elz <kre@munnari.OZ.AU> writes: > Date: Wed, 27 Feb 2002 21:32:21 -0800 > From: Randy Bush <randy@psg.com> > Message-ID: <E16gJAz-000G7v-00@rip.psg.com> > > | i see it as 'your' responsibility to live with the altered physics when > | you, irregardless of transport, *leave* the internet by moving behind > | a firm boundary. > > There's a presumption in this that leaving the internet is a voluntary > choice, made by a party that is excluded. > > If your ISP decides that they don't like your mail behaviour (perhaps > you say bad things about them in public...) and filters you - whose > responsibility is it to then "hide" the problem by rearranging MX records > so mail can get delivered without delaying other sites MTAs attempting to > connect to filtered addresses? > > Of to be even more bizarre, if I decide to filter (the metaphoric) you, > so my MTA cannot reach your MX target, who should be dealing with that > as a problem? You're the filtered one after all. > > Filtering creates all kinds of issues, and simply deciding that someone > else should "fix" it seems to me to be outside reality. Well said. > kre > > ps: is this discussion about the "don't publish unreachable" draft, > or about the "v4-v6 namespace fragmentation" draft? I would have > guessed the former, but some of the comments make me believe that it > might be about the latter - it would be nice if the Subject header > gave some clue... I know that the meeting agenda item is v4-v6, but > most of the discussion seems to be about whether unreachable > addresses should be published. All very confusing... It is really about both, in a general sense. They are connected in devious ways, which I think is becoming obvious to anyone who has had the stamina to follow us this far. But at a higher level it is not about any of them, but rather about the definition of "the Internet" (wow, how's that for a meta-discussion?). There are two choices of definition: * "1-1 mutual reachability for any pair of nodes, adjust the namespace to fit". Let's call this "Internet Classic". * "same namespace, regardless of reachability". Let's call this "Internet Noveau". and their respective consequences are: * with the classical, definition the v4/v6 migration goes down in flames (since that definition makes v4 and v6 *different* internets rather than different transports for the *same* Internet). Then they *should* have different namespaces. * with the latter definition v4/v6 migration becomes at least teoretically possible, since nodes are allowed to be unable to reach each other while still inhabiting the same Internet. But instead split-DNS becomes bad form, since the unreachability characteristics should not be tracked in the DNS namespace. I really don't think that I can make the choice any clearer than that. It should also be pointed out that with the latter definition we raise the requirements for the v4/v6 DNS transport bridging ickiness, while with the former we lower them. Johan