[gnso-rds-pdp-wg] [renamed] Key early questions

[Andrew Sullivan]

Hi,

On Wed, May 11, 2016 at 11:54:11AM -0400, Sam Lanfranco wrote:
> As for Andrew's point that operators have need for data to conduct their
> business, that "need" breaks down into three categories:
> 1. There is the data needed to meet ICANN's contracted requirements.
> 2. There is the data needed to conduct the business of dealing with clients
> as registrants (invoices, billing, payment, etc.).
> 3. There is desired data to conduct other marketing and innovative aspects
> of being a contracted operator (registry, registrar).

That list is missing one, and it's the one I am concerned about.  It's
the data needed for people who have no contract with one another, but
who are trying to exchange packets and having a problem.

The Internet works (and has taken over basically all other
telecommunications systems) in part because it enables easy
interoperation with no (or at most minimal) prior arrangement.  There
are lots of communications technologies that have risen and fallen in
the past that all depended on pre-existing contractual arrangements
between the parties or else bilateral contractual relationships that
resulted in a chain of contractual responsibility.  But the Internet
doesn't work that way (even mostly), and that's part of why it has all
but replaced the phone system and the various proprietary online
communications systems that didn't depend on the Internet Protocol.

The way that the Internet can be kept working is by direct
co-ordination between end points when the need arises.  To allow that
to happen, people used an RDS called whois.  (Because whois as a
protocol and a mechanism is just completely bonkers bad, geeks came
together and decided to build RDAP.  The present PDP group, as far as
I can tell, has decided not to take the analysis done by the WEIRDS WG
as given, and instead wants to redo much of that.)  In numbering
resources, the RDS allows one to figure out the person to contact for
a given IP address or AS number.  And for names, the RDS allows one to
figure out whom to contact when their domain is not working and all
their mail is bouncing or so on.  The technical details -- the name
servers and DS records and so on -- are all there so that one can do
adequte troubleshooting from one's own network (the fault might be
your own, for instance -- a stale cache, say).  And some contact
information (maybe not all we have) is necessary so that one can find
an alternative method of contact when the usual well-known aliases
(abuse, noc, webmaster, and so on) don't work because the domain is
broken.

This isn't a little matter: if Comcast's customers are all trying to
watch the Mars lander and suddenly nasa.gov goes DNSSEC-bogus, then
Comcast has an expensive support burden to deal with even though it
has no contractual path through to NASA to work out the problem.  (Go
ahead, ask them how expensive, because it happened.)  On the Internet,
we don't use contracts for this: we use the RDS instead.  Technical
failures and abuse are the operationally critical cases for which we
need (and always needed) the RDS.

If I ran the circus, I would arrange things so that ICANN's
contractual requirements were aligned with the things necessary to
support that technical function.  That's the particular thing I want
to make sure we don't miss in all this, because that's one function
that the system must support.

Best regards,

A

-- 
Andrew Sullivan
ajs at anvilwalrusden.com