[139763] in North American Network Operators' Group
Re: IPv4 address exchange
daemon@ATHENA.MIT.EDU (David Conrad)
Tue Apr 19 12:16:53 2011
From: David Conrad <drc@virtualized.org>
In-Reply-To: <2F4BCAD2-CA18-4121-95AA-3947AD1DE714@istaff.org>
Date: Tue, 19 Apr 2011 09:16:35 -0700
To: John Curran <jcurran@istaff.org>
Cc: NANOG list <nanog@nanog.org>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
John,
On Apr 19, 2011, at 3:46 AM, John Curran wrote:
> Does it have to get worse simply because there is change? =20
Have to? No. However, historically, entropy has generally increased.
> I see no particular=20
> reason that the Internet number registry system can't evolve into =
something
> with multiple registries including overlapping service regions and =
competition=20
> if that's what folks actually want.
We already have multiple registries, albeit with arbitrary (and =
increasingly unjustifiable and unsustainable) geographical service area =
monopolies. This actually points to one of the symptoms of the =
underlying problem: a near terminal case of NIH syndrome. For example, =
just for fun, compare/contrast the results of the following 5 commands =
(to pick a prefix at semi-random):
% whois -h whois.afrinic.net 128.8.10.5
% whois -h whois.apnic.net 128.8.10.5
% whois -h whois.arin.net 128.8.10.5
% whois -h whois.lacnic.net 128.8.10.5
% whois -h whois.ripe.net 128.8.10.5
Note the wildly differing response structure/schemas/tags/values/etc. =
Being objective, doesn't this strike you as insane? Even ignoring the =
simple brokenness of everybody having their own registry data =
schema/response, I keep hearing from anti-spam folks, law enforcement, =
network operators, etc., that the quality of the data actually returned =
is simply abysmal. And soon, network operators are going to be asked to =
make routing decisions on this data not just at customer acceptance =
time.
However, as far as I can tell, multiple registries isn't what is =
implicitly being proposed. What appears to be eing proposed is =
something a bit like the registry/registrar split, where there is a =
_single_ IPv4 registry and multiple competing 'post-allocation services' =
providers. A single registry with a single database schema and data =
representation would seem to me to be infinitely better than what we =
have now (and what it looks like we're moving towards). I personally =
don't have a strong opinion on the competitive address registrar idea as =
long as there is a consistent set of registration requirements, but in =
my experience (reasonably regulated) competition tends to bring higher =
quality/lower prices vs. monopolies.
> Registrants may have exclusive use of their=20
> numbers, but the network operators also have a right to know the =
registration
> of any given piece of address space. =20
I'm not sure I see that there should be a difference in the operational =
requirements for the DNS registration data, but that's a separate topic.
> As you know, multiple IP registries=20
> would definitely pose some coordination challenges in being able to =
reliably
> account for all of the address space at any given moment.
Which is exactly my point. Given that market forces are driving the =
establishment of (presumably) competitive "address registrars", of which =
the first two now apparently exist, how are network operators going to =
deal with the proliferation of whois databases they're going to need to =
query to establish 'ownership' of prefixes?
> What we lack is any meaningful proposals on how to restructure the =
Internet
> number registry system, including what are the goals of doing such, =
how are=20
> those goals and the existing requirements are met, and what =
protections are=20
> needed for integrity of the system.
Unfortunately, I suspect we are past the time in which a well thought =
out, global consultative action (even assuming an agreeable venue for =
such a consultation can be identified) would result in a plan of action =
before being overtaken by events. There are already two "address =
registrars" and at least 5 (6 if you count IANA) address whois =
databases. I expect there to be more in the future, particularly now =
there is an existence proof that you can sell addresses and the Internet =
doesn't explode.=20
Hoever, perhaps I'm being too pessimistic. What venue do you propose =
for a global consultative action to be taken in an open, transparent, an =
unbiased manner?
> Personally, I do not see it as inevitable that "alternative =
registries" must=20
> have a detrimental impact to the WHOIS database, unless they are =
introduced=20
> in an uncoordinated manner and without global discussion of the actual =
goals.
This coming from the CEO of the RIR that decided to come up with their =
own (and yet another) completely new replacement for the whois protocol =
(maybe the 5th attempt will be the charm)...
Regards,
-drc
