ICANN/DNSO
DNSO Mailling lists archives

[ga]


<<< Chronological Index >>>    <<< Thread Index >>>

Re: [ga] Names Policy Development Process


Dear Todd,
You rise an important issue. It has however a long legal, technical and 
political roots and experience we are trying to build upon right now. You 
might be interested to join as some other serious people about the 
international stability, security and development of the network (see below).

On 17:28 27/07/02, todd glassey said:
>I still think that ICANN's biggest political and technical hurdle to leap 
>is really what to do about other ROOT's.This may seem like a simple 
>business question but it has far reaching ramifications that stretch 
>throughout the entirety of what we know as the Internet.
>
>Further - it needs to be noted that ICANN can develop whatever it wants 
>internally but if its processes are too oppressive and too painful to deal 
>with, then these other ROOTS will certainly gain significant numbers of 
>ICANN's existing customers and that is a serious issue to deal with.
>
>ICANN's trying to stop the operations of these other ROOTS is equally 
>problematic since it ***will*** result in law suits and like restraining 
>orders against ICANN, its officers and its agents (the Registrars and the 
>ASO members) from prohibiting these other roots from functioning. 
>Restraint of Trade is a pretty easy claim to prove here under today's 
>circumstances.


A root is the plug of a sub namespace into the global namespace. So .arpa 
is the ARPANET plug into the global namespace. The plug of the global 
namespace into itself (the inclusive root) is the "." alone.

This notation comes from the fact that the DNS to support the Internet 
current practice and parallel internal access and public service access had 
to reverse the current global namespace semantic (ucla.arpa standing for 
ARPAUCLA). This is the same as in ENUM. It created a confusion between IP 
and X.121 naming schemes: the IP 12.34.56.78 could be confused with X121 
78563412. Hence the global root sign at the end of the names. It meant the 
string (even numeric) was part of the namespace and not from the IP 
addressing. This way 90077.3106 from Internet was 301690077 (Dialog on 
Tymnet). That reversing method at a "." has become quite universal in the 
DNS (look at ENUM support). You will find echoes of this in the initial 
RFCs using "root" (general) and "roots" (TLDs) wording.

The "." or global namespace was initially coordinated by Tymnet under FCC 
valued added network services license through its IRC agreements with the 
foreign monopoly services. All of them together formed the International 
data network services system ("Intlnet"). This Intlnet delegated the 
".arpa" sub namespace to the Internet team, the same it delegated many 
other root names to others countries and organizations (this was only 
providing a password to that name into the Tymnet validation system,  under 
standard FCC approved Tymcom or Tymnet/Tymnet rated services). Further on 
it became also regulated by the ITU through the X.121 and E.164 standards.

Tymnet (Telenet and Uninet, the other public system Arpa connected) also 
accepted the transit/access of calls from/to other sub namespaces, as per 
the normal practice of open subnetworks. This is where com (Tymnet), net 
(Telenet) come from.

All this was under rates, licenses, public interconnect agreements subject 
to ITU and to international laws and rules and FCC licences for the US 
parts. The constraints on the Internet and the possibilities for the 
Internet community have been globally summarized in the RFC 920 by Jon 
Postel in Oct 1984. This RFC has been strictly respected (in a restrictive 
way) by the IANA up to now, so even if all this is 18 years old, it is the 
current rule and the basis of the DNS: ICP-3 is wrong on many points, but 
it is absolutely legitimate because it intends to describe the permanent 
IANA policy. If it was not sticking to the 1984 agreement, it would have 
needed a review by the DNSO, what the BoD denied. When Kent, Roberto oppose 
the "alternative roots" they are right: there cannot be other roots within 
he ".arpa" sub-namespace.

But there can be as many other non".arpa" roots in use (and the agreement 
was that they would be respected) in the ".arpa" IP address system. To 
understand that you have to recall that ARPA were the machines associated 
into the US ARPA project. Yet among the identified users groups there were 
the foreign Local Internet communities (ccTLDs) as /customers of the local 
monopoly services.

These people were by no means local operators, but local customers of their 
telecom services and connected registrants communities. They therefore were 
only subject to the local association or university laws, with the task of 
validating usernames with the local PTT (physical connections) and 
registering names into the global naming system though the .arpa 
sub-namespace. This means that if I wanted in France to access "inria.fr", 
I had to connect Transpac, go into an Internet US gateway, and through an 
Internet channel (may be using Transpac) to reach the INRIA system. Many 
suffered from that limitations!

The ccTLD names were accepted as the ISO 3166 2 letters code because usage 
on the namespace was to root local monopolies either/both in ISO 3166 3 
letters codes and 4 numerics (DNIC) from the X.121 list. The RTC 920 
mentions that ccTLDs and multiorganization TLDs (.arpa groups of 
registrants wanting to form their own TLDs) were only registered and not 
administered. This because their names had to be included into the global 
TLD list (root names on the Tymnet supervisor or in the X.121 ITU list) and 
because they could maintain relations outside of the control of Jon 
Postel's team..

You have to realize that all this had much to do with rights, but with 
billing. Root names were introduced to make sure we could easily bill the 
issuer (registry) of the name in sorting the statistics. Anyway, this means 
that ICANN is perfectly entitled to manage its root the way it does, to 
support its own naming system except that it should include the roots into 
the other sub namespaces - that was the deal - instead of refusing the 
interconnects.

A part from being in an old forgotten deal, the reason why is that the 
reasons of the deal are still here. And in not taking care, it leads to a 
conflict with the ITU, NGO, Govs, Business communities it will lose.

The situation we face comes from changes in the situation:

1. the public systems deregulated, so national authority is no more with 
PTT but with States. Mike and Stuart were technically correct in calling on 
States but wrong in asking them to do what they should have imposed on them.

2. the ARPANET is no more, and there is now an IP system infrastructure 
never paid by the ARPA or the USG, IANA still consider as its own - 
creating an usurpation feeling.

3. the Tymnet supervisor is no more acting as the global root ".", the 
".arpa", "gbr", ".fra" etc.. could plug into.

4. the national root names have been using their DNIC as a name (".fra" is 
currently ".2080" +++) but due to the success of TCP/IP the ITU members are 
interested in reusing them with most of their legitimate holders ignoring 
their rights (borne from stability), creating confusion, uncertainty. 
Example: obviously the EEC has nothing to do ask the ICANN about ".eu". 
Just to inform it.

5. the telephone access which was included in the initial deal (Out-Dial 
service) was not supported by the DNS until the ENUM project.

This calls for a lot adaptations. During the last 18 years the respect of 
the RFC 920 created the "status quo" which benefits to the main Internet 
providers. But the system is a 18 years old system shaped to serve 100 
times less users, and expanding. Either its administrators get real and 
reform the system accordingly or the system (which is a distributed 
consensus) will reform itself (as China started doing).

IMHO the only way to do it in good order is to rebuild the Supervisor (the 
global root system) so there is a stable and secure namespace reference 
again (the "dot"). And the ".arpa" root may keep going the way it does 
now.  We have several good technical, operational, political, commercial 
reasons to do that now. The ccTLDs, China, EEC, Govs, etc. have major 
interests in it being done. Or already have started.

What is urgent IMHO is that we make sure it is done "a la Internet", by a 
voluntary effort and not by on an ITU or USG managed budget. This is why we 
are launching the "dot-root" project for a three parallel asynchronous root 
servers systems next generation DNS+ experimentation. We call on who wants 
to participate, sharing his DNS competence and one on-line machine. The 
target is to experiment and then consensually implement a DNS.2 
architecture, operations, policy, at root and user system level - able to 
support the full "." namespace. Do you want to join?

>As a simple example, I allege that it is possible that ICANN is playing 
>antitrust by locking out other Internet Standards Processes and 
>organizations. This is simply demonstrated by that IANA will not issue a 
>system port except to an organization that has an IETF RFC number. So no 
>one from ITU or any of the other standards orgs can submit anything for 
>the issuance of a System Port on the Global Internet unless they play 
>ICANN's PSO Game and that is clearly anti-trust since ICANN does not own 
>the Internet. Which simply says, that without the IETF/IESG/IAB processes 
>in place, nothing gets codified as an Internet Standard and personally 
>that is the largest load of BS anywhere.

May be we can have it another way. ICANN manages the Status Quo because 
sticking to the RFC 920 and alleging that no one is now in charge of the 
namespace, but the ITU which did not really took over yet, is helping some 
to make stable returns. So it does not lead people to go ahead with major 
innovations and make them better organize a technical status quo. And 
freeze innovation.

Exemple: the key development to free the Internet from ICANN is a Windows 
dns-resolver plug-in. No one on sourceforge even proposed it.

>Another concept here is that the Domain Owners are more the friend of the 
>Registrars than ICANN is, and if you don't believe me, then ask Verisign 
>how many bodies to GoDaddy they lost because GoDaddy is more friendly to 
>end users at the wallet level. And most of the Domain Registrars don't 
>realize this yet because we spend so much time arguing about personal sh*t 
>and not the goals of the group.

May be the main thing is that we do not need the Registrars at all nor to 
pay for DNs. It is surprising that no one works on the necessary development.

The "root intox" works well. This is brainware: what counts is not the way 
the network is, but the way the peole believe it is to be used.
jfc



<<< Chronological Index >>>    <<< Thread Index >>>