Re: [dnsop] WG Last Call: draft-ietf-dnsop-misbehavior-against-aaaa-00.txt

["J-F C. (Jefsey) Morfin" <[e-mail]>]

At 03:27 25/03/04, David C Lawrence wrote:
>Someone, "J-F C. (Jefsey) Morfin" <[e-mail]>, I think, said:
> > the meaning of the word "global" differs in American and in English.
>
>This was in response to:
> >> DNS has been designed to present a single, globally unique name space

Yes, I am the "someone" :-). You might look at an American and at an
English Dictionary. However, it has so many technical implications to the
network, the DNS nd cybernetics that I will explain in PS for those interested.

If you are not interested, please disregard.
jfc


PS. In American as in other languages "global" means "from the whole
globe". This is confusing because this is understood as:
"universal, catholic" in American.
"all the parts of the globe" in French, English etc.

This is the very basis of every misunderstanding among network
architecture for 20 years (since we introduced the word in datacoms).
Probably one of the reason why the IAB never produced an
Internet model. Because by nature (DoD Jan 1974 specs), the
Internet is the internetting of several networks (and technologies
according to Vint) what fits the Anglo/French multiple/globality but
less the American global/universality.

So the global Internet is understood in American as the universal
TCP/IP system :  legal definition of the Internet (47 USC 230 (f)(1)).
In English, French and most of the other languages it will be
understood as the addition of all the different entities using TCP/IP.
The currently voted Franch law on the internet does not even care to
define "internet" it considers as a generic name, saying "by internet"
as you say "by train", "by car". This translates into current spelling
when you use a big or a small "I".

Due to the general usage of Basic American as a lingua franca this
is not noticed on a daily basis, but when we come to architecture
and governance organization the split is very important.

One the result is the difficulty to introduce granularity and
subsidiarity in an universal (unilateral) model while they are
the very basis of a global (multilateral) model. The problem
also of its cybernetics - (or governance, same Plato's word):
the way to manage datacom systems as per DARPA today -
the intelligent ubiquity of the eTEN European family of projected
services. We speak quickly of different things.

This is the whole dispute over RFC 1591, ICANN, ccNSO and
current UN hearings. This was the problem wich emerging at the
WSIS in December. The US culture understands the DNS as a
hierarchical system where the IANA delegates its authority to
TLD Managers as the trustees of their community (meaning the
people mutually trusted to fulfill the mission). While the rest of
the world understands it as a first level flat plan of equally
sovereign TLDs listed in the root file, managed by trustees
delegated by the registrants or their representatives (the States)
[BTW this is the way we built the international name space].
When VeriSign introduced PathFinder what chocked us the
post was the reaction by ICS and Paul Vixie to develop a
patch to block a decision by a TLD Manager. Because it said
where they considered the authority was. And the current
legal action is over that, the same as the UN meeting of today
in NY. They do not understand/say it fully yet, but the WSIS
shown they learn very fast/speak up more easily.

When I quote RFC 882 and RFC 883, it is because in RFC
883, Mokapetris describes the DNS as a true global (common
sense) system perfectly able to cope with the Anglo/French
meaning. While Jon Postel in the RFC 882 describes the way
he is going to manage for the ARPA Internet. Then they went
global (plugging into the international system). The result of the
international consensus is in the RFC 920, which is correctly
considered by ICANN as the source of its global legitimacy
on the legacy. But only on the Legacy.

In 1996 the US consensus (reflected into the USC) was not
yet stabilized (you see that in the definition of the data
services for the Telcos, leaving a lot of flexibility to the FCC
in order to define them). The same non finalized policy lead
to ICANN. Jon Postel had obviously understood that his
system was becoming the global network - and I suspect
from the Memo #1 to ccTLDs he understood it in the European
way (but he kept control in a very clever way, ICANN still
follows however it is now outdated). And he started acting in
the proper way (conflicting with many if I am correct) -
proposing new TLDs, playing with the root. But he died.

When I quote ICP-3, it is that who ever wrote it has done
a very very good job at respecting the ideas Jon Postel
certainly had, in our today context. The only thing I disagree
with (coming from the global side rather than going to it) is
that we must keep the concept of a single root. ICP-3
think that we might abandon it: it is true that we need to
abandon it as a "file", and make it a complex "matrix",
but we need to keep the concept of single (and to stop
confusing it with centralized). Single means to respect
the single global multiple diverse reality (Franco/English
meaning), not an unique universal vision (American).

Single and diverse means flexible and we may keep for
ever a generalized naming service. Unique and
universal means a non acceptable centralized control.

Now, the big step ahead of IPv6 is that the addressing
may now also become global and diverse in its turn.
This means that the restrictions Jon Postel imposed in
RFC 882 are no more necessary and that the system can
really be used as Mokapetris describes it in RFC 883.
Hence what I say.

However, not the way the IPv6.001 numbering plan
is designed as global (unique) and centralized. Leading
to concepts such as HIP etc. quite rewarding when selling
addresses and an IP DNS service. Hence the increasing
uneasiness here. But IPv6.010 or even IPv6.011 make no
difference in DNS and network connections. They are
transparent to the technology (they can only make sure
that IPv6 is transparent and help it to be consistent).

Now, there is another reason why I am interested in
this proposed RFC. IPv4 transition to IPv6 is the first
real life interoperability of IPvXs. There are still many
simple things to make accepted in here and elsewhere.
But if we want to proceed towards the continuity of the
digital ecosystem, we need to make IPvX interoperable
with many other systems (like Telephone, TV, Radio,
domotic, space, automation, etc). This means to make
the technology global.

But again you may think of a global technology as the
NGN (new generation network, everything under a
superIP, at layer 3 or 4) or as internetted technologies
which will cooperate at layer 7/8.

To end with this. You will note that the global unique
namespace by John Klensin's definition is the same
Domain Name must resolve to the same site.

How a sad world :-)
- no local emergency number
- no follow me
- no anycast
- no Dynamic DNS


.
dnsop resources:_____________________________________________________
web user interface: http://darkwing.uoregon.edu/~llynch/dnsop.html
mhonarc archive: http://darkwing.uoregon.edu/~llynch/dnsop/index.html