Re: How the IPnG effort was started
2004-11-20 09:23:24
Dear Stephen,
there are two things necessary first to accept:
- that what we name "names" and "addresses" are two of the three main ways
to identify objects. The way linked to the object, the way attached to the
system, the way related to the users. And because this analysis is not
worded clearly and there only two accepted levels there is a lot of
confusion. This confusion is increased by the routing Internet system.
- a identification plan is something far more complex than a protocol which
is not to be designed a minima, in a few mails but which calls for
experience, studies, consensus of a lot of authorities, worldwide. For
example, you cannot say that you treat all the countries in the world with
"9" bits. There are scores of standards, study, needs,billions of daily
affected usages, transactions, records, legal consequences, etc. etc ... to
be taken into account. This is not something where a Government "can apply"
for a /32 to some RIR. This is something where States sovereignty
_decide_through a carefully designed treaty. This is why IPv6 does not
fulfill the need the way it is proposed. Again, the problem is not the
protocol, the problem is the lack of interest in the way the numbering is
to be managed. Which makes it useless for most of the uses it could be also
used for, and the world need to see addressed. Add to the resulting Vint/US
centrric way it is perceived.
I accept that we did it quickly, and alone, 20 years ago for X.121. But we
did because we were licensed to do it by the FCC, within an ITU (CCITT)
consensus aggregating all the telecom monopolies which were our customers
(delegating us) and partners for that (we were the leading public operator
and the source/end of their international traffic and the manufacturer of
their technology). We kept them informed and seek their direction in
bi-yearly meeting we made _them_ to organize and chair. Also we only
proposed a framework, we discussed it with every monopolies listening to
and embodying their remarks, respecting the standard they had voted. We
also only partly committed 20% of the numbering space (0-- and 9--) and
make it easily reversible or mendable by them. We listen to the users and
to their dealy changes and developments, we did not evangelize them with a
15 years old theoretical gospel. I do not say this to criticize, but taking
advantage from a positive experience may be useful and helpful when one are
mudded in a problem. Obviously these are two different situations.
On 06:39 20/11/2004, Stephen Spunk said:
There is a huge difference between knowing someone's phone number and
knowing their exact legal identity. Phone numbers, even ones portable
between carriers, are inherently temporary things. And, let's not forget,
many people pay their phone companies not to be listed in directories and
pay again when too many people (or even a single ex) know their number.
MAC addresses were proposed for the lower 64 bits of IPv6 autoconfigured
addresses, and privacy advocates threw an amazing tantrum about how that
would lead to invasion of privacy -- and that just tied an address to the
NIC of a particular computer, not to a publicly known legal identity.
Please do not confuse what is possible and what is mandatory. The rule is
not what YOU decide, but what EACH USER decides for himself. This is not to
be an obligation but a service.
The real product is the addressing plan. And the reasons why no one is
excited are that:
- these addresses are managed "a la IPv4", as a unique Vint Cerf's/ICANN
numbering area. This is what they want to correct with ITU. I submit
there is no conflict. IPv6 has 6 different numbering plans. Let say that
001 is for the US Vint's legacy and 011 for international. That Vint can
manage the 001 area and the ITU the 011 area. This is status quo.
If the IP routing infrastructure were regulated (as the telephone one is)
so that all US ISPs had to carry for non-US routes was a single prefix,
this would work. Currently, it doesn't work that way, and many things
would need to be changed for that to happen.
This is your point of view as a technologist, who represent may be
1/million of the people of this planet. What you mean when you say "many
things would need to be changed" is that many things should be changed in
your vision, but you may fail to see that for your approach to happen there
would many things to change in the way the other billions people interact.
This may be why IPv6, using that plan, does not catch.
- the way the countries will manage their numbering space is up to them.
But if I refer to the telephone solutions, my guess is that many will
differentiate routing and addressing in a very simple way (and this is
certainly what the ART (French FCC) wants to hear about - because this is
what users want : IP addresses are to be independent from the ISP). This
means that they will allocate national IDs that you will be able to use
as a NetworkID or as a UserID. And you will probably get the UserID for
free at birth or creation, probably additional ones on a small fee and
you will pay for the routing to your NetworkID.
Why the heck should users care what their IP address is in the first
place? They are not intended to be seen by users, and there is a
user-friendly replacement called DNS that is manageable by end-users for
their needs.
I am afraid you confuse what these two systems are doing. See below.
Addresses inherently represent position in the topology. Any attempt to
abstract addresses out of the topology will simply mean that a new, less
transparent layer of addressing will be created underneath and a mapping
mechanism will be added, increasing latency and decreasing reliability.
Network identifiers are not absolute coordinates and identifiers. An
address is an index in a table. What is important is that this index is
fully controlled. The less transparent added layer is the additional
'addressing DNS' some propose (how to do it for mobiles other wise). You
are right: this would be absurd and this is what a lack of addressing
organization will necessarily call for, to compensate it. But let
understand that addressing is not a directory. One want to access an
address in a table, not to document a whois. A whois is only necessary when
you sell host addresses to host owners. Please do not think in 1983
Internet and IPv4 Host centric academic network terms. I know this legacy
is difficult to overcome. But we are in a real demand world, not in RFC
legacy.
Just refer to the basic Internet architecture principle of Brian
Carptenter: there is only one principle that will never change : that
everything may change. And changes do not occur because IETF technologists
want them or the way they want, they occur because the users want them. The
IETF is not to invent the Internet but to help users to make it to work.
How does that fit into a /128? Very simply. The final global network
address will result from the concatenation (probably described in 0-Z
numbering) of :
- a numbering plan header (like 011) + may be one or two additional
digits to qualify plans, documentation, anycast, multicast, and he type
of service/network (like telephone, Internet, TV, Radio, Posts, etc. )
- DCC+national number as a Network ID. Global routing is made at this level.
- DCC+national number as a UserID. Network local routing is made on that one.
- users interfaces.
This means that when I subscribe to a network or another I will keep my
same UserID in each network, but my number starts with the NetworkID of
the access provider. So, if I concatenate the traffic of several ISP this
makes no problem. I can even pay these ISP on the basis of the datagrams
they carried. A mobile changing from network will be easily followed. The
/128 address of my French mobile when I travel in Korea will be will be
011xx+KoreanISP+FrenchID.
We already have this in effect today; there is a numbering plan ID, a
network ID, and a host ID. It'd be trivial to assign EUI-64s to users
instead of to NICs, except that we'd need to make allowances for multiple
numbers per user.
You have to make allowance of roughly 1000 numbers by users, multiplied by
1000 for management (you never know what is going to increase), multiplied
by 1000 for innovation, multiplied by PI like in any project multiplied by
an uncertain number which will decide of the duration of the plan. There
are rules, experience, examples, obligations in that sort of game I never
even seen discussed in here. To sell IPv6, is for us to make sure people
concerned are getting involved and convinced that IPv6 can foot the bill.
My worry is that they already did just that in CCITT (ITU-T) in the mid-80
and they decided that they will not use addresses longer than 32 digits.
With the experience and development we known in 20 years I feel we should
base our thinking on an _open_ /256 base. But the we can manage for a while
with /128 if we provision future /256 expansions in new RFCs.
Of course, since configuration is no longer automatic, we'd see
billions of cases where users would type in this information incorrectly,
and we'd need to figure out what to do when multiple users share a
computer. And then there's the privacy issues...
No different from having jefsey.com, jefsey.org, jefsey.net.
My email address (stephen(_at_)sprunk(_dot_)org) would work just fine if I were to
move to France or Korea; in fact I've used it from France myself and one
of the other users in that domain has lived in Korea and Thailand. Worked
perfectly, and there was no need for government-issued IDs or mangling of
IP addresses.
I am afraid I was confusing. I was just saying that my personal nickname
"jefsey" worked the same in the DNS under various TLD as my 280810001
UserID would work the same in the 3110022 NY ISP and in my 280820087 French
ISP.
If you consider the way famous names are treated in the DNS they do exactly
that. Whatever the TLD, IBM or MERCEDES are like universal user names. This
is not an IETF proposition. This is a WIPO decision.
The priority is easiness of simple, stable, sure, secure fool proof use,
not to do what is technically possible. For that everyone is ready to waste
a lot of numbering space not to lose a lot of time and money in bug
tracking and in calling user support. Even the well documented Jordi's site
on IPv6 is a total darkness for 999.995 people out of a million in the world.
This has a cost that obviously no government (except may be the US DoD with
cybersecurity), no organization (except research and some fancy minded
corporation or academic institutes) or entrepeneurs betting on a 15 years
old novelty has seemingly found an ROI for.
This means that everyone has an address for his web/mail, for
broadcasting TV or cognitive radio, etc. You can discuss international
agreements, establish treaties on content, on address-back (feed back on
an address?) payment authentication, establish usage warranties and
insurances, etc, etc. We are in regalian (Government role) business.
We already have universal addresses for these functions; SMTP, SIP, etc.
all have DNS-based addressing schemes that allow users to keep their
identity(ies) with them when they change network-level providers.
I am afraid you confuse DNS and IP addressing. I am not against the concept
(less than 50% of the internet exchanges are using the DNS) but this will
lead to confusion. Actually what you propose is OSI. The international US
system during a time (mid-80s) supported OSI addresses in using numeric
names. But our target was to introduce naming back into OSI, though usage
(like in DNS today), not to replace it. The OSI problem was that confusion.
Obviously there are objections. And these objections are what has to be
worked on to sell "IPv6".
0. there is no more way to make money worldwide because I have been given
an Excel table to fill. IDs will be allocated by Govs the way they want.
What will be paid to RIR, NIR or LIR will be their real service with QoS
control. This calls probably for a new economic model.
I'm sure if a national govt came to IANA (or an RIR) and asked for a /32
to address everything in their country, and arranged for a national-level
routing infrastructure to use those addresses exclusively and efficiently,
it would happen. They'd then have 96 bits to do whatever they wanted
locally without interference.
I am afraid that no Gov will ever came to IANA and ask... The same as when
Stuart Lynn asked Gov to advise, or IAB asks Govs to finance, etc. the Govs
say they are not that business, but in the business of deciding and to
protect people when they trust their decisions. This is what is named
"regalian duty" (what goes with regalia).
Govs have understood there is a problem. They started with names. Now they
start considering numbers. Govs have signed a treaty nearly 140 years ago
to jointly address their communications issues through a universal agency
(it was the first one). For years this agency has been told inappropriate
to manage the Internet - while it supported well the semi-private
initiative of the first international network and erred in parts in
specifying a too rigid, too monopoly oriented technology (OSI). They
observed. Now they are trying to analyze what they should do, to make sure
their digital infrastructure works properly.
As you may recall the IETF and the IANA owns nothing. People, Govs,
Corporations owns the equipment. And they increasingly think the solutions
proposed by IETF do not match the way they want to use them. So, ITU has
started proposing resuming its role of center of reference. For names
(ccTLD) and Numbers (DCC). These are not /32 but /16 with accepted rules in
part of this /16.
1. there is no room enough in /64 as actually (if I understand well) /128
addresses are just /32 addresses extended to /64 with a user subaddress
payload. User addresses will probably requires /80 or /96. Less than half
in the routing tables. But structuring may permit clever thinking.
Enumerating all the humans on the planet only takes 33 bits today, and
even with 9 bits for a country code and a few bits for multiple devices
per user, we still have nearly two dozen bits left unused. Please explain
why you think /80 or /96 will _ever_ be needed to count people.
This means that you can probably name all the people of the planet with
unambiguous 8 characters names. So why not to change to an DNSv6 8
characters system and see if it takes off.
The first rule of a numbering scheme (like in database indexing) is to keep
as much as possible free space. That you ask to me instead of both of us
referring to an ISO standard shows that out debate is out of scope. I am
not arguing what should be done. I am telling what will not work.
2. there are much more needs to address virtual objects than just
computer ports. So wee need to establish a numeric root of the numbering
schemes accessible through the network, to give them an addressing
capacity (this is what we called the Uninum proposition) of an unlimited
size (their purpose is not necessarily to number network entities, but to
number entities which can be reached through the network). They may
eventually be supported by numeric names. These addresses will become
more and more important as unique lingual and time independent
references. But this is another aspects of the changes we needs.
Why do you think people want numeric names for things? We already have a
textual naming system that meets all of your requirements of unlimited
length, country (and often province) identification, use for multiple
purposes, etc. Even you state your theoretical numbering system will be
conveyed in alphanumeric representation, so why not use the naming system
_that already does that today_?
Sorry for the confusion. Please refer to my initial remark. There are three
levels of identification. Only 2 are documented by IETF but the three are
used by the users. We name "brainware" the way users collectively uses the
network and think the network is. The point is not so much what the
hardware and the software are, but what the people think the network is and
the way it should react to them.
I used "eventually" with a French flavor. I mean both that this will happen
(English meaning) and that it can also be used (French flavor). I say that
necessarily numeric names will be used to support some addressing (because
today the confusing IETF wording associates names with DNS and numeric with
IP addresses). Because they already are (cf. use of numeric names by non
roman character set users).
Please refer to my initial remark. As long as there are no _clear_ words to
describe all the functions carried in a network system and possibly a clear
model to describe where they fit, that can of misunderstanding and waste of
time will happen. Sorry. All the words we use today are inherited from the
way the function they correspond to was supported in 80s Internet.
Today, interfaces with some functions can only be carried with numeric
names, how absurd it can be. Example: ENUM (a direct addressing relation
would be far more efficient with a /xx-Numbering plan addressing.
We're already seeing a move away from numeric (i.e. address) based phone
systems towards textual (i.e. name) based communication systems; we're
only a few years away from users not having phone numbers at all, but
rather SIP URIs that look the same as email addresses. Users don't like
numbers, and they shouldn't be expected much less forced to remember tham,
particularly long ones.
Sorry, this is not a move away from "numeric". This is a move towards
simplification in aliasing the information used to called a telephone
device which is identified by an address in a network. IP addresses are not
to be good mnemonic, they need to be stable and globally consistent. There
is much demand the other way around. To send mail to telephone number (if
there was no spam). There is no mystery every information which is not in
the address is to be in a table. Look at the difference of finding a place
in NY, Paris and Tokyo. On the network it is the same. Please let stop
considering DNS and IP addresses. There are object location and access
routing information which can be aliased by users for usage simplification,
increased control or added value reasons.
jfc
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- Re: How the IPnG effort was started, (continued)
- Re: How the IPnG effort was started, Carsten Bormann
- Re: How the IPnG effort was started, Kurt Erik Lindqvist
- Re: How the IPnG effort was started, Valdis . Kletnieks
- Re: How the IPnG effort was started, JFC (Jefsey) Morfin
- Re: How the IPnG effort was started, Stephen Sprunk
- Re: How the IPnG effort was started,
JFC (Jefsey) Morfin <=
- Re: How the IPnG effort was started, Kurt Erik Lindqvist
- Re: How the IPnG effort was started, Eric A. Hall
- Re: How the IPnG effort was started, Paul Vixie
- Re: How the IPnG effort was started, Eric A. Hall
- Re: How the IPnG effort was started, Franck Martin
- Re: How the IPnG effort was started, shogunx
- Re: How the IPnG effort was started, Joe Abley
- Re: How the IPnG effort was started, Jon Allen Boone
- Re: How the IPnG effort was started, shogunx
- Re: How the IPnG effort was started, Joe Abley
|
|
|