🚀 go-pugleaf

Randall Atkinson <atkinson@itd.nrl.navy.mil> writes:
|
|   Erik,
|
|   Several of your statements above just are not true.  For example,
|
|     * I most certainly *CAN* buy Internet access off the shelf.  In
|       fact there are at least 3 commercial Internet providers locally.
|       Access costs vary with data rates provided and the low end costs
|       for low data rate connections are quite reasonable.  I can even
|       buy off the shelf dial-up IP service using my normal modem.

OK, I have received much mail telling me the same.  I regret to say that
it's a remarkably US-centric comment.  (And that from a US-centric
European!)  Internet connectivity rests on advertising your existing so
that people can reach you, and getting "connected" status with the
biggest player.  That's NSFNET at present.  You can't buy connected
status from NSFNET.  There are other players in the market, and you may
be able to "bypass" NSFNET, but if you can't reach all the connected
hosts on the Internet, you have a network island.  Network islands means
you have to have cables to the each of the other islands, with each
major player in the market, and you end up with prohibitive costs which
effectively shut people out of the Internet.  CCITT and ISO intend with
OSI to give access to all buyers on an equal footing, with predefined
and operational infrastructure that the customers won't have to build
themselves.  The Internet is very far from this situation, and looks
like early telephone history where a big company would have to have
phone lines to each carrier to be reached by all his customers.

|     * There are several places on the Commercial Internet (the set of
|       commercial IP networks connected via the Commercial Internet
|       Exchange or CIX) that RFCs are available online.  The host
|       ftp.uu.net is one obvious example of a source for RFCs.

CIX is the way to go.  However, it's not very relevant where you can
find RFC's _after_ you're hooked up.  I thought I already covered that.

|     * The commercial Internet connects to far more places than you
|       seem to realise.  The existence of a commercial backbone means
|       that lack of access to the NSFnet backbone isn't nearly as much
|       of a problem as it used to be.

I know.  I've been watching a major information source in Oslo find a
network supplier and have lots of inside information on their ordeal in
looking for one, and discarding several over-priced and under-featured
suppliers.  I'm waiting for the time when access to the NSFNET backbone
has ceased to be a problem altogether.  As long as someone you want to
talk to is only reachable through the NSFNET, you need to have a cable
to his mid-level to talk to him, and that's the infrastructure that each
customer is required to pay for that I mentioned above.

|     * The telephone number of the DDN Network Information Center is
|       published (you can ask directory assistance for it) in normal
|       ways just like ISO's phone number is available.  Within the US
|       and Canada, there is even a toll-free 800 number.  Also, SRI
|       will be happy to take your order from their private NISC and
|       they will happily take your credit card.  The cost of getting
|       printed copies is quite small, CD-ROM copies are also available.
|       ANSI and ISO and CCITT, by contrast, will NOT take my credit
|       card and have outrageously high costs.

CCITT doesn't have outrageously high costs.  They're very inexpensive
for such a big organization.  IBM is also have ridiculously inexpensive
documentation compared to ISO.  Beware of copies of ISO standards where
you pay for both duplication and processing costs as well as the ISO
royalty.  I've seen copies cost twice to three times that of the
original.  The same goes for ANSI standards.  These things are not very
expensive if you know where to shop.  Somehow, I think you've given up
before you found a good supplier.  The fact that you can choose over-
night delivery anywhere in the world from one supplier and have to wait
three months from another is just part of the free market.

To ask a pointed and slightly pointless question: Where's the European
(or Norwegian) subsidiary of NISC?  Can I go and "preview" the standards
I'm looking for before buying them?  I can with most ISO and CCITT
specifications.  ANSI is trickier, but I did visit ANSI in New York.

The U.S. of A. is incredibly important in the information technology
business, but even I can begin to appreciate the desire among OSIfists
for a non-US set of standards with the line of argumentation I'm getting
here.  Face it, the world is more than the U.S., and as long as people
outside of the U.S. try to do anything in the Internet, they meet all
sorts of barriers and hindrances.  "It works for us" has never been a
good criterion for quality.

|     * There are no restrictions on email on the Internet and so folks
|       without any Internet connectivity can get RFCs free using the
|       RFC-by-mail servers run by DDN NIC or NSFnet.  Commercial email
|       providers (e.g. MCI Mail) provide access to Internet mail and if
|       you have anything that talks UUCP you can get access essentially
|       for the cost of a modem.

Well, this is probably great if you're already on the network, somehow.
It's getting on it which is the hurdle, remember?  We have some really
anal-retentive UUCP suppliers in Europe who make it almost impossible to
get decent service.  Monopolistic, monolithic service providers in the
worst tradition of the government-run PTT's, that's what we've got to
fight with over here.  Don't even mention IP to them, because they're
_so_ happy with OSI they just squeal with delight.  (OK, so some of them
also have brains, and have seen the light, but far from all.)

|     * In summary, it is easier and cheaper for folks not on the net to
|       get paper copies of RFCs than it is for them to get paper copies
|       of ANSI, ISO, or CCITT documents.

Easier depends on the location of the eye of the beholder, I guess.
Cheaper?  No argument with that.  Unless you count ECMA, which sends you
all the stuff you want for free.  (Yes, that _can_ work, but only
because ECMA has its (commercial world) members pay for the secretariat
and the dissemination of standards.)

|     * If ANSI, ISO, and CCITT would make their documents available
|       online as RFCs already are, then there would probably be more
|       people implementing the specs and greater interoperability
|       amongst independent OSI implementations (there is near zero
|       interoperability between two independent implementations right
|       now, much though we might wish it were otherwise).

I don't think the interoperability problems would be solved, because the
specs are chock full of options and ambiguities which can't be resolved
just like that.  Of course, these options and ambiguities come from the
process of having people sit in committees and decide things without
knowing what what they talk about will result in in practice, because
they don't have interoperating implementations to play with.  This comes
partly from the size of the problems they have chosen to fight, and
partly from the very real threat of having someone get a financial head
start.  That's also why they couldn't choose TCP/IP, because European
manufacturers would do almost as much as George Bush would do to win the
election to stop the U.S. proposals, because of the "unfair advantage"
of U.S. router manufacturers.  (Not that the U.S. router manufacturer of
choice didn't win the OSI routing competition, but at least they can go
back to their respective sponsoring companies and say "I tried my
best".)  OK, so I'm a cynic, but that doesn't change the attitude of
several European companies and the moronic politicians who sponsor the
design of government OSI profiles.

Now, whether we would have more implementations just because the
standards would be free is dubious at best.  The conclusion from "RFC's
are free, RFC's are widely implemented" to "free -> widely implemented"
doesn't hold water.  (D)ARPA sponsoring of University projects helped a
lot to boostrap this process.  There's nothing like that for the OSI
stack.  Also, the working documents are available almost for free and
are free (at least over here) if you get involved in the process.
Another major difference between TCP/IP and OSI is that TCP/IP evolved
in academia (due to heavy sponsoring), while OSI evolved in the
commercial world.  There are so many factors involved here that the
inherent and superior value of free standards is a pipe dream.

Now, all of this said, please keep a note in your Rolodex or whatever
that I'm solidly entrenched in the TCP/IP world myself, and consider
most of this OSI stuff to be utter nonsense.  This has nothing to do
with the availability of standards, but with their genesis and the level
of political involvement in the OSI process.  Add to this a footnote on
my skepticism towards having CCITT mess with user applications, based
mostly on being very skeptical of monolithic, monopolistic organizations
deciding what people need.  (Add another note stating that I have the
highest respect for CCITT's knowledge and expertise on OSI layers 1, 2,
and 3, in that order.)

This, however, does not mean that I have to refuse to understand how
things operate on both sides of the Berlin wall between academia and the
commercial world.  I wish more people could understand 	it, and also try
to understand the much more global context that CCITT and ISO have to
work with.

I hope this article has shed more light than heat, because I got pretty
worked up when I started this thing.

Best regards,
</Erik>
--
Erik Naggum             |  ISO  8879 SGML     |      +47 295 0313
                        |  ISO 10744 HyTime   |
<erik@naggum.no>        |  ISO 10646 UCS      |      Memento, terrigena.
<enag@ifi.uio.no>       |  ISO  9899 C        |      Memento, vita brevis.