[comp.protocols.misc] On the TCP/IP versus ISO "debate"

ken@gvax.UUCP (06/10/87)

Many groups, like my ISIS project, The Ameoba project, and the V project
at Stanford see a need for broad support of protocols that conform to
the ISO standards but go beyond the basic TCP/IP stuff (for example, to
include multicasting).  This has lead to a bit of a debate, and as a result
DARPA has asked the DSAB to set up a task force on protocols, headed by
Dave Cheriton.  People interested in this issue may want to follow the
work of this task force.  We have met once so far, and will meet again
later this year.

To simplify dicussion, I think people should keep in mind that the ISO
standards really don't specify any particular protocol and hence it doesn't
make sense to talk about ISO versus TCP/IP.  At the same time, there is
a need to think about TCP/IP versus XNS, for example, and also to try and
consolidate the many "standards" that are being created for session level
protocols and for multicasting before things get out of hand.  I don't
want to get drawn into a long bboard discussion, but at some point in the
future the task force will certainly produce an RFC that people could
get to fairly easily and perhaps a copy could be posted to this news group too.
An RFC is a request for comments, so comments would be appropriate then.

My personal interest is in seeing less of an emphasis on RPC and more of
an emphasis on multicasting and reliability issues.  We really need to
watch out or we will have to live with a generation of networks that are
really good at RPC just as we all discover that reliable multicasting is
where its really at... (this is a red herring).  Other members of the
task force have their own biases, but I think we cover quite a spectrum of
interests reprsenting industry as well as academics.  Stay tuned, there will
be interesting stuff to debate.

Ken Birman

mrose@GREMLIN.NRTC.NORTHROP.COM (Marshall Rose) (06/11/87)

> To simplify dicussion, I think people should keep in mind that the ISO
> standards really don't specify any particular protocol and hence it doesn't
> make sense to talk about ISO versus TCP/IP.

    Uh?  What?  Wrong.

    ISO standards specify gobs of protocols.  Where do you want to
    start?

	ISO Transport Protocol (ISO 8073)
	ISO Session Protocol (ISO 8327)
	ISO Presentation Protocol (ISO 8823)
	ISO File Transfer, Access, and Management (ISO 8571)

    These are four.  I've got over fifty sitting back in my office, and
    I'm a very selective reader!  

    It makes a great deal of sense to talk about the ISO protocol suite
    vs. the TCP/IP protocol suite.  Although they both have layered
    architectures, they cut up the cake differently.  Further, the
    protocols they use diverge significantly once you start climbing to
    the higher layers.

/mtr

yamo@AMES-NAS.ARPA (Michael J. Yamasaki) (06/11/87)

> From: ken@gvax.cs.cornell.edu  (Ken Birman)
> To simplify dicussion, I think people should keep in mind that the ISO
> standards really don't specify any particular protocol and hence it doesn't
> make sense to talk about ISO versus TCP/IP.  At the same time, there is
> a need to think about TCP/IP versus XNS, for example, and also to try and
> consolidate the many "standards" that are being created for session level
> protocols and for multicasting before things get out of hand.  I don't

Whew.  Perhaps, I'm confused.  In my world there is a thing called the ISO
Reference Model.  This is a thing which describes a seven layer "onion", uh,
er, encapsulated model in which the layers are theoretically interchangeable.
According to marketing literature in the industry, everything follows this
model - TCP/IP, Ethernet, SNA, DECNET, XNS, HYPERchannel...

There is this other thing called ISO Protocols which generally refers to a
collection of protocols some refered to by the CCITT numbers X.this and
X.that, some refered to NBS or IEEE numbers, some are even referred to by
such names as MAP or TOP or ISDN, but the real ISO protocols are 
ISO WXYZ numbers usually followed by caveats to their draft status or 
acceptance level, they also have names, like FTAM, TP4, TP0, and JTM.

All of the above are refered to as ISO.  Hmmm.

                                           -Yamo-

lamaster@pioneer.arpa (Hugh LaMaster) (06/11/87)

In article <8706111658.AA14605@ames-nas.arpa> yamo@AMES-NAS.ARPA (Michael J. Yamasaki) writes:

>
>Whew.  Perhaps, I'm confused.  In my world there is a thing called the ISO
>Reference Model.  This is a thing which describes a seven layer "onion", uh,

This is actually the OSI Reference Model (Open Systems Interconnect).  ISO is
the International Standards Organization.  It is confusing.  But OSI does not
specify any protocols.  There are many ISO protocols however.

*******************

Moving on to something else, I have been looking at the ISO protocols the last
few days, and it finally occurred to me exactly what I didn't like.  The
problem, for an internet user, is that X.25 is not, and cannot ever be, a
complete layer 3 protocol.  In fact, it could be a layer 2.5 protocol, which
is sort of the way that the current Arpanet uses it: as the host interface
protocol, which IP sits on top of.  Now, I haven't seen any ES-IS proposals,
so I can't legitimately complain about it, but the published intention was to
write it as a GATEWAY to X.25, which I think is entirely incorrect.  Instead,
X.25 should be moved to its proper place in level 2 and be one of the many
media that IP can run on top of.  Then, in layer 3, there needs to be a
definition of "The Internet Addressing Standard" as applied to the new ISO IP,
with re-implementations of name service, domain name service, gateways, etc.
X.25 might be used by some gateways to talk to each other, or, the gateways
might use something else (a lot of people are putting TCP/IP on T1 now; is
there a "standard encapsulation"?).  Comments?




  Hugh LaMaster, m/s 233-9,  UUCP {seismo,topaz,lll-crg,ucbvax}!
  NASA Ames Research Center                ames!pioneer!lamaster
  Moffett Field, CA 94035    ARPA lamaster@ames-pioneer.arpa
  Phone:  (415)694-6117      ARPA lamaster@pioneer.arc.nasa.gov

("Any opinions expressed herein are solely the responsibility of the
author and do not represent the opinions of NASA or the U.S. Government")

kre@munnari.UUCP (06/20/87)

In article <1744@ames.UUCP>, lamaster@pioneer.arpa (Hugh LaMaster) writes:
> ISO is the International Standards Organization.

No its not.

ISO is the "International Organization for Standardisation"

its also "Organisation Internationale de Normalisation".

Its also something else in a language I don't recognise (probably
Russian) and which is in a script I can't copy (ascii doesn't
have the characters).  (Exposing my ignorance of culture here.)

Understanding this fine point is *crucial* to understanding ISO,
and the way the OSI protocols have been developed.

OSI *cannot* be the International Standards Oriagnzation, because
that would make it an acronym in English, but not in French, Russian,
German, Japanese, ...

International politics simply won't allow this kind of bias, all things
must be equal, and fair to everyone, and seen to be.

kre

ps: also remember that ISO standardizes more than computer protocols.
There are ISO standards for more things that you can imagine - how do
you think they got to numbers like 9542 (the network routing protocol)?
All of these standards have been adopted under the same kinds of constraints.

lamaster@pioneer.arpa (Hugh LaMaster) (06/23/87)

In article <1712@munnari.oz> kre@munnari.oz (Robert Elz) writes:

>In article <1744@ames.UUCP>, lamaster@pioneer.arpa (Hugh LaMaster) writes:
>> ISO is the International Standards Organization.
>
>No its not.
>
>ISO is the "International Organization for Standardisation"
>

I stand corrected.  What is worse, I have made this mistake several times
before.  However, I don't think it is very important.  The meaning
(semantically speaking) is IDENTICAL.


>Understanding this fine point is *crucial* to understanding ISO,
>and the way the OSI protocols have been developed.

It DOES make sense for an international organization to use multiple languages
because it shows that the organization is not NECESSARILY biased towards the
interest of one particular organization or country.  It does not mean that the
process of standardization with respect to some particular set of standards is
"fair to everyone".  Furthermore, what did ISO care about being "fair" when it
adopted wholesale CCITT protocols that don't fit the OSI model and aren't
interoperable with other protocols in the suite?  I'm not saying ISO or
politics are bad.  I'm saying ISO is political and politics isn't "fair".

>
>OSI *cannot* be the International Standards Oriagnzation, because
>that would make it an acronym in English, but not in French, Russian,
>German, Japanese, ...

I am a chess player.  The international organization which handles master
level chess and above is FIDE which stands for something in French, but
nothing in English.  That doesn't bother me at all.  What bothers me is that
I'm not a grandmaster :-(

>
>International politics simply won't allow this kind of bias, all things
>must be equal, and fair to everyone, and seen to be.
>

What does ISO stand for in Tagalog or Bahasa Indonesia or (Polynesian) Fijian?
I guess ISO isn't fair to native speakers of these languages then, is it?

Personally, I would be perfectly happy to learn Esperanto or some other
synthetic belongs-to-no-one language for international use (and even usenet
use :-)   ); now THAT would be FAIR.  Unfortunately, the owners of English,
French, Russian, (Spanish, Chinese, Arabic, ??? I forget) and the other
"official" languages of the U.N. (That is United Nations in English) didn't
want to either use Esperanto or support the development of a new language for
international use.  But then, maybe we should be happy we don't have an UN
designed language.  It would probably be layered, and come in several mutually
incompatible subsets.  We might not even be interoperable ;-)


  Hugh LaMaster, m/s 233-9,  UUCP {seismo,topaz,lll-crg,ucbvax}!
  NASA Ames Research Center                ames!pioneer!lamaster
  Moffett Field, CA 94035    ARPA lamaster@ames-pioneer.arpa
  Phone:  (415)694-6117      ARPA lamaster@pioneer.arc.nasa.gov

("Any opinions expressed herein are solely the responsibility of the
author and do not represent the opinions of NASA or the U.S. Government")