The Internet had not reached my part of the UK in the 1970s.
From my perspective 'The Internet Architecture' is primarily what
emerged and what succeeded. And in particular what differentiated the
Internet from competing schemes. I have however heard Vint (amongst
others) state on at least two occasions was that the one design
principle that was constant was the insistence that a network of
networks was not merely a network and that IP be the only inter-domain
routing protocol.
There were plenty of networks offering file transport. And there were
plenty offering some form of email. The features that differentiated
the Internet over other packet switched networks were:
* Independent administration - no flag days, no central control
* Open Standards (not controlled by one manufacturer)
* Single address space / Single name space
* Large address space (> 16 bits)
* DNS
* Store and forward email (* copied by others)
I do not believe that the concept of end to end pure IPv4 was either a
differentiator or a factor in the success of the Internet. What was a
distinguishing factor was having a large enough address space that
initially every host and later as IPv4 exhaustion took place every
server could have a unique IPv4 address in a single address space.
What worries me about certain assertions being made with respect to
IPv6 transition is that some people seem to have made IP-purity an
absolute requirement while ignoring the principle of independent
administration.
As for FTP, I think we have a really different idea of what is meant
by 'non-specialist'. My meaning would be a secretary/typist.
My criteria for usability come from having worked in the video game
industry. A video game is a program which people will pay to interact
with.
On Thu, Sep 2, 2010 at 4:42 AM, John C Klensin <klensin(_at_)jck(_dot_)com>
wrote:
--On Tuesday, August 31, 2010 17:36 -0400 Phillip Hallam-Baker
<hallam(_at_)gmail(_dot_)com> wrote:
Surprising as it may seem, I was aware of the prior existence
of FTP and Telnet.
I actually assumed that. Where I think there was a disconnect
is in your understanding of how they were used and by whom. See
below.
The point I was trying to get to was email was the only
application that was useful to people beyond the type of
people who were building the Internet. Hence for the purposes
of the original discussion, the fact that SMTP was the only
protocol that is not end to end pure IPv4 indicates that it is
a paradigm to follow rather than a mere anomaly.
But the "useful to people beyond the type of people who were
building..." conclusion is just not true.
* Telnet (and my old friend SUPDUP) were used extensively by
non-specialists for remote online access to computing resources.
* FTP was used by a number of communities to access online
repositories of data and documents, not just programs.
* I spent much of the 70s and 80s in design and management roles
in a sequence of projects and consulting activities involving
computer applications for (in different projects) social and
behavioral scientists of various types, architects and urban
planners, nutritionists and food chemists, folks doing strategic
management and planning for various resources, etc. -- none of
those people being what I assume you mean by "the type of people
who were building the Internet". Many of those projects were
using the Internet, some for remote access, some for data
repositories, some for distributed computing activities that
quietly built and ran their own protocols to meet the needs of
their applications. They made extensive use of email too --in
some cases, taking the advantage of the one feature that makes
email almost unique among application-level Internet protocols,
the ability to relay and queue through multiple systems and
multiple transport protocols.
The thing that made email special --especially subsequent to RFC
974-- was a clean, end-user-transparent, model for routing email
traffic into and out of systems and networks that were not using
TCP/IP or connected to the Internet at the IP layer. That
architectural property of email was extremely important in
enabling message communications among a diverse collection of
networks and network-like arrangements (e.g., systems that
received email, printed it, and got the paper into envelopes for
delivery came into being in both commercial and some
academic/research settings). The same queue-and-relay
architecture was of immense help in dealing with getting
messages (at least) through to systems that, while directly
connected to the Internet and running TCP/IP to endpoints, were
behind relatively unstable or intermittent links. But those
features were a piece of email architecture that functioned to
work around systems that were not end-to-end and permanently
connected to the Internet, not design features of the Internet
architecture as your notes seem to suggest.
john
--
Website: http://hallambaker.com/
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www.ietf.org/mailman/listinfo/ietf