Technically, the Internet is the endpoint of a series of
steps. What makes it stand apart is not the technology, it
is that the technology was non-proprietary.
There were several ways of structuring bits across a
network. Many proprietary: remember SNA, BNA, DDCMP, etc.
In that evolution, a very important progress was the quasi
simultaneous creation of bit-based protocols (as opposed to
byte-, character- or word-based ones), and the use of
"envelopes" to separate control and actual data (counter-
examples: IBM BSC, Burroughs Poll/Select, etc.).
The CCITT (now ITU) was investing a lot in the 7-layer OSI
model, which was intended to be for computers what the
telephone is for humans. There still are some parts of that
work floating around (e.g., X.400, X.500/LDAP at the higher
layers), and some that just died (X.25 message-switching and
the Minitel).
The CCITT work was impressive and hugely complex, as it
tried to do it all for everybody and was not assuming
English (and Latin-1/ASCII) as a universal given. For
parallel reasons, CCITT protocols were emphasizing
connection-oriented traffic; establishing a connection was
quite a monumental affair, involving subprotocols and
languages/conventions (ASN.1 comes to mind).
DARPA was, on its side, not planning an universal fabric.
Its goal was to interconnect computers from unrelated
vendors, mostly for C&C purposes but, while they were at it,
simple tasks like file transfers, expecting humans to do the
bulk of assigning meaning to the conversations. Bits were
bits. If bytes, they were expected to be ASCII, the 8-bit
variety - but that wasn't even necessary. Not 7 layers:
with these simplified objectives, 4 were amply sufficient.
So the IMP (interface message processor) was conceived and
built, and eventually debugged. It used IP - because "IP
(works) on everything" ((c) V. Cerf). TCP was added on top
of IP to speed up file transfers. And IP grew; we're using
version 4 and trying to move to 6.
The decisive factor that cemented the "win" of the DARPANET,
now Internet, over the CCITT approach was that it was put in
the public domain, unencumbered. Another factor was that
the IP protocol is expandable through the IETF, which is
much less constraining and heavy than CCITT or OSI
equivalents. Not mentioning: waaay simpler to implement.
Now we at the same time enjoy and lament that, as the pat
taken by the Internet protocols emphasizes raw performance
over any concern for security; the CCITT was way more
cautious and expected some semi-centralized control with
more power than the ISPs have.
All of that took years. CHILL ~1972; Actual X.25 networks,
~1979; I still remember active commercial projects based on
OSI in 1992 (CTOS/BTOS' FTAM), when "http://" was beginning
to show everywhere.
--------
The above, as memory serves. Finding appropriate references
is left as an exercise to the reader :-)
---
bck[ Reply to This | Parent | # ]
|