[ih] The UCLA 360/91 on the ARPAnet/Internet

[Vint Cerf]

Jack this is helpful!

I am reasonably sure that Bob Kahn nudged me to create a "cabinet" in
1979 and that I did do so - but it is possible that the group was not
convened separately and  formally until 1981. Plainly we would be in
planning mode for the big cutover by that time.

Berkeley BSD4.2 had the first TCP/IP code in it from Bill Joy who did
not use the BBN code.

Kirstein chaired the International Coordination Board (ICB) rather
than serving on the ICCB. ICB was notably focused on SATNET access to
the ARPANET hosts using TCP/IP at least a year ahead of the big
ARPANET cutover.


v



On Sun, May 13, 2012 at 6:45 PM, Jack Haverty <jack at 3kitty.org> wrote:
> [Please preface every sentence with "As far as I can remember..."]
>
> I did a little garage archaeology and found my notebooks from the late
> 70s/80s.  The earliest mention I found of "ICCB" was notes from the
> ICCB meeting of 9/21/1981 at UCL - the day before the full quarterly
> Internet Group meeting, which by then had become quite large.
>
> Since I don't have any ICCB notes from the previous Internet meeting
> at COMSAT in June 1981, I suspect that the UCL meeting was the first
> ICCB meeting.  Vint had asked a small group of people who also
> attended the regular Internet meeting to come a day early and help him
> brainstorm some longer-term and architectural issues, which were
> increasingly difficult to do in the ever-larger Internet meeting.  But
> the 1981 date conflicts with Vint's 1979 date.  Perhaps the ICCB
> started earlier, but I think I recall being at the formative meeting
> where the term "configuration control" was selected.  Maybe someone
> else knows more....?
>
> I recall that the name "Internet Configuration Control Board" was
> explicitly chosen to make the activity sound boring and unattractive -
> otherwise everybody would have wanted to be there.  This had already
> happened in the prior working groups which had gotten unwieldy.
>
> As I recall, the ICCB membership was:
> Vint Cerf - DARPA
> Ed Cain - DCA/DCEC
> Ray McFarland - DoD
> Jim Mathis - SRI
> Jon Postel - ISI
> Bob Braden - UCLA
> Dave Mills - Comsat?  Udel?
> Dave Clark - MIT
> Steve Kent - BBN
> Jack Haverty - BBN
>
> I can't recall whether or not Peter Kirstein and/or John Laws was
> involved.   Danny Cohen and Dave Reed were not on the ICCB.   There
> were many meetings in various Internet-based projects with highly
> overlapping membership, so it's hard to remember who was in what
> groups any more.   Maybe a little more garage archaeology will help.
>
> I don't think Dan Lynch was on the ICCB, but he was everywhere so I
> could be wrong.   He did solve one of Vint's problems rather neatly.
> Everyone wanted to go to the Internet meetings, so it became difficult
> to "get a ticket" from Vint to attend.  Dan noticed this, and being a
> true entrepreneur solved it by booking a conference center and
> charging hundreds of dollars to attend - plus inviting and encouraging
> all of the regular Internet meeting denizens to present papers, etc.
> Problem solved.  As more and more people attended, it just required a
> bigger and bigger conference facility.  That's how the Interop shows
> got started.
>
> The ICCB continued as a regular meeting colocated with the expanding
> Internet meeting, acting as a sort of steering committee/advisor for
> Vint, and we could go back after the meetings to our various
> organizations and try to get the whole crew to head in the same
> direction based on the ICCB consensus.
>
> I've attached a scan of my notes from that first meeting.   If anybody
> can read my horrible handwriting, they might prove interesting.  The
> motivation for the ICCB seemed to be the need to plan out the "January
> 1983 System" which would be able to support "heavy load".    That of
> course turned out to be the milestone when the Arpanet was converted
> to TCP, and the transformation of the research Internet into the
> operational service net.  At some point along the way, the ICCB "came
> out of the closet" and was renamed the Internet Activities Board.
>
> I'm pretty sure that no one had any idea that this would lead to what
> we have today....we would have run away screaming in disbelief!
>
> /Jack
>
> PS - my introductory role to the Internet in the 1977/78 timeframe was
> to implement TCP 2.5 for a PDP-11/40 running Unix, using Jim Mathis'
> implementation for the LSI-11 as a base.  AFAIK, that was the first
> TCP implementation for any Unix system.  Because the 11/40 was so
> limited, my implementation was done in user address space, which
> severely hampered performance.  Mike Wingfield subsequently did an
> implementation in C for the PDP-11/70, and Rob Gurwitz did one for the
> VAX.  Rob's code was supplied to Berkeley for incorporation in BSD,
> but whether they used it in the BSD TCP for anything other than a
> bookend is unknown.   John Sax did TCP for the HP-3000.  Bob Hinden
> did TCP for the Arpanet TIP/TACs.  Bill Plummer did the PDP-10 TOPS
> and Tenex implementations.   All of these were done at BBN
>
> On Sun, May 13, 2012 at 3:10 AM, Vint Cerf <vint at google.com> wrote:
>> it was international and it did include louis pouzin. However, the
>> TCP/IP development was undertaken first by bob kahn and me (and we
>> briefed INWG in Sept 1973 at University of Sussex), then by my group
>> at Stanford University during 1974 (including yogen dalal, carl
>> sunshine, dick karp, judy estrin, jim mathis, darryl rubin and seminar
>> attendees john shoch and occasionally bob metcalfe. Gerard LeLann came
>> from Louis Pouzin's group for a year; Dag Belsnes from Univ of Oslo,
>> Kuninobu Tanno from Japan, Paal Spilling from NDRE; I am sure I have
>> left out a few others); and then Ray Tomlinson and Bill Plummer at BBN
>> as well as Peter Kirstein and his group at UCL (there is a long list
>> here but I can't reproduce it from memory) in 1975. In 1976 we start
>> seeing more implementations and tests - the big one in Nov 1977 with
>> all three networks. We generated Internet Experiment Notes. I don't
>> think we had a name for the group of implementors sponsored by ARPA.
>> By 1979 we are well on the way to standardizing version 4 including
>> the split. By 1980 or so, BBN and Berkeley are working the Unix
>> version; ultimately BSD 4.2 is released with TCP/IP by Bill Joy (among
>> others). I don't recall exactly when you did the IBM 360/91 and 360/75
>> versions but it must have been 1976 or later? Dave Clark did his IBM
>> PC version probably around 1980? Jim Mathis did a version for the DEC
>> LSI-11/23 that we used for the packet radio testing in the 1976-1980
>> period. Bob Kahn urged me to create the ICCB, which I did in 1979 with
>> Dave Clark as chair. After I left ARPA, Barry Leiner assumed
>> responsibility for further Internet development and created the
>> Internet Activities Board again with Dave Clark in the chairman's
>> post.
>>
>> As for the group that did the original tcp/ip design, implementation
>> and testing, I think the principals were on the ICCB  - so that
>> included Bob Braden, steve kent (security - BCR project w/NSA and
>> DCEC), Dave Clark, Dan Lynch, Jon Postel, Jack Haverty, Dave Mills,
>> who else? Danny Cohen and David Reed were proponents of splitting off
>> IP but I don't think they were on the ICCB (boy, memory is hazy). I
>> don't remember whether Ed Cain was on the ICCB but he was the active
>> technical proponent of TCP/IP at the Defense Communications
>> Engineering Center in Reston and was involved in the testing of the
>> BCR packet Encryptors. Ray McFarland was the primary contact at NSA
>> for BCR and for the Internet protocol development starting around
>> 1975, if memory serves.
>>
>> regarding the term "Internet" it was applied to RFC 675, December
>> 1974, the first full TCP spec that had three authors: vint cerf, yogen
>> dalal and carl sunshine.
>>
>> i am copying the history list hoping they will add to this summary
>> and, in particular, pick up names I've missed.
>>
>> vint
>>
>>
>> On Sat, May 12, 2012 at 11:32 PM, Robert Braden <braden at isi.edu> wrote:
>>> Vint,
>>>
>>> I had the idea that INWG was international and included eg Louis Pouzin.
>>>
>>> There was a group of ARPA contractors and a few others ( e.g. , ??? from
>>> DCEC) , which I think you formed and which you certainly led,
>>> that worked out the TCP/IP protocol specs. You subdivided it into the TCP
>>> sub-group (to which you assigned me) and the IP sub group. From this
>>> group came 5 (or 6?) prototype implementations of the developing TCP
>>> spec. What was this group called? I don't think we had settled on the
>>> term "Internet" yet; I recall an ICCB meeting where that issue
>>> as settled.
>>>
>>> I have never read any recognition of this group, nor seen its membership
>>> recognized.
>>>
>>> Bob
>>>
>>>
>>>
>>>
>>>
>>> On 5/12/2012 12:45 PM, Vint Cerf wrote:
>>>>
>>>> i think we settled on "international network working group" (INWG) in
>>>> October 1972 but IEN 48 was titled "The Catenet Model" as I recall -
>>>> and credit was given to Louis Pouzin and his group for inventing that
>>>> term.
>>>>
>>>> v
>>>>
>>>