[ih] Early History of the Internet

[John Day]

Keep in mind that CYCLADES was not intended to be a production network, but a network for doing research on networks. Hence, they started with the idea of doing the least that was possible and then seeing what else was needed and how. In a sense, CYCLADES was the first ‘clean-slate’ attempt. 

Also, CYCLADES assumed that hosts were not located near the CIGALE switches and that a Host could be connected to more than one switch. It supported multihoming as a consequence of the architecture. The Transport Protocol carried its own addresses and had a network address for each switch it was connected to.

They started with fixed routing and moved later to dynamic routing later. Cigale was implemented by Jean-Louis Grangé.  They had also done a contract in 1972 with Univ of Waterloo to investigate routing and congestion.  Eric Manning had a brilliant grad student named Merek Irland. He did a lot of simulations. However, by 1979, when IRIA sponsored a conference on flow and congestion control, he had died of lung cancer. The proceedings are dedicated to him.

Furthermore, the designers of the Transport Protocol (Zimmermann and Elie) had realized that the protocol naturally cleaved into data transfer and feedback functions (ack and flow control). Data Transfer writes the state vector, Feedback reads it and generates the ack and flow control packets. There is very little interaction between the two. During connection establishment, it was possible to choose whether the feedback functions were present. So in effect, CYCLADES Transport included both the UDP and TCP functionality and later didn’t break fragmentation at the gateways.

CYCLADES was exploring a different problem than the ARPANET or the Internet.

Take care,
John Day

> On Jan 9, 2024, at 21:09, Brian E Carpenter via Internet-history <internet-history at elists.isoc.org> wrote:
> 
> Vint,
> 
> Thanks; you were there and I wasn't. I just re-read McKenzie's paper from the Annals about INWG (https://ieeexplore.ieee.org/document/5723076) and I assume it's accurate.
> 
> I've never looked into the details of Cyclades and Cigale. But from what Pouzin wrote in 1972 (https://dl.acm.org/doi/pdf/10.1145/800280.811034, published January 1973), the addressing scheme was rather simple and I guess they had fixed routing tables. He certainly discussed interconnection issues - in fact I find the following insight very perceptive:
> 
> "Inter-network communications have still to demonstrate
> their practical feasibility... In particular,
> any function except sending packets is probably just
> specific enough not to work in conjunction with a
> neighbor."
> 
> That's still true. For example, the Internet is notoriously opaque to newly specified IPv6 extension headers.
> 
> Regards
>   Brian
> 
> On 10-Jan-24 14:15, vinton cerf wrote:
>> Brian,
>> Bob Kahn and I were working on the interconnection of networks starting in 1973. The first I saw the idea of dropping the underlying networks and just using routers with IP knowledge to build a network was when Cisco System announced their router design. Maybe Pouzin had that idea (no distinct network below the gateway) but I think all our implementations tended to have some kind of underlying network. Even the Ethernets had gateways (and ARP for the hosts). I thought Cigale was the packet network and it had an underlying routing system (did it?) and that hosts and gateways would be needed to interconnect the networks. Of course, Pouzin had only the one CIGALE network (kind of like the Arpanet but with fixed datagrams rather than variable length message transport. ) I don't recall that he was working on a multi-network system?
>> vint
>> On Tue, Jan 9, 2024 at 4:00 PM Brian E Carpenter via Internet-history <internet-history at elists.isoc.org <mailto:internet-history at elists.isoc.org>> wrote:
>>    Jack,
>>    Thanks, that's a nice article.
>>    But there is a bit of a category mistake. The APRPANET wasn't the
>>    Internet. I couldn't use the APRPANET in 1971-73 when I took my
>>    first baby steps in networking at CERN. I couldn't use it in
>>    1974-1976 when we tried to set up an inter-university network in
>>    New Zealand, and I still couldn't use it in 1977 back at CERN.
>>    (There were a few lucky users in the UK and Norway by then,
>>    of course, but it was still a network, not a catenet.)
>>    The conceptual leap forward happened, as far as I can tell, in
>>    ~1974 thanks largely to Pouzin and it became reality in 1981
>>    (or a little bit earlier if you admit uucp).
>>    Regards
>>         Brian Carpenter
>>    On 10-Jan-24 06:33, Jack Haverty via Internet-history wrote:
>>     > IMHO, this essay provides a good overview of the early history leading
>>     > to the development of the Internet:
>>     >
>>     > https://www.freaktakes.com/p/the-third-university-of-cambridge <https://www.freaktakes.com/p/the-third-university-of-cambridge>
>>     >
>>     > Professor Licklider was my thesis advisor, and later boss, at MIT, and
>>     > from 1977 to 1990 I worked at BBN in the same group that built and
>>     > evolved the ARPANET.   The history told in the essay agrees with my
>>     > recollections of the time span when I was involved at MIT and BBN.
>>     >
>>     > As told in the essay, Lick's vision of a "galactic network" was using a
>>     > collection of computers, communicating amongst themselves over some kind
>>     > of electronic means, to assist people in doing everything people do.
>>     > That was the mantra that drove creation of the ARPANET, and that we
>>     > tried (are trying) to evolve into today's Internet.
>>     >
>>     > Jack Haverty
>>     > (MIT 1966-1977; BBN 1977-1990)
>>     >
>>     >
>>    --     Internet-history mailing list
>>    Internet-history at elists.isoc.org <mailto:Internet-history at elists.isoc.org>
>>    https://elists.isoc.org/mailman/listinfo/internet-history <https://elists.isoc.org/mailman/listinfo/internet-history>
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history