[ih] ARPANET pioneer Jack Haverty says the internet was never finished

[John Day]

Jack,
Only now getting around to responding to this. 

BBN was correct. It wouldn’t work with just that. TCP was based on the work in 1971-72 by CYCLADES at IRIA in France, now INRIA.

Having seen the ARPANET, CYCLADES was building a network to do research on networks. Dave Walden spent some time working with them. The team’s plan was to determine the minimal assumptions to get a packet through a network. Then their research program was to see what else was needed. Start with the simplest and see how much more was needed. (What we would call a clean-slate approach today). What they found was that a 'best effort' datagram with an end-to-end transport in the hosts did everything at the time but two open questions: routing and congestion control. Of course some work had been done on routing but we are still looking for better ways. They recognized immediately that congestion would be an issue and began work on that. (Remember Baran had laid had said there would distinct advantages to a network dedicated to data, rather than voice. It wouldn’t be until much later that more would be needed when that view was relaxed.)

Around 1972 or so, CYCLADES awarded a contract to the University of Waterloo to do research and simulations on those two topics. The primary people were Merek Irland, his advisor Eric Manning and a few others.  They got some interesting results which were factored into the CIGALE network implementation. (CIGALE were the switches for CYCLADES.)  However, CYCLADES was shut down in the late 70s and Irland died of lung cancer in ’78 and as near as I can tell the work was forgotten.

Irland did publish results at the 4th Data Communications Symposium in 1975 in Quebec and was co-author on two papers at a IRIA sponsored conference, Flow Control in Computer Networks in 1979. (The proceedings are dedicated to Irland’ memory.) There are a couple of other papers by him, but mostly there are his thesis and reports at Waterloo and INRIA, so far as know none of which are on-line.

Take care,
John

> On Mar 2, 2022, at 23:32, Jack Haverty via Internet-history <internet-history at elists.isoc.org> wrote:
> 
> IMHO, many things also happen for non-technical and non-business reasons.  Since multicast was needed for some uses of the 'net, but it didn't actually get deployed widely in the Internet (whatever happened to the Mbone...?), people figured out another way to provide it by putting it in separate boxes (the CDNs) from the switches themselves.
> 
> I've always wondered if that same pattern drove the creation of TCP and use of datagram mode.   The ARPANET was the only WAN of the day, and its gurus were extremely reluctant to allow use of "uncontrolled packets" (aka datagrams) for fear of bringing down the whole network.   I recently found a 1975-era BBN report analyzing the TCP proposal and concluding for DCA that it couldn't work.
> 
> So TCP was implemented in the host computers, where mere mortals could get at the code.   Of course, TCP mechanisms duplicated the mechanisms already in the ARPANET.   That's what I meant by "moving mechanisms from switches to hosts:.  But that did enable us a few years later to simply interconnect routers with wires, cutting the ARPANET out of the picture.
> 
> Jack
> 
> On 3/2/22 20:03, John Levine via Internet-history wrote:
>> It appears that Noel Chiappa via Internet-history <jnc at mercury.lcs.mit.edu> said:
>>>    > On Tue, Mar 1, 2022 at 8:46 PM Jack Haverty wrote:
>>> 
>>>    > One that I used in the talk was TOS, i.e., how should routers (and TCPs)
>>>    > treat datagrams differently depending on their TOS values.
>>> 
>>> I actually don't think that's that important any more (or multicast either).
>>> TOS is only realy important in a network with resource limitations, or very
>>> different service levels. We don't have those any more - those limitations
>>> have just been engineered away.
>> That's not it, they came up against the impenetrable barrier of a
>> business model. We understand how to price peering and transit of
>> traffic where all packets are the same, but nobody has any idea how
>> you do it where some packets are more valuable.
>> 
>> I never figured out why multicast failed.  It is bizarre that people are dumping
>> cable service which has 100 channels multicast to all of the customers in favor
>> of point-to-point service where you frequently have a zillion people streaming
>> separate copies of the same thing, e.g., a football game.  We fake it with CDNs
>> that position servers inside retail networks but really, it's multicast.
>> 
>> R's,
>> John
> 
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history