[ih] OSI and alternate reality

[Brian E Carpenter]

On 16-Mar-24 09:14, John Day wrote:
> Brian,
> 
>> On Mar 15, 2024, at 16:04, Brian E Carpenter <brian.e.carpenter at gmail.com> wrote:
>>
>> John,
>>
>>> CLNP - as the internetwork protocol
>>> TP4 for transport.
>>
>> Really, the critical OSI issue in the mid/late 1980s was that
>> there were two OSI candidates for each of those layers, with
>> European profiles using a connection-oriented internetwork
>> protocol (X.75/X.25) and a dumb transport protocol (TP2).
> 
> Only to the uninformed. ;-) 

Who included most of the people handing out the money, and the
people supplying international leased lines. I remember having
it explained to me (probably in Brussels) that CLNP was an
American plan to take over the universe and destroy the European
computing industry (or words implying that). CERN wasn't popular
among the European GOSIP promoters because we had chosen CLNP
and TP4 as the way to go, long before we became more unpopular for
switching horses midstream to TCP/IP.

> Only the PTTs kept trying to push X.75 but everyone knew it could never sustain a network. That was implicit in 8648. We all knew the PTTs were a dead horse so why waste time on it. From the time of CYCLADES and before, it was clear that the game had to be dynamic resource allocation, which virtual circuit wasn’t doing.
> 
> Yea, the Brits had to push TP2. I have never understood why. Just so they could say they had done something. It wasn’t any more efficient than TP4 under the same error conditions and if those errors occurred (which could be eliminated) TP2 didn’t work and TP4 did. Also TP4 incorporated Watson’s result on synchronization which made it more robust and more secure. The Brits were always a pain. TP4 was a major advance, besides being less overhead.
>>
>> And that was what gave TCP/IP its chance - no competing profiles,
>> plus running code.
> 
> There was running code by the demos in 1985, I believe.

I'm sure there was, but I remember trying to find viable CLNP
code for the Motorola 68000 around that time, and the price
was absolutely ludicrous. (Some small company in Santa Monica
was trying to make megadollars out of US GOSIP.) But by the time
DECNET Phase V was deployable, it was already "game over".

     Brian

> 
> Take care,
> John
>>
>> Another thing I'd mention, which was a direct consequence of the
>> above, is that when Tim Berners-Lee designed HTML and HTTP,
>> there was exactly one deployed transport layer available.
>> (Although the official CERN policy was TP4/CLNP at the time,
>> the CERN reality was TCP/IP.)
>>
>> So, getting back to David's counterfactual question, if we
>> had all simply waited for OSI, Tim would have had no substrate
>> for the Web, it would at best have been a lab project, and
>> none of us would be where we are today. In other words,
>> that was a really major branching point in history.
>>
>> David, I assume you're familiar with Olivier Martin's history
>> of the period? Very relevant to EARN.
>> http://ictconsulting.ch/reports/European-Research-Internet-History.pdf
>>
>> Regards
>>    Brian Carpenter
>>
>> On 16-Mar-24 06:17, John Day via Internet-history wrote:
>>> Catching up.
>>> The core OSI Protocols were:
>>> Ethernet (all 802 standards are ISO standards)
>>> (Network Layer-was intended to be technology dependent and potentially non-standard. This is following the INWG and ISO 8648 model of internetworking.)
>>> CLNP - as the internetwork protocol
>>> TP4 for transport.
>>> Fast-byte for the collapsed upper layers.
>>> ACSE - for creating application connections including authentication and was designed to be recursive.
>>> Then most anything that was proposed could be done (remember standards are bottom up, so if support could be generated for an application it could be done.)
>>> The application layer structure was designed to be modular, so base application protocols could be mixed and matched with supporting protocols. Among the things proposed were:
>>> Virtual Terminal - which was mostly obsolete by then but was on the write track until DEC screwed it up.
>>> FTAM - for file transfer
>>> JTAM - for Job Transfer
>>> CCR - for commitment, concurrency and recovery.
>>> RPC - for Remote Procedure Call
>>> TP - for Transaction Processing.
>>> CMIP - for network management (but really the base application protocol since all application protocols act on objects external to the protocol. The only difference is what the object models are.)
>>>> On Mar 15, 2024, at 09:52, Olivier MJ Crépin-Leblond <ocl at gih.com> wrote:
>>>>
>>>> Totally agree -- hence the only possible evolution from there, which I followed up in my next email:
>>>>
>>>> https://www.bing.com/images/create/a-bright-green-sneaker-with-osi-logo-on-it/1-65f44c2966d7477884c76d7914be2bf6?id=oiufq8FtuQ9dPQT%2f2ETwVA%3d%3d&view=detailv2&idpp=genimg&idpclose=1&thId=OIG3.IQig2FDd_t3cKvUjIr8X&frame=sydedg&FORM=SYDBIC
>>>>
>>>> :-)
>>>>
>>>> On 15/03/2024 13:45, John Day wrote:
>>>>> These were all crap.  X.25/X.75/X.29 were all trying to preserve the PTT monopoly.
>>>>>
>>>>> The PTTs never got (and still don’t) that this was all about dynamic resource allocation, not static allocation.
>>>>>
>>>>> X.400 was far too complex.  X.500 was trying to be the white pages and the yellow pages, when all that was necessary was a simple protocol that mapped application names to network addresses.
>>>>>
>>>>> These were all illustrate how the PTTs didn’t get what was going on.
>>>>>
>>>>> The mistake OSI made was inviting to do the work jointly with CCITT (ITU). However, given that there was no telecom deregulation even being talked about in Europe the Europeans felt they had no choice, especially given the interference they had already shown with EIN and EURONET.
>>>>>
>>>>> The Europeans proved to be their own best enemy.
>>>>>
>>>>> Take care,
>>>>> John
>>>>>
>>>>>> On Mar 15, 2024, at 09:34, Olivier MJ Crépin-Leblond via Internet-history <internet-history at elists.isoc.org> <mailto:internet-history at elists.isoc.org> wrote:
>>>>>>
>>>>>>
>>>>>>
>>>>>> On 15/03/2024 10:19, David Sitman via Internet-history wrote:
>>>>>>> Would we have seen the same rapid and universal adoption of computer
>>>>>>> networking with OSI? Could the Web have flourished? Would address space and
>>>>>>> security issues be alleviated? Would "OSI on Everything" have become a meme?
>>>>>> Having actually used X.3, X.21, X.25, X.400, X.500... sorry, let me re-phrase this... having actually struggled with the aforementioned X. based services and also programmed stacks according to these protocols a loooong time ago, my prediction of a Green Internet based on computing networking with OSI would be resumed as this:
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> Kindest regards,
>>>>>>
>>>>>> Olivier
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> -- 
>>>>>> 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
>>>>
>>>> -- 
>>>> Olivier MJ Crépin-Leblond, PhD
>>>> http://www.gih.com/ocl.html
>