[ih] GOSIP & compliance

[Karl Auerbach]

We ought not to forget the force that was the Air Force's "ULANA" 
(Unified Local Area Network Architecture) program of the mid 1980's.

It was a quite large procurement for the time, and it pushed for 
commercial, off-the-shelf, components, mostly TCP/IPv4 based.

ISO/OSI stuff had been published.  I had done some ASN.1/BER stuff which 
ended up turning being incorporated into my company's (Epilogue 
Technology) SNMP engine.

GOSIP was lurking and forming in the background, but the Air Force 
essentially ignored it.

I worked on the TRW team bid.  We won, but AT&T protested and the whole 
program ended up going down in flames.

During the preparation of our response we pushed hard on some very new 
and small companies - such as Cisco (which was then operating out of a 
garage), etc.  (Including my own company for SNMP engines.)

Because it was COTS we got TRW to design and build for us a smart 
ethernet board for PC's - to do essentially on Ethernet what the Sytek 
boards did for Netbios on the Sytek ring networks.  To be COTS we had to 
sell it to the public.  Don't know whether anybody bought any.

The bid required that things would work together, and this was before 
the Interop show network, so we created fairly extensive 
interoperability testbeds.  This created a powerful forcing function 
that really underscored that vendors needed to make sure that their 
products played nice with the competition and other pieces of the net.

It's been my feeling, as well as the feeling of others who I worked with 
on the project, that ULANA was one of the important things that made TCP 
based systems commercially viable, and at the same time created a "prove 
to me that it works and interoperates" silver bullet that eventually 
helped to bring down the GOSIP momentum.

On the first couple of years of the Interop show network we deployed 
TCP/IP, ISO/OSI, XNS/Netware, and Decnet.  It was a maddening collection 
of diversity.  The TCP/IP based stuff came together on the shownet quite 
well, as did Netware and Decnet. But we always had trouble with 
ISO/OSI.  (We also had a lot of stuff at that time that used the 
overwrought IEEE 802 Ethernet frame header mess with SNAP headers and all.)

All of that said, however, the ISO/OSI stuff had a lot of really 
interesting ideas (wrapped in expensive documents filled with 
indecipherable gibberish and more unnecessary bells and whistles than a 
circus Calliope.).  I think the IETF's hostile attitude towards ISO/OSI 
created an atmosphere of auto-rejection in which those ideas were too 
often ignored.

         --karl--


On 3/18/22 10:02 AM, Bob Purvy via Internet-history wrote:
> I was around for all this, but probably not as much as some of you. So many
> memories fade...
>
> I've been reading this
> <https://courses.cs.duke.edu//common/compsci092/papers/govern/consensus.pdf>.
> This passage...
>
>
> *By August 1990, federal agencies were required to procure
> GOSIP-compliantproducts. Through this procurement requirement, the
> government intended to stimulate the market for OSI products. However, many
> network administrators resisted the GOSIP procurement policy and continued
> to operate TCP/IP networks, noting that the federal mandate, by specifying
> only procurement, did not prohibit the use of products built around the
> more familiar and more readily available TCP/IP.*
>
> ... in particular stuck out for me. Admins were required to go OSI, but
> somehow it never happened.  Does anyone have any personal stories to relate
> about this, either your own or someone else's?
>
> *Disclosure*: I'm writing historical fiction, mostly because that's what I
> want to do. So there won't be any actual names in whatever I write. I'm
> interested in the private choices people make, not the institutions,
> towering figures, and impersonal forces that most historians write about.