[ih] Historical fiction

[Louis Mamakos]

On May 15, 2012, at 6:26 PM, Noel Chiappa wrote:

>> From: Alex McKenzie <amckenzie3 at yahoo.com>
> 
>> Sytel asked what an ARPAnet message looked like.
> 
> I'm working on a page about them - it's partway done (have all the images,
> both Host-IMP and IMP-IMP - only two of the latter, I imagine there must have
> been more, but that's what I could find), but I got distracted to work on an
> ARPANET maps page (almost done, hope to have it out soon), and then the
> archives stuff! Popping my way back to it, though... :-)
> 
> 	Noel

In some respects, the syntax of an 1822 frame on the host interface is less
interesting than the service model that the IMP/PSN subnet exposed.  I think
it's interesting to see what the capabilities of that L2 subnetwork were as
compared to what followed, later with the Ethernet and other multipoint
network technologies like FDDI, Frame Relay and ATM networks.  Sure, X.25
networks had some of these capabilities, and perhaps it was in that context
and lack of the layered architecture we take for granted today that naturally
led to a "thicker", more feature-rich IMP subnetwork.

What other subnet layers deployed could transport arbitrary bit-length messages?
That had end-to-end flow control (e.g, RFNM messages)?  

Of course, I still have nightmares of X.25 host/PSN connections.  Not a good fit
for a "host" that was actually a "gateway" (router) between the ARPANET and
the NSFNET, with more flows to ARPANET hosts than available X.25 VCs.  
Thrash, thrash, lose, lose.

Louis Mamakos