[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ih] inter-network communication history

On 11/6/2019 12:38 PM, John Gilmore wrote:
> Telenet began service in August 1975.  This was a public
> packet-switching network built by BBN.

Although Telenet used packet switching, its business model was for 
dial-up service, like the earlier Tymnet.  It did X.29 over X.25 to make 
remote dial-up access (a lot) cheaper.  (Perhaps oddly this constraint 
matched the business approach that my first Silicon Valley employer, 
Ungermann-Bass used, mostly at the LAN level, for replacing lots of 
terminal wires with many fewer 'LAN' wires.)

Among the examples of Telenet's use, another one that is relevant to 
Internet history is with CSNet, which can be classed as a pre-cursor to 
NSFNet -- I usually refer to it as market research for the later effort.(*)


While the project had lofty goals, its original service was more 
mundane, simply providing email relaying between sites not (otherwise) 
on the Arpanet (and then Internet) and those that were on the net.

It did this with email relaying over dial-up lines and sometimes that 
was made cheaper by using Telenet.

This was where I had my first direct experience with mismatching packet 
sizes at different layers.  It's a classic CS training point, about 
buffer sizes, and the like.

The email transfer mechanism (MMDF) had its own reliable packet layer, 
on top of the "phone" connection and below its email transfer mechanism.

When I initially set such connections up, I discovered they were 
massively inefficient and massively slow.  And massively expensive.  All 
because the MMDF packet size was about 120% of a Telenet packet size. 
As soon as aligned mine with theirs, things went swimmingly. And a lot 


(*) At the NSFNet meeting that created the CSNet effort, the start of 
discussion was among people who thought that Telenet really was a 
replication of Arpanet functionality, with full and easy host-to-host 
services.  They were quite dismayed to discover that its service model 
was far more limited...

    CSNet established a funding model that was repeated for NSFnet, 
starting with NSF money but requiring self-sufficiency with some years.

Dave Crocker
Brandenburg InternetWorking
Internet-history mailing list
Internet-history at elists.isoc.org