[ih] Correct name for early TCP/IP working group?

[John Day]

That would be my guess, that there really was no change.

Prior to sending out the ARPANET RFQ, Roberts had a meeting of the potential sites at Ann Arbor. There he got a lot of pushback that the sites' systems were already so overloaded that they couldn’t possibly support a network. Roberts left the meeting very unhappy. ;-) On the way to the airport, Wes Clark suggested putting a minicomputer in front of each host as a front-end, which is what Roberts wrote into the RFQ. Hence, IMPs.

The initial distance restrictions of 1822 and the IMP-Host protocol reflect that front-end functionality as well as that IMPs delivered whole messages to the Hosts, reassembling packets into messages. (To satisfy the objections from the Ann Arbor meeting.)  The host requested that its IMP create a ‘connection’ to the destination IMP and host. Then NCP operated process-to-process over that. NCP carried socket ids, not host ids.  

(There is a good chance the terminology then is the same as the terminology now.)  ;-)

Take care,
John

> On Jan 25, 2025, at 09:58, vinton cerf <vgcerf at gmail.com> wrote:
> 
> John, that might depend on what you mean by "front-end" - 
> Maybe you are distinguishing between the host/imp interface which accepted "messages" and turned them into packets and the part that just routed packets. Similarly on receipt, reassembling packets into messages before delivering them to hosts. 
> 
> If that is the correct parse, I am not sure there was ever a change. Host/IMP and IMP/IMP interactions were always pretty distinguishable, I think. Once we put in "real" gateway between Arpanet and Milnet, the Host/IMP interfaces were used to serve the gateway hosts, presumably.
> 
> v
> 
> 
> v
> 
> 
> On Sat, Jan 25, 2025 at 9:48 AM John Day <jeanjour at comcast.net <mailto:jeanjour at comcast.net>> wrote:
>> This brings up the question, when did IMPs cease to be front-end+router and at least some of them become just routers?
>> 
>> I know that we started to think of them that way long before they actually were that way, if they were.
>> 
>> Take care,
>> John
>> 
>> > On Jan 25, 2025, at 09:43, vinton cerf via Internet-history <internet-history at elists.isoc.org <mailto:internet-history at elists.isoc.org>> wrote:
>> > 
>> > Thanks for that, Andy - I had not realized that this was used as a
>> > scaffolding - must have affected routing in some way and created a kind of
>> > gateway IMP notion for IMPs lying along a border between two connected
>> > subsets?
>> > 
>> > v
>> > 
>> > 
>> > On Sat, Jan 25, 2025 at 9:28 AM Andrew G. Malis via Internet-history <
>> > internet-history at elists.isoc.org <mailto:internet-history at elists.isoc.org>> wrote:
>> > 
>> >> Jack,
>> >> 
>> >> - There was some work however within the ARPANET IMP software to
>> >>> acknowledge the need for multiple networks.   For example, some of the
>> >>> formats of data as it passed through the ARPANET included fields
>> >>> labelled "Network Number".  AFAIK this was never actually fully
>> >>> implemented so the ARPANET itself never achieved connectivity between
>> >>> multiple networks until TCP was deployed.
>> >> 
>> >> 
>> >> As I recall, the ARPANET's "network number" field was used during the
>> >> ARPANET/MILNET split to allow logical separation of the IMPs and hosts
>> >> sharing the same backbone infrastructure until the physical split could be
>> >> completed.
>> >> 
>> >> Cheers,
>> >> Andy
>> >> --
>> >> 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
>> >> 
>> > -- 
>> > 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
>>