[ih] Protocol numbers (was IP version 7)

[Louis Mamakos]

There was also an X.25 synchronous serial interface that appeared later. 
  When we got ARPANET IMP 17 at the University of Maryland as part of 
the NSFNET Interconnections project, our host interface was X.25.  This 
was matched to an X.25 Q-bus interface in a MicroVAX-II with the most 
complex device driver I ever had seen to date.

We had a synchronous null modem between the IMP (referred to as a PSN at 
the time, I suppose) and our host.  It had some thumbwheel switches to 
select the clock speed; 56K bps worked, sadly, selecting 78K bps did 
not.

As it turned out, I think the X.25 interface ended up being a bad fit 
for this particular application (router between two networks).  We also 
had one of the root DNS servers running at our site, so lots to sites 
wanted to exchange traffic with stuff behind our IMP port.

There was essentially an X.25 VC that would be signaled to send traffic 
to a particular remote IMP/port combination.  And there often were more 
active endpoints than number of active X.25 VCs that were supported, so 
there was some amount of thrashing that would happen.  I don't recall if 
this was a limitation of the IMP, the X.25 board in the MicroVAX or the 
driver.

louie


On 25 Dec 2020, at 9:58, Alex McKenzie via Internet-history wrote:

>  Nope.  NOT 100kbs.  The speed of the data transfer was governed by 
> the SIGNALS "ready for next bit" and "here's your bit".  The IMP made 
> NO assumptions about how fast the Host could send or receive, and no 
> assumptions about whether the speed would be constant or sporadic (for 
> example, there might be a longer delay when accessing a new Host 
> memory word than the delay between bits within a word).  Eventually 
> there was a "Very Distant Host" interface designed to run over a 
> clocked modem, but I don't think any modem with a speed higher than 
> 50kbs was ever used.
> Alex
>
>     On Friday, December 25, 2020, 7:13:41 AM EST, Steve Crocker via 
> Internet-history <internet-history at elists.isoc.org> wrote:
>
>  Nope.  Bit serial 100 kbs.  Specifically designed to be as simple 
> as possible because each site had to design its own half.  100 kbs 
> was easy to implement and fast enough to keep up with the 50 kbs 
> IMP-IMP circuits.
>
> I assume the description you’ve copied was just an unconscious 
> writing error.
>
> Steve
>
> Sent from my iPhone
>
>> On Dec 25, 2020, at 6:57 AM, Noel Chiappa via Internet-history 
>> <internet-history at elists.isoc.org> wrote:
>>
>> 
>>>
>>> From: Brian E Carpenter
>>
>>> ** "Early Experiences with the ARPANET and INTERNET in the UK", 1998
>>
>> Very interesting, thanks for the pointer.
>>
>> It contains one thing that I'm curious about: "the Interface Message 
>> Processor
>> (IMP). This were initially attached locally to Host computers by a 
>> parallel
>> interface." (bottom of pg. 2) I'm unaware of this stage; I thought 
>> they always
>> used the 1822 bit-serial interface? I'd look at my copy of the BBN 
>> proposal
>> (in response to the DARPA RFP), which probably talks about whether 
>> the plans
>> were always for that interface, but alas my son and his wife are 
>> asleep in the
>> room it's stored in, so I can't. I suppose it's possible a first IMP 
>> or so
>> were prototypes, and initially used a parallel interface, later 
>> replaced by
>> the 1822 interface, but I don't recall hearing about that. Can anyone 
>> expand?
>>
>>     Noel
>>
>> PS: That site, ban.ai/multics/, is completely mind-blowing. Check it 
>> out.
>>
>> -- 
>> Internet-history mailing list
>> Internet-history at elists.isoc.org
>> https://elists.isoc.org/mailman/listinfo/internet-history
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history
>
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history