[ih] when did APRANET -TIPs become known as -TACs

Barbara Denny b_a_denny at yahoo.com
Sun Sep 28 23:10:03 PDT 2025


  I am not sure in Germany you had an option.  I think you had to use x.25 to get the physical connection between sites. I could be very wrong about this though. (The US army may have also already selected to use x.25 to connect sites.) . As a reminder I was helping Cisco debug their AGS router with x.25  when I was in Germany. BTW I am pretty sure Cisco's first product was not a router. I have seen websites only talk about routers in the history of Cisco.
In my memory I only associate PSN as an IMP.,  sorta like a router was originally called a gateway. 
Ppbarbara

    On Sunday, September 28, 2025 at 03:14:41 PM PDT, Jack Haverty via Internet-history <internet-history at elists.isoc.org> wrote:  
 
 Hi Barbara,

X.25 made it easier to connect hosts to IMPs.   ARPA had funded lots of 
computers used in the research community to have someone build 1822 
interfaces.   But there were lots of other computers used outside the 
research environment.  X.25 was a key part of the OSI vision, and 
computer manufacturers were much more likely to create an X.25 interface 
for their products than an 1822 interface.

I don't think this really affected the choice of IMPs versus directly 
connecting routers with wires.  Part of an X.25 interface was the basic 
physical connection for a wire, and it was straightforward to just 
interconnect routers with a wire by using that same physical interface 
with no X.25 connection management sofware needed.

At one point at BBN we noticed that a wire was really just a very basic 
"network" that could be used to interconnect gateways.  A wire was 
effectively a "class F" network with just 2 possible attached computers 
- "this end" and "the other end".  We actually tried connecting two 
gateways together with a wire instead of using the ARPANET and it worked 
fine.

At one point (can't remember exactly when) I was responsible for "DDN 
System Engineering", and frequently got called to Washington for various 
meetings.   One of them was to hear some startup's pitch for how DDN 
could use their products.  After the pitch, everyone turned to me and 
the guy in charge asked "Will this work?".   I think they expected me to 
say it was a silly idea and they really needed to use BBN's solutions.  
But, as a DDN consultant, I said "Yes, it should."   They got a testbed 
running, and the startup no doubt realized the same thing that we (and 
SRI later) did - you didn't really need the IMP in the picture.  BTW, 
that startup was Cisco Systems.

I'm not sure when I first heard PSN as the acronym for Packet Switched 
Node.  My recollection is that the terminology came out of the OSI 
vision which had become very popular.  IMP stood for Interface Message 
Processor but that name was always confusing.  I remember there was an 
IMP somewhere with a clipping pasted on its front panel -- a message had 
come in from a US government Senator, congratulating someone (ARPA?  
BBN?) on their successful creation of the "Interfaith Message 
Processor"  (read the last three words carefully).   There's a writeup 
at 
https://foxmancommunications.com/the-interfaith-message-processor-and-the-tower-of-babel/

It was an interesting time.

Jack

On 9/28/25 12:19, Barbara Denny via Internet-history wrote:
>  Do you think the obsolescence of 1822 in favor of X.25 just made it easier to get rid of IMPs in the networks?
> I am thinking about the military testbed for USAREUR where I think the original deployment had CXXs (don't remember if it had C30s or C70s but leaning towards C30s).  It was deployed in that timeframe. It originally had IMPs and Cisco AGS routers  but I think SRI pulled the IMPs and just used the routers very soon after it was originally installed.
> BTW, what is the story for replacing the IMP term for PSN (Packet Switching Node) and when was this done?
> barbara
>      On Sunday, September 28, 2025 at 10:08:29 AM PDT, Jack Haverty via Internet-history<internet-history at elists.isoc.org> wrote:
>  
>  On 9/28/25 06:31, Noel Chiappa via Internet-history wrote:
>>        > From: Jack Haverty
>>
>>        > Much of this history was probably well-documented in the reports
>>        > submitted by BBN ... It may be available on discover.dtic.mil
>>
>> I took your suggeation, and turned up an answer to one question: there
>> were C/30-based TACs, as well as one-time-TIP-based TACs.
>>
>>      Combined Quarterly Technical Report No. 22
>>      https://apps.dtic.mil/sti/pdfs/ADA104931.pdf
>>      Combined Quarterly Technical Report No. 23
>>      https://apps.dtic.mil/sti/pdfs/ADA108783.pdf
>>      The DDN (Defense Data Network) Course
>>      https://apps.dtic.mil/sti/tr/pdf/ADA173472.pdf
>>
>> I found nothing about what physical interface any of them used, though,
>> but I'll bet it was an 1822.
>>
>>      Noel
> Hi Noel,
>
> Yes, good detective work.  There's lots of information in contractor
> reports.  At every Internet Meeting there were a lot of contractors, so
> the reports from each were limited to 15 minutes or so.   Much of the
> detail wasn't even presented in the meetings, and of course not captured
> in Jon's minutes.
>
> There's lots of technical detail in those old reports that probably
> should have been issued also as RFCs or IENs.  The reports went to
> various parts of the government, and to the people inside BBN who had
> worked on the projects, but probably not much beyond those groups.  For
> example, I assume all the other ARPA contractors had to submit similar
> reports.   But I don't recall ever seeing a report from SRI, MIT, UCLA,
> Linkabit, Collins, or any of the other contractors who attended the
> various Internet meetings.   I still haven't seen more than a handful of
> non-BBN reports, but I suspect some might be in DTIC.
>
> At some point I was given responsibility for all of the ARPA and related
> contracts in our part of BBN.   That meant I became the "author" of the
> BBN reports.  Pragmatically what it meant was that I had to badger all
> of the project leaders to write down what their teams did during the
> quarter.   Getting blood out of a stone would have been easier than
> getting documentation out of an engineer.  For many of our contracts,
> the only required deliverables were the Quarterly Reports.   Until the
> Report was submitted, the government wouldn't pay the bill.
>
> BTW, re C/30 et al.  Internet History has probably never been told about
> that part of the history:
>
> The C/30 hardware was based on a BBN project called the MBB -
> Microprogrammable Building Block.  As the name implies, the hardware was
> microprogrammable.  The C/30 microcode was designed to make an MBB look
> exactly like a Honeywell 316.  So the same code that had been developed
> for the 316-based IMPs (or TIPs) would also run on a C/30.
> Effectively, a C/30 looked exactly like a Honeywell 316 to the software
> that ran on it.
>
> Similarly, a C/70 was a Unix minicomputer also built on an MBB, but with
> an interface to disk storage and probably more RAM.  The MBB microcode
> used for a C/70 was optimized for code written in the C language, which
> was the language used by the Unix OS.
>
> BBNCC started life as BBN Computer Corporation, with a plan to sell Unix
> boxes to the world.   Competing with DEC was probably always a bad idea,
> so later BBNCC became BBN Communications Corporation, selling IMPs to
> the marketplace as ARPANET clones, and a few C/70s operating as NOCs.
> Didn't even have to change the logo.
>
> There was also a C/60, but I can't remember what it did.....
>
> There's probably lots of detail in other old BBN reports, as well as
> reports from others.  For example, I just searched in DTIC for "BBN MBB"
> and found this discussion about formal verification of the C/30
> microcode:https://apps.dtic.mil/sti/tr/pdf/ADA510573.pdf   -- which
> even found bugs in the microcode.  Another timeline bit in that report -
> the 1822 interface was obsolete on DDN by 1986, in favor of X.25 for the
> Host/IMP interface.

  


More information about the Internet-history mailing list