[ih] Early interconnection between Ethernets and Arpanet

Matthias Bärwolff mbaer at cs.tu-berlin.de
Fri Mar 12 21:46:17 PST 2010 

Thanks, this is helpful.
(ih-list cced, just for the record)

John Shoch wrote:
> --In the VERY early days PARC first built Maxc, the time-shared pseudo-PDP-10, which ran the regular Tenex operating system;  we used terminals to access the machine.  That provided initial access to the Arpanet, via the Tenex network code.  We could Telnet to other sites, exchange email, FTP, etc.
>
> --As soon as the lab started to envision hundreds of Altos, on dozens of Ethernets, it was clear that the Arpanet protocol design was not appropriate.  Thus, we designed our own internet architecture, PUP;  created at about the same time as TCP, but actually implemented and refined well in advance of TCP (and its successor, the more layered TCP/IP).
>
> --This internet architecture allowed Alto users on our multiple networks (all around the world) to access our own set of PUP file servers, print servers (first laser printers), mail servers, etc.
>
> --Our PUP-based mail system supported our internal users, and could exchange mail externally, via the mail system on Maxc, to Arpanet mail sites.
>
> --I think you could Telnet from an Alto to Maxc (via a PUP-based internet telnet protocol), and from there Telnet out to a legacy Arpanet site;  but there was little need or desire to do this.
>
> --So, all these machines had indirect access to Arpanet facilities, via Maxc.  But there were no native Arpanet/NCP implementations on the end-user Alto machines.
>
> --We also had a gateway link from the PUP architecture using the Arpanet as a transit network.  So, for example, after we gave some Alto computers to MIT, CMU, and others we could communicate with them using our internet architecture, and the Arpanet as a transit network.
>
> Hope this is helpful.
>
> John Shoch
> Alloy Ventures
>
> -----Original Message-----
> From: Matthias Bärwolff [mailto:mbaer at cs.tu-berlin.de]
> Sent: Thursday, March 11, 2010 11:03 PM
> To: John Shoch
> Subject: Re: [ih] Early interconnection between Ethernets and Arpanet
>
>
>
>
> John Shoch wrote:
>   
>> Friends,
>>  
>> As I recall, Parc's initial Arpanet connection came via an IMP provided by Arpa.  The Xerox MAXC machine (a micro-programmed pseudo-PDP10) had a Data General Nova as a front-end, and that was outfitted with an IMP interface.  Thus, we had Arpanet access for Maxc -- and email!  But that was not an internet.
>>   
>>     
>
> Did you have other computers connect to the Maxc via Ethernet back then?
> And, in which ways would the Maxc serve them as a gateway to the Arpanet
> -- not in the Internet sense, but in the pre-Internet "here I am at
> Xerox and want to access some host on the Arpanet via Telnet or FTP (or
> whichever pendants applicable inside Xerox)" one? I am asking because I
> am interested in pre-Internet network connections to the Arpanet and how
> they fared.
>
> Thanks,
> Matthias
>
>   
>>  
>> Although the work started much earlier, the PUP architecture was first publically described in a Xerox PARC blue-and-white report in 1979, available here:
>> http://www.bitsavers.org/pdf/xerox/alto/pupArch.pdf
>>  
>> The following year, the full paper was published in the IEEE Trans. on Comm.
>> D. R. Boggs, J. F. Shoch, E. A. Taft, and R. M. Metcalfe, "Pup: An Internetwork Architecture," IEEE Transactions on Communications, April 1980.
>> [I'm not sure what it takes to access an online version of this, through the IEEE.]
>> [We really do need to get a scanned, OCR'd version of this online......]
>>  
>> The system initially used a set of "gateways" (now called routers) connecting end-user machines on Ethernets, on the Data General MCA, and other systems.  The gateways were connected by leased lines, to provide our own packet-switched backbone;  we also used the ARPANET and the Packet Radio Network as transit networks, between PUP gateways.
>>  
>> Some excerpts from the paper:
>>  
>> "This work serves as the basis for a functioning internetwork system that provides service to about 1000 computers, on 25 networks of 5 different types, using 20 internetwork gateways."
>>  
>> "Long-haul communication facilities include the ARPANET, the ARPA packet radio network, and a collection of leased lines implementing an ARPANET-style store-and forward network.  These facilities have distinct native protocols and exhibit as much as three orders of magnitude difference in bandwidth."
>>  
>> "We distinguish two kids of gateways:  media translators and protocol translators. Media gateways are hosts with interfaces to two or more packet transport mechanisms among which they forward internet datagrams, using the appropriate encapsulation for each. ... Protocol gateways are hosts which speak two or more functionaly similar but incompatible higher-level protocols used to transport information within networks, mapping one higher level abstraction into the other."  [Our gateways were packet-style media translators;  I don't believe we ever built a PUP-to-TCP protocol translator.]
>>  
>> "ARPANET:  To cover longer distances, Pups can be routed throug the ARPANET;  the format for encapsulating a Pup in an ARPANET message is shown in Fig. 2."
>>  
>> "Packet Radio Network:  On an experimental basis, the ARPA packet radio network has been used to carry traffic among local networks in the SF Bay area."
>>  
>> "The Pup archiecture emerged against a background of ARPANET protocols.  Many of its important ideas -- and those of its key relative, TCP -- first appeared during the coure of a series of meeting[s] of the International Network Working Group (IFIP TC-6 WG6.1) during 1973.  Pup and TCP share a number of important principles....."
>>  
>> Hope this helps.
>>  
>> John Shoch
>> Alloy Ventures
>>  
>> PS:  Use of the Packet Radio net (based at SRI) as a transit network for Pup was driven by "the other Larry" -- Larry Stewart.  That work was also published in 1979:
>> http://portal.acm.org/citation.cfm?id=800092.802993  Thanks again, Vint, for helping to make the PRNet available for our experiment!
>>  
>> PPS:  I met Danny Cohen through the TCP working group meetings;  he and I had fun figuring out how you might encapsulate Pup packets in TCP packets, and TCP packets in Pup packets -- what we called mutual encapsulation.  The paper, with Ed Taft, was published in 1981:  Shoch, Cohen, and Taft, Mutual Encapsulation of Internetwork Protocols", Computer Networks 5, North-Holland, 1981, 287-300.
>>
>>
>>  
>>  
>>  
>> -----Original Message-----
>> From: Vint Cerf [mailto:vint at google.com]
>> Sent: Thursday, March 11, 2010 3:32 AM
>> To: Matthias Bärwolff; Bob Metcalfe; John Shoch; Yogen Dalal; Larry Masinter
>> Cc: internet history
>> Subject: Re: [ih] Early interconnection between Ethernets and Arpanet
>>
>>
>> there were gateways into the ARPANET from the Xerox PARC Ethernet. At least one of gateway was built using a packet radio as part of an early internet experiment. What I don't recall is whether this was a protocol converting gateway (from Xerox' PARC Universal Packet and associated TCP-like protocol to TELNET/NCP for example or whether there was an early TCP running on, e.g., either an ALTO work station or the MAXC server). PARC also had an ARPANET IMP (or TIP?) and connected the MAXC into the ARPANET as an NCP host. The Ethernet would have linked the ALTOs into the MAXC machine. PUP was being developed at the same time as TCP/IP so I expect there was protocol conversion going on somewhere; possibly in MAXC.  
>>
>> I am not sure whether Bob Metcalfe, Yogen Dalal and John Shoch are on the history list so I am copying them directly. Bob and John were at PARC during the TCP/IP and PUP developments; Yogen was at Stanford working on TCP/IP during the 1974-1976 period and later joined PARC to work on XNS (Xerox Network System) that grew out of the PUP experiments. If I remember correctly, Larry Masinter did the work to link the Ethernet at PARC with the Bay Area Packet Radio network with a packet radio on the premises. 
>>
>> vint
>>
>>
>> On Thu, Mar 11, 2010 at 6:00 AM, Matthias Bärwolff < mbaer at cs.tu-berlin.de> wrote:
>>
>>
>> Dear all,
>>
>> is there any documentation (or tacit knowledge on this list) on early
>> Ethernet-Arpanet gateways (obviously other than those built using IP in
>> the late 1970s)? I found some documentation on how Alohanet terminals
>> connected to the Arpanet, and how the NPL guys in the UK connected their
>> IBM machines to their TIP, plus some notes on how people struggled to
>> get the SRI port expanders working; but nothing on the much more obvious
>> candidate, Ethernet, given its early popularity.
>>
>> Sorry if I've just been lazy. Thanks for all hints.
>>
>> Matthias
>>
>> --
>> Matthias Bärwolff
>> www.bärwolff.de <http://www.xn--brwolff-5wa.de> 
>>
>>
>>
>>
>>
>>   
>>     
>
>   

-- 
Matthias Bärwolff
www.bärwolff.de

More information about the Internet-history mailing list