[ih] Internet-history Digest, Vol 50, Issue 6

[John Day]

Packet switching had many advantages, but from the point of view of the inventors (Baran and Davies) was the improvement over message switching.

Message switching (and torn tape systems, which message switching was emulating) were analogous to the FIFO Batch systems in computers. Short messages could get stuck behind long ones. Packet switching was analogous to multiprogramming. Short messages were still delayed some, but their 'completion time' was much shorter. (Just like in processor scheduling).  This is what attracted Davies to the idea and must have been noticed by Baran, although I haven’t found him mention it explicitly. His focus was mostly on survivability and redundancy. But it seems implicit in much of what he wrote.

That makes Virtual Circuit, multiprogramming with contiguous (static) allocation; and datagrams multiprogramming with dynamic allocation.

And a couple of years later (1968), Denning showed that dynamic was orders of magnitude less likely to run out of buffers than static allocation.

Take care,
John

> On Jan 10, 2024, at 19:38, Vint Cerf via Internet-history <internet-history at elists.isoc.org> wrote:
> 
> thanks John -
> the last point about speed adaptation is a key value in packet switching as
> long as there is also adequate flow control.
> The ability to link at different speeds and do store-and-forward is one of
> the major attractions of packet switching.
> v
> 
> 
> On Wed, Jan 10, 2024 at 3:26 PM John Shoch via Internet-history <
> internet-history at elists.isoc.org> wrote:
> 
>> I've gotten a little behind on the Internet History list.....let me try to
>> catch up.
>> 
>> John Day:  Thank you for providing more background on Cyclades, and its
>> underlying transport system, Cigale.
>> Louis Pouzin's group at IRIA was an early advocate of datagram-based
>> networking, and had the vision to see the importance of connecting diverse
>> networks into a "catanet" -- although they were not able to implement that
>> in France.
>> Vint has noted that Gerard Le Lann was a visitor at Stanford (and is
>> acknowledged on the Stanford Internet Plaque that Vint organized):
>>  http://ana-3.lcs.mit.edu/~jnc/history/Internet_plaque.jpg
>> The Cerf/Dalal/Sunshine early draft of a TCP spec., Dec. 1974, acknowledges
>> input from IRIA:
>> "In the early phases of the design work, R. Metcalfe, A. McKenzie, H.
>> Zimmerman, G. LeLann, and M. Elie were most helpful in explicating the
>> various issues to be resolved."
>>  https://datatracker.ietf.org/doc/html/rfc675
>> 
>> Dave Crocker:  You noted that, "... I guess PARC was doing
>> gatewaying/translation between Arpanet/XNS peers...."
>> Perhaps I can elaborate a little bit:
>> --PUP was the first generation of internetwork protocols, developed at
>> Xerox PARC;  XNS was the second generation, primarily developed at Xerox
>> SDD (building upon what was learned in Pup -- adding large unique
>> addresses, etc.).
>> --When the full PUP architecture was first implemented within Xerox, one
>> could use PUP FTP to move a file from an Alto to an account on PARC's
>> MAXC/Tenex machine, and one could use Arpa FTP to move that to another
>> Arpanet host.  I do not think there was a higher-level protocol translation
>> to automatically link PUP FTP with Arpa FTP.
>> --The Arpanet was added as a network within the PUP architecture, but
>> primarily as a transit network (as was the Packet Radio Network).  The
>> encapsulation of PUP internet packets within Arpanet messages is described
>> and diagrammed in the PUP paper by Boggs, et al.:
>> 
>> http://bitsavers.trailing-edge.com/pdf/xerox/parc/techReports/CSL-79-10_Pup_An_Internetwork_Architecture_Jul79.pdf
>> https://ieeexplore.ieee.org/document/1094684
>> --When Xerox gave Alto/Ethernet/Dover/PUP systems to a number of
>> universities (Stanford, MIT, CMU, CalTech, Rochester), I think it included
>> Nova-based PUP "gateways" (which later led to the Stanford/Cisco
>> multi-protocol routers).  Thus, a user on an Alto at Stanford could use PUP
>> FTP to retrieve a file from a PUP file server at CMU -- and the PUP
>> internet packets would transit through the Arpanet.  [Xerox, Arpa, and the
>> universities all worked together to enable this.]
>> --The universities had Xerox PUP file servers;  I don't know if any of them
>> ran the PUP code on their PDP-10s...maybe someone was there?
>> --XNS came later....
>> 
>> John Levine made a good observation:  "It occurs to me that it might not
>> have been obvious that you could run the same network protocol over a 56K
>> DDS line and a 3Mb Ethernet, glue the two together using simple minded
>> gateways, and it'd work."
>> --This is, of course, one of the things that motivated the separate
>> development of Xerox PUP -- we knew that we had to support 3Mbps local
>> Ethernet connections to file servers and print servers, while also
>> supporting 9.6 Kbps inter-site leased phone lines.  It was a design
>> objective, and it had to be made to work!
>> --Again, from the Boggs paper:
>> "The communications environment includes several different individual
>> network designs. The
>> dominant one is the "Ethernet" communications network, a local-area
>> broadcast channel with a
>> bandwidth of 3 megabits per second [Metcalfe & Boggs, 1976]. 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 performance."
>> 
>> Cheers,
>> 
>> John Shoch
>> --
>> Internet-history mailing list
>> Internet-history at elists.isoc.org
>> https://elists.isoc.org/mailman/listinfo/internet-history
>> 
> 
> 
> -- 
> Please send any postal/overnight deliveries to:
> Vint Cerf
> Google, LLC
> 1900 Reston Metro Plaza, 16th Floor
> Reston, VA 20190
> +1 (571) 213 1346
> 
> 
> until further notice
> -- 
> Internet-history mailing list
> Internet-history at elists.isoc.org
> https://elists.isoc.org/mailman/listinfo/internet-history