[ih] Hourglass model question

John Day jeanjour at comcast.net
Fri Jul 5 17:37:17 PDT 2019


Thanks Steve for that. Just to add.

Keep in mind computing was still very small (there was only one or two computer conferences a year, when did NCC split into Fall and Spring Joint?) The networking field was even smaller. Publishing a paper was considerably more work and the criteria considerably higher than they are now. A lot of work and a lot of discussing went on that never appeared in publications or even in RFCs or other samizdat circulations. (I have all sorts of papers from this period that were not part of any even informal publication series.

In 1968, Dykstra published his paper on THE and layered OSs. And it was all the buzz.  Most, if not all, of the NWG were OS guys. You needed OS guys to figure out how to introduce the IMP-Host protocol and then the Host-Host on top of that in the OS. By 1970, layer diagrams of IMP-Host, Host-Host(NCP), (Telnet, DTP), FTP, RJE were common. (DTP was Data Transfer Protocol, the part of FTP that did the actual transfer.)

By 72/3, the layers of Physical, Data Link, Network, Transport from CYCLADES were pretty common as well as a general characterization that wasn’t specific to a given network. INWG began in 72 after ICCC ’72 and these layers were common by then. There is also strong evidence that because CYCLADES was building a network to do research on networks (very different from what the ARPNET was)*, they had figured out a lot more about layers than most of us knew at the time.

John

*Remember the ARPANET was built to lower the cost of research but not really to do research on networks. That could be a side benefit and a lot of us thought there was a lot to do, but it wasn’t ARPAs main focus for the ARPANET. Once it was built, ARPA considered the network part done! (At least for awhile they did.) BBN couldn’t take the net whenever they wanted to do some experiment. The ARPANET was in a fairly real sense, a production network to support ARPA research.

> On Jul 5, 2019, at 17:35, Steve Crocker <steve at shinkuro.com> wrote:
> 
> Layering was part of the earliest discussions we had in 1968-69.
> 
> On Fri, Jul 5, 2019 at 5:34 PM Craig Partridge <craig at tereschau.net <mailto:craig at tereschau.net>> wrote:
> Related but not quite on target.
> 
> The hourglass/margarita glass is a representation of layering.  And back in 1988 I tried to figure out the origins of the layered model for a collection of networking papers I edited.  At the time, the best answer I found was that layering, from a networking perspective, originated with a paper by Davidson et al. on the ARPANET TELNET protocol from the DATACOM conference in 1977.  It portrays layering as a fan, in which different protocols layer on each other as needed.  But it clearly articulates the notion of layering and how layers interact.  (And there's a narrow window between the 1977 paper and the Cerf/Kahn 1974 paper on TCP/IP, which presumably would have mentioned layering if the concept was in wide use).
> 
> Craig
> 
> On Wed, Jul 3, 2019 at 8:10 AM Andrew Russell <arussell at arussell.org <mailto:arussell at arussell.org>> wrote:
> Hi everyone - 
> 
> You might have seen the CACM featured an article in the most recent issue “On the Hourglass Model” - https://cacm.acm.org/magazines/2019/7/237714-on-the-hourglass-model/fulltext <https://cacm.acm.org/magazines/2019/7/237714-on-the-hourglass-model/fulltext>. 
> 
> It’s not a history paper, but it raised a history-related question for me.  As far as I know the visual representation in question started with a drawing of a margarita glass in 1979, in the context of an OSI committee meeting and the 7-layer model. I reproduced the image on page 214 of my book “Open Standards and the Digital Age” - it’s visible to me here: 
> https://books.google.com/books?id=jqroAgAAQBAJ&pg=PA214&lpg=PA214 <https://books.google.com/books?id=jqroAgAAQBAJ&pg=PA214&lpg=PA214>.
> 
> My question for the list has 2 parts:
> 1) when/where did the margarita glass turn into an hourglass?
> 2) when/where did the TCP/IP community borrow it from the OSI community?  (I’m assuming this is how it happened, would be very interested in evidence or recollections to the contrary)
> 
> My hunch, without doing a fresh round of research, is that I should look first to papers by David Clark and co-authors in the 1980s to answer a third question, which is how this illustrated concept morphed into a “Theorem” (as the CACM essay puts it).  But that’s just a hunch, and I’d really appreciate pointers or recollections.
> 
> Thank you!
> 
> Andy
> _______
> internet-history mailing list
> internet-history at postel.org <mailto:internet-history at postel.org>
> http://mailman.postel.org/mailman/listinfo/internet-history <http://mailman.postel.org/mailman/listinfo/internet-history>
> Contact list-owner at postel.org <mailto:list-owner at postel.org> for assistance.
> 
> 
> -- 
> *****
> Craig Partridge's email account for professional society activities and mailing lists.
> _______
> internet-history mailing list
> internet-history at postel.org <mailto:internet-history at postel.org>
> http://mailman.postel.org/mailman/listinfo/internet-history <http://mailman.postel.org/mailman/listinfo/internet-history>
> Contact list-owner at postel.org <mailto:list-owner at postel.org> for assistance.
> _______
> internet-history mailing list
> internet-history at postel.org
> http://mailman.postel.org/mailman/listinfo/internet-history
> Contact list-owner at postel.org for assistance.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://elists.isoc.org/pipermail/internet-history/attachments/20190705/4929a3d8/attachment.htm>


More information about the Internet-history mailing list