Re: "the net"

[David Farber <dave () farber net>]

Begin forwarded message:

From: Steve Crocker <steve () shinkuro com>
Date: September 11, 2009 1:35:35 PM EDT
To: John Day <jeanjour () comcast net>
Cc: Steve Crocker <steve () shinkuro com>, dave () farber net
Subject: Re: [IP] "the net"


On Sep 11, 2009, at 7:41 AM, John Day wrote:

> Steve,
>
> I am really trying to understand this.  Because I think the major  
> insights of this field often go unnoticed.  But I don't quite  
> understand what the "dramatic breakthrough" was.  To me it always  
> seemed that connecting the various networks that had been built  
> together was the "logical progression."  The next obvious thing to  
> do.  The idea of using layers to move from hardware specific to  
> hardware independent was well-established in Operating Systems by  
> 1970 and its application to networks was "obvious."  ;-) (Well, we  
> all thought so) We kind of thought in layers for the ARPANet, but it  
> was really CYCLADES that settled us on physical, data link, network  
> and transport, which does make the transition from hardware  
> dependent to independent.  (Of course, that view turns out to have  
> been overly simple.  The world was more complicated than that.)

Perceptions vary.  While it might have seemed obvious to some of us  
extend the lower layers, there really were significant conceptual,  
technical and political issues involved.  And though layering is a  
straightforward idea, choosing the layers and getting the details  
right is considerably trickier than it might seem.  ARPA built the  
Arpanet and kept adding nodes, and it could have easily stopped  
there.  Had it done so, similar sized networks would have sprung up  
and we probably would have had chaos.

"Dramatic breakthrough" might be more laudatory than you'd choose.  I  
wanted to make the point that the very substantial work on protocols,  
process and community was and continues to be important, and it  
transcends the packet forwarding service provided by the lower layers,  
but I did not want to diminish the importance of the packet forwarding  
service.

> Of course saying and doing it are two entirely different things and  
> there was a lot of work to make it happen.  But not what I would  
> call a "breakthrough."  For me, a breakthrough is an unexpected  
> result that opens up new vistas.  Things like datagrams, or that  
> bounding 3 timers is both necessary and sufficient for reliable data  
> transfer, nothing else required.  Breakthroughs are rare.
>
> In one sense I think we went awry someplace.  The early Internet  
> discussions were about *internet* protocols operating over network  
> protocols. (One of Vint's early IENs talks about it.)  Hence an  
> Interent. Somehow we ended up with internet protocols mainly over  
> data link protocols.  Strictly speaking, we have more of a catenet  
> than an internet.  (The "Internet" today puts far more constraints  
> on the internal operation of a subnet provider than the ITU ever  
> did. Not some thing that makes me feel good.) I don't know where it  
> happened, but it is clear that an internet would have been more  
> advantageous than the current catenet masquerading as an internet.
>
> Also, what new user applications were developed and deployed in the  
> Internet between the demise of USING in 1974 and the advent of the  
> web?  (I consider DNS and SNMP as internal to the network, not user  
> applications.)
>
> As for the 'Net taking root and flourishing where others didn't, be  
> careful. It is far from clear that we can take credit for that.   
> When a vastly overprovisioned facility is made available essentially  
> for free, it is hard for it not to take root.  Most of the other  
> efforts were required to make a business case very early usually  
> before they were built.  Only the ARPANet and only because it was  
> such a minor part of the DoD budget was able to exist on government  
> subsidies.  I fear that when the infatuation wears off, people will  
> realize that the Internet's success has had more to do with  
> protective economic policies than good science and engineering.

There had been several smaller networking projects prior to the  
Arpanet.  I worked on one at UCLA that consumed a substantial amount  
of money and faltered because of some weaknesses in key personnel plus  
intramural politics within UCLA.

Overprovisioning is an important idea.  Underplaying a venture is  
often a path to failure.  If the Arpanet had depended on 100%  
cooperation from all of the sites and complete competency and  
execution from each of those sites, the venture would have been much  
riskier.  Also, BBN's orientation toward very high reliability, though  
deeply flawed in some respects, was also pivotal in its success.  And,  
of course, embodying the packet forwarding in the IMPs instead of  
within each host, was also pivotal.

Having it all subsidized was also important.  That gave it room the  
flower on technical merits, not short term ROI.

> But I have to admit, that I too refer to the whole thing as "the  
> 'Net" and contribute to promulgating its myths.  ;-)
>
> Take care,
> John
>
>
>
>
> > Begin forwarded message:
> >
> > From: Steve Crocker <steve () shinkuro com>
> > Date: September 10, 2009 2:16:11 PM EDT
> > To: dave () farber net
> > Cc: Steve Crocker <steve () shinkuro com>
> > Subject: Re: [IP] Internet still reshaping history
> >
> > For IP.
> >
> > On Sep 6, 2009, at 1:42 PM, David Farber wrote:
> >
> > > Why do people say the internet is 40 years old. Is the Arpanet the  
> > > Internet -- I think not. djf
> >
> > Dave,
> >
> > Both the "Arpanet" and the "Internet" are two things at the same  
> > time.  The Arpanet was a single network with uniform set of  
> > routers, a single "vendor" (BBN) and a single network operations  
> > center (at BBN).  At the same time, it was a heterogenous network  
> > that connected a diverse of host computers with an open ended and  
> > growing set of protocols.  The Internet was a dramatic breakthrough  
> > on the lower levels -- a diverse set of routers, multiple vendors,  
> > multiple operators, operating over multiple media.  The work you,  
> > Bob Kahn and Vint Cerf, and many others did in the 70s that came to  
> > fruition in the 80s created the technical and operational  
> > foundation for today's Internet.  At the upper layers, the user  
> > community, the open protocol architecture and the expansion of  
> > applications continued more or less continuously.  From this  
> > perspective, the Arpanet and Internet are part of a continuous  
> > arc.  There had been smaller networking attempts prior to the  
> > Arpanet, but none took root and flowered the way the Arpanet did.   
> > In casual conversation with non-technical people, I find myself  
> > using the term "the net" to cover the whole arc.
> >
> > Steve
> >
> >
> >
> >
> >
> > -------------------------------------------
> > Archives: https://www.listbox.com/member/archive/247/=now
> > RSS Feed: https://www.listbox.com/member/archive/rss/247/
> > Powered by Listbox: http://www.listbox.com





-------------------------------------------
Archives: https://www.listbox.com/member/archive/247/=now
RSS Feed: https://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com