[ih] Internet analyses (Was Re: IPv8...)

Brian E Carpenter brian.e.carpenter at gmail.com
Thu Apr 23 14:31:56 PDT 2026


> Second, TCP/IP was free of cost.

This needs to be underlined. OSI packages were of course available but at eye-watering cost. I can't recall the exact numbers, but I remember looking into OSI/CLNP for Motorola 68000 microprocessors running the RMS68K operating system, and a site license was tens of kilodollars from some company in Santa Monica. They never had a chance. (I think that was in 1983.)

Regards/Ngā mihi
    Brian Carpenter

On 24-Apr-26 08:54, John Gilmore via Internet-history wrote:
> Jack Haverty via Internet-history <internet-history at elists.isoc.org> wrote:
>>     	     	 		  	 I recall for example seeing a
>> huge semi-truck filled with Sun workstations delivering to a Wall
>> Street investment house.  The computer industry figured out that new
>> characteristic of their marketplace, TCP became the effective
>> Standard, regardless of what the official bodies said.
> 
> I was at Sun for a number of early years.  It was a bit obvious how Sun
> standardized on TCP/IP.  First, TCP/IP was built into Berkeley UNIX,
> which was the origin of SunOS.  It was there and it worked, and so we used
> it in-house too ("eating our own dogfood" to ensure that it kept working
> well).
> 
> Second, TCP/IP was free of cost.(*) Sun customers could buy third-party
> products (some even resold by Sun) that implemented DECnet, SNA, X.25,
> OSI, etc.  But they were all extra-cost products, and they typically
> cost the customer hundreds or thousands of dollars for EVERY
> WORKSTATION.  The percentage of customers who would pay that toll was
> very small.
> 
> DARPA's contract to pay Berkeley to embed a working TCP/IP into the
> freely available(*) Berkeley UNIX 4.3BSD distribution was certainly good
> for driving TCP/IP acceptance among their academic research partners.
> It had the perhaps unanticipated follow-on effect of driving the
> acceptance of TCP/IP in the much broader commercial UNIX world that came
> a very few years later.  Indeed, that story is a classic textbook
> example of the power of freely available software to move markets in
> positive directions for end-user benefit.(**)
> 
> 	John
> 
> (*) BSD was free of per-copy cost if you already had an AT&T UNIX
> license.  Academic sites got very cheap source code licenses from AT&T.
> Commercial companies paid a "UNIX tax" per copy shipped in binary.  But
> since that tax was built into the purchase price of the workstation, and
> since in its early years Sun didn't "unbundle" its OS, there was no
> further customer cost for the Berkeley-added TCP/IP.  Much later, the
> BSD OS had all of its AT&T-licensed parts removed and replaced, becoming
> fully both free-of-cost and free-as-in-freedom.
> 
> (**) The BSD+TCP example informed my later business model when founding
> Cygnus Support.  Cygnus drove the acceptance of the GNU C and C++
> compilers in the 1990s, by providing commercial support, customer-funded
> improvements, and a train of increasingly reliable and capable
> free-as-in-freedom releases.  The result was that for a few decades they
> became and remained the default C compilers throughout the non-Microsoft
> portions of the whole computer industry.


More information about the Internet-history mailing list