[ih] Installed base momentum (was Re: Design choices in SMTP)

[Barbara Denny]

 Let's not forget about ATM. I  think ATM was also a big area of focus for many people in this time frame.  
barbara

    On Friday, February 10, 2023 at 06:48:47 AM PST, Craig Partridge via Internet-history <internet-history at elists.isoc.org> wrote:  
 
 On Thu, Feb 9, 2023 at 7:16 PM Jack Haverty via Internet-history <
internet-history at elists.isoc.org> wrote:

>
>
> At the time, in the 1990ish timeframe, there was a huge installed base
> of network technology.  Hundreds of thousands of computers utilizing
> networks based on SNA, SPX, XNS, Decnet, etc. etc.  TCP existed, but
> was a small player, confined largely to the academic and research
> communities.
>
> ...
>
> So how did TCP manage to blast through that momentum of the installed
> base, creating such a chaos in the collision?  And how did it do it so
> rapidly?
>
>
Hi Jack:

I'll start with a shout out to Brian's point that the transition was
already well underway by 1990.  Absolutely
fits my experience.

 I would argue that a critical issue was communicating outside one's
organization and/or over long distance.  The various technologies you list,
except for DECNET, did not focus on solving problems across organizational
boundaries.  Recall Netware was the
biggest networking technology of the time and, while it adapted somewhat,
was designed to connect an office or suite
of offices.

Meanwhile, by 1987, we'd built a relatively homogeneous email environment
across the Internet, USENET, CSNET, and
(thanks to BITNET and EARN) the academic SNA networks.  I remember at a DC
Interop c. 1990, someone observing
that they had discovered couldn't hire new computing graduates if they
weren't connected to the RFC 822/domain name email
network.  So the tech mindset, among the younger generation, was that they
should be able to communicate with anyone via
email.  This pushed folks towards TCP/IP -- or, at least, email
compatibility with the Internet.

At a bits-and-bytes level, long-distance reliable communications networks
are hard to do.  I remember Dave Clark talking about
this around 1985 and discussing how protocol suites designed around the
local network didn't scale.  He used the struggles by
the LOCUS distributed file system (which worked great on a LAN) to work
over the ARPANET as an example.  In the late 1980s,
only two networking architectures had engaged with and worked through those
issues: TCP/IP and DECNET.  Nicely, the most prominent and
complementary papers on congestion issues, one by Van Jacobson (TCP/IP) and
one by Raj Jain and KK Ramakrishnan (DECNET),
were presented back-to-back at the ACM SIGCOMM conference in 1988.  So if
you were looking to build (or soon after via NSFNET, connect
to) a sturdy wide-area network, unless you were a DEC VMS organization,
your best choice was TCP/IP.

I'll note it was, in my view, a near thing sometimes.  NSFNET was a
tremendous gamble and for parts of 1987 and 1988 was not
a very good service (I'm told a scientist complained loudly at the National
Academy about this non-functional network they were
trying to use for important science).  We figured out congestion collapse
well enough for the time (pace buffer-bloat folks) just as
it was threatening to make the Internet unusable.  But I distinctly
remember that roughly around the end of 1988 or beginning of 1989,
Internet folks began to realize that when they were talking with engineers
building other networking technologies there was a whole
suite of community knowledge that the Internet folks had and nobody else
(except the wonderful DEC networking team) did.

Craig


-- 
*****
Craig Partridge's email account for professional society activities and
mailing lists.
-- 
Internet-history mailing list
Internet-history at elists.isoc.org
https://elists.isoc.org/mailman/listinfo/internet-history