How the Internet transformed from concept to colossus, from theory to ubiquity.
By Jim Duffy
Network World, 03/26/01
that the musings of university researchers 40 years ago would bear
the Internet that enterprise networks depend on today for business
communication and commerce is almost unfathomable. Even though most
inventions begin this way - the light bulb, the automobile, the telephone
- we lose the perspective of awe for these advances because they've
been around so long, we often take them for granted.
Enterprise networks have yet to take the Internet for granted. Users
have yet to experience all that the Internet is capable of.
Internet's origins date back to 1961, when Leonard Kleinrock, a Massachusetts
Institute of Technology student, published the first paper on packet
switching theory. He proposed using packets rather than circuits for
communications, believing that packets allowed for better sharing
"When you try to send data over a network, most of the time you're
sending nothing," Kleinrock says. "In circuit switching you dedicate
the entire communications link, or sequence of links, to your conversation.
The key idea that had to be introduced was the notion of resource
sharing, where you only got to use the communications facility when
you had something to send."
In 1962, J.C.R. Licklider of MIT wrote a series of memos on the concept
of a "Galactic Network" - a globally interconnected set of computers
through which people could access data and applications from anywhere.
Later that year, Licklider headed the computer research program at
Defense Advanced Research Projects Agency (DARPA), the organization
that would develop the computer network that some say gave birth to
The Licklider and Kleinrock theories were put to the test in 1965
when Lawrence Roberts and Thomas Merrill connected a computer in Massachusetts
to one in California over a 2.4K bit/sec telephone line. The researchers
found that time-shared computers could work together to run programs
and share data.
The computer network was born.
"My goal was to link all of the computers so that we'd all be able
to get at anything without having to have compatibility with the other
computer and be able to get all the knowledge instantly across the
world," Roberts says.
In late 1966, Roberts went to DARPA to develop the computer network
concept. He conceived his ARPANET plan and published it in 1967.
Roberts initially resisted going to DARPA. Had he not gone, or joined
later than he did, the Internet would be much different than it is
today because three separate investigations into packet switched communications
- one at research institute RAND, another in the United Kingdom -
were going on at the same time, unbeknown to other researchers.
"Donald Davies in the U.K. would have done it first," Roberts says.
"England would have had the first Internet and all of the economic
impact of the Internet, instead of the U.S."
In 1968, DARPA awarded BBN the contract to develop packet switches
for ARPANET, which were called Interface Message Processors (IMP).
Roberts worked on the ARPANET topology, and Dr. Robert Kahn of BBN
worked on the architectural design of the network.
Kleinrock, meanwhile, headed a team at the University of California-Los
Angeles to develop an ARPANET measurement system.
In September 1969, BBN installed the first IMP at UCLA, connecting
the first host computer to what would become ARPANET. A second node
was added at the Stanford Research Institute (SRI).
The first host-to-host message exchange, between UCLA and SRI, occurred
in October 1969.
"We had no idea this was going to be a monumental event," Kleinrock
says. "We just didn't make much of it. What I never imagined was that
my 92-year-old mother would be on the Internet today."
UCLA issued a press release in 1969, however, quoting Kleinrock as
saying that once this nationwide network is operational, it will be
possible to gain access to a "computer utility" as simply as accessing
electricity or telephone service.
Two more nodes were added, at the University of California-Santa Barbara
and the University of Utah. By the end of 1969, ARPANET consisted
of four interconnected host computers.
In December 1970, Steve Crocker developed the Network Control Protocol
(NCP), the initial host-to-host protocol for ARPANET. NCP enabled
three disparate networks - the time-shared computer network, a packet
satellite network and a packet radio network - to interoperate. In
1972, ARPANET e-mail was developed by BBN and expanded by Roberts.
From there, ARPANET either grew into the Internet or was merely a
predecessor that contributed to the development of the 'Net. There
is a debate on these points among four of the founding contributors
to the Internet.
"The ARPANET was a single network run by one party that could allow
just a single machine or two to connect at every node," Kahn says.
"If the ARPANET were viewed as the Internet, nobody else could ever
connect to it as a network provider. The only thing you could connect
was an end-user computer. BBN would be running the world today."
"To say that the Internet is different from the ARPANET is to say
that the invention of the automatic transmission changed what an automobile
was," Kleinrock says. "It enhanced it, it made it easier to use, the
interface was better, it added capability… but it didn't change the
essential idea that you had a vehicle that was moving and carrying
people across the country."
The Internet is based on an open architecture concept - introduced
in 1972 by Kahn as "Internetting" - in which several disparate networks
can be connected to share data and applications. The open architecture
concept was based on Kahn's work with packet radio.
To implement Internetting, Kahn developed a new version of NCP specifically
for the open architecture network. This protocol ultimately became
TCP/IP, and this, according to Kahn and TCP/IP co-author Vinton Cerf,
gave birth to the Internet. "Most people basically view the creation
of TCP/IP as the thing that was the genesis of the Internet," Kahn
says. "I think [the Internet] was my idea, not to say that all the
other parties weren't critical."
"The underlying networks [such as ARPANET] didn't know anything about
Internet packets," Cerf says. "They didn't know anything about TCP."
The spread of computer networking and the Internet took off in 1973,
when Bob Metcalfe developed Ethernet LAN technology at Xerox PARC.
From there, the industry saw widespread development of LANs, PCs and
workstations in the 1980s, and this helped the Internet flourish and
"Until the personal computer and the workstation were available, there
was no way to envision the world that you have today," Kahn says.
Momentum for the Internet continued to build in 1980, when TCP/IP
was adopted as a defense standard. ARPANET transitioned from NCP to
TCP/IP in 1983, solidly entrenching the protocol in defense organizations.
By 1985, the Internet was established as the network supporting a
broad community of researchers and developers. It was also beginning
to be used by other organizations for daily computer communications.
E-mail was the communication method of choice by these users, and
soon commercial enterprises would follow suit.
That year, the National Science Foundation's NSFnet program announced
its intent to fund Internet access for the higher education community.
The NSF also co-authored with the Internet Activities Board RFC 985,
which specified requirements for Internet gateways - also known as
"routers" - and helped ensure interoperability between the DARPA and
NSF partitions of the Internet.
NSF also enforced an "Acceptable Use Policy," which prohibited use
of the national NSFnet backbone for purposes "not in support of research
and education." This, however, op-ened local and regional usage to
commercial network traffic, spawning the creation of service providers
PSI, UUNET, ANS CO+RE and others.
To ensure fair access and use among educational, research and commercial
interests, Kahn's Corporation for National Research Initiatives formed
the Internet Society in 1991, and Cerf headed it up. This is a year
after ARPANET was decommissioned, and three years after U.S. Sen.
Al Gore (D-Tenn.) lobbied for federal funding of the "information
superhighway" based on the NSF-commissioned report "Towards a National
While heading the Internet Society, Cerf sought permission to interconnect
MCI Mail, a commercial e-mail service, to the NSFnet backbone. Permission
from the Federal Networking Council was granted.
"It was the first time that anybody was allowed to put a commercial
service up on the Internet," Cerf says.
The floodgates for commercial and enterprise network traffic on the
'Net burst open after that. The World Wide Web, which was invented
by Tim Berners-Lee, was developed and widely deployed on the Internet
making the network "user-friendly" for a broad community of nontechnical
users. The WWW Consortium was formed to develop protocols and standards
associated with the Web.
In 1993, National Center for Supercomputing Applications programmers
Marc Andreessen and Eric Bina developed the Mosaic user interface
for the X, Microsoft Windows and Apple Macintosh operating environments.
This development spawned the creation in 1994 of Mosaic Communications,
a maker of graphical user interfaces for the Internet and Web, or
"Web browsers." Mosaic changed its name to Netscape later that year.
The Web began to show enterprise users the potential of the Internet.
"I think the World Wide Web is the best example of [the Internet]
vision," Roberts says. "People can now [access data from virtually
any computer in the world] very effectively."
In 1995, the NSFnet backbone was opened to commercial traffic through
for-profit service providers such as UUNET and PSINet. Welcome to
the Age of the Internet. Roberts and the three other "founding fathers"
of the Internet agree that today's capabilities are barely scratching
the surface of the 'Net's potential. The full realization of that
potential is in the mind of the user.
"I am increasingly convinced we will put all of our communications
media, including television and radio, over the Internet," Cerf says.
Contact Edge Managing Editor Jim Duffy
Other recent articles by Duffy