History of the Internet:Where Did the Internet Come From?


Don't worry, there won't be a test on this stuff, but we think it's kind of interesting.

The Prehistoric Period

The ancestor of the Internet was the ARPANET, a project funded by the U.S. Department of Defense (DOD) in 1969. The ARPANET was both an experiment in reliable networking and an effort to link DOD and military research contractors, including the large number of universities doing military-funded research. (ARPA stands for Advanced Research Projects Administration, the branch of the DOD in charge of handing out grant money. For enhanced confusion, the agency is now known as DARPA -- the added D is for Defense, in case anyone had doubts about where the money was coming from.) Although the ARPANET started small, connecting three computers in California with one in Utah, it quickly grew to span the continent.

The reliable networking part involved dynamic routing. If one of the network links became disrupted by enemy attack, the traffic on it could be rerouted automatically to other links. Fortunately, the Net rarely has come under enemy attack. Cutting a cable during road construction (known in the biz as backhoe fade) is just as much of a threat, however, so it's important for the Net to be backhoe-resistant.

Because the ARPANET was wildly successful, every university in the country wanted to sign up. This success meant that the ARPANET began getting difficult to manage, particularly with the large and growing number of university sites on it. It was broken into two parts:

  • MILNET, which had the military sites
  • The new, smaller ARPANET, which had the nonmilitary sites

The two networks remained connected, however, thanks to a technical scheme called IP (Internet Protocol), which enabled traffic to be routed from one network to another as necessary. Because all the networks that are connected in the Internet speak IP, they all can exchange messages.

Although it started with just two networks, IP was designed to allow for tens of thousands of networks. An unusual fact about the IP design is that every computer on an IP network is, in principle, just as capable as any other, so any machine can communicate with any other machine. (This communication scheme may seem obvious, but at the time most networks consisted of a small number of enormous central computers and a large number of remote terminals, which could communicate with only the central systems, not with other terminals.)
Technical Stuff

Can the Internet really resist enemy attack?

It looks that way. During the Gulf War in 1991, the U.S. military had considerable trouble knocking out the Iraqi command network. It turned out that the Iraqis were using commercially available network routers with standard Internet routing and recovery technology. In other words, dynamic routing really worked. It's nice to know that it works, although perhaps this was not the most opportune way to find out.

Meanwhile, back at the classroom

Beginning around 1980, university computing was moving from a small number of large time-sharing machines, each of which served hundreds of simultaneous users, to a large number of smaller desktop workstations for individual users. Because users had gotten used to the advantages of time-sharing systems, such as shared directories of files and e-mail, they wanted to keep those same facilities on their workstations. (They were perfectly happy to leave behind the disadvantages of time-shared systems. A sage once said, "The best thing about a workstation is that it's no faster in the middle of the night.") Most of the new workstations ran a variety of UNIX, a popular (and, for many versions, free or close to it) type of operating software that had been developed at AT&T and the University of California at Berkeley. The people at Berkeley had a government contract to improve computer networking, so their version of UNIX included all the software necessary to hook up to a network. Because workstation manufacturers also began to include the necessary network hardware, all you had to do to get a working network was to string the cable to connect the workstations, something that universities could do for cheap because they usually could get students to do it.

Then, rather than have one or two computers to attach to the ARPANET, a site would have hundreds. What's more, because each workstation was considerably faster than an entire 1970s multiuser system, one workstation could generate enough network traffic to swamp the ARPANET, which was getting creakier by the minute. Something had to give.

Enter the National Science Foundation

The next event was that the National Science Foundation (NSF) decided to set up five supercomputer centers for research use. (A supercomputer is an extremely fast computer with a hefty price, like $10 million apiece.) The NSF figured that it would fund a few supercomputers, let researchers from all over the country use the ARPANET to send their programs to be "supercomputed," and then send back the results. The plan to use the ARPANET didn't work out, for a variety of reasons -- some technical, some political. The NSF, never shy about establishing a new political empire, then built its own, much faster network -- the NSFNET -- to connect the supercomputing centers. Then, it arranged to set up a bunch of regional networks to connect the users in each region, with the NSFNET connecting all the regional networks.

The NSFNET worked like a charm. By 1990, in fact, so much business had moved from the ARPANET to the NSFNET that, after almost 20 years, the ARPANET had outlived its usefulness and was shut down. The supercomputer centers the NSFNET was supposed to support turned out to be a fizzle: Some of the supercomputers didn't work, and the ones that did were so expensive to use that most potential customers decided that a few high-performance workstations would do just as well. Fortunately, by the time the supercomputers were on their way out, the NSFNET had become so entrenched in the Internet that it lived on without its original purpose. By 1994, several large, commercial Internet networks had grown up within the Internet, some run by large, familiar organizations such as IBM and Sprint and others run by such specialist Internet companies as PSI and Alternet. The NSFNET has been wound down, with its traffic taken over by commercial networks.

Although the NSFNET permitted traffic related only to research and education, the independent, commercial IP network services can be used for any legal purpose. The commercial networks connect to the regional networks just as the NSFNET did, and they provide direct connections for customers.

Outside the United States, IP networks have appeared in many countries, either sponsored by the local telephone company (which is usually also the local post office) or run by independent national or regional providers. The first international connections were in 1973 with England and Norway. Almost all countries are connected directly or indirectly to some U.S. network, which means that they all can exchange traffic with each other.

The term Internet first appeared in 1982, with the DARPA launch of the Internet Protocol (IP). So know you know!

If you want to know more, be sure to check out these sites:

If you'd like to be notified when we update this area of Internet Gurus Central, leave your e-mail address here. (We'll only use this to send you updates about this site, and will not give your address to anyone else, not even our beloved publisher. We'll send you a message to which you must respond to confirm that you want updates, which also tells you how to get off the list if you change your mind later.)

Your e-mail address:
Your name:
(optional but appreciated)

Updated: Jun 3, 2005

Internet Gurus Home Search This Site Our Books
Internet Gurus Central
"...For Dummies" is a registered trademark of Wiley Publishing, Inc.
© Copyright 1998-2011 I.E.C.C. Last update August 5, 2009.