This chapter provides a brief history of the Internet. And yet, no exact history can be written about the Internet because the Internet is not an easily definable thing. It is a consensus of ideas, an agreement among friends and colleagues, a reflection of technological trends. It is evidence of the notion that communication among peoples is a good thing; it is a quiet affirmation of individual initiative. In short, the Internet is a very large concept.
Therefore, what you will read in the following pages is the history of a number of discrete events that, when combined in history, resulted in the Internet.
The 1960s were a peculiar time in the United States. The start of the decade saw the arrival of nuclear missiles in Cuba. The simmering Cold War with Russia rose to a near boil; the threat of nuclear annihilation was a constant in nearly everyone's daily life.
Concurrent with the blockade of Cuba, the beginning of the Vietnam conflict, and political intrigue in many Third World countries, the Cold War was being fought in research labs, fueled by federal spending and public fear. It was thought that the ability to create and keep a technological edge would determine the winner of the war. Technological advances were coming in a rush, and nowhere were they coming more quickly than in the field of computers.
By the late 1960s, every major federally funded research center, including for-profit businesses and universities, had a computer facility equipped with the latest technology that America's burgeoning computer industry could offer.
The idea developed quickly that these various computer centers could be connected to share data. But the actual means by which they would be connected was colored by the ever-present Russian threat. Any network linking these defense-related centers had to be capable of withstanding disruption by a nuclear attack.
The Advanced Research Projects Agency (ARPA) within the Department of Defense was charged with finding the best way to interconnect these various computer sites.
The government's research did not start in a vacuum. Both the National Physics Lab in the United Kingdom and France's Societé Internationale de Télécommunications Aeronautiques were experimenting with a means of intercomputer communications called packet switching, which provides tremendous flexibility and reliability in moving commands and data from one computer to another.
PACKET SWITCHING VERSUS CIRCUIT SWITCHING
ARPA funded a study by the firm Bolt Beranek and Newman (BBN) to find out how communications between these research centers and military installations could be maintained in spite of a nuclear attack. By 1969, BBN had come up with a packet-switching network protocol called the Network Control Protocol and had designed a network controlling computer called an Information Message Processor (IMP), which could manage the network tasks for mainframe computers. The very first IMP was installed at UCLA that same year. By 1970, the first packet-switched computer network in the United States had been created. As shown in Figure 2.2, ARPAnet connected the University of California at Los Angeles, the University of California at Santa Barbara, Stanford University, and the University of Utah in Salt Lake City.
FIGURE 2.2. The original ARPAnet connected four university campuses.
This was the start of the Internet—four universities connected by a packet-switching network funded by ARPA. If any one link of the network failed, information could still be routed along the remaining links. This satisfied the original criteria for developing a computer network that could withstand hostile attack.
By using packets for communications, each computer was at a peer level with every other computer on the network. This arrangement decentralized network control. No one computer was the master and all had equal standing on the network. This fundamental design element was key in encouraging the growth of networks throughout the world and the eventual linking of many of these networks into one world-wide Internet.
By 1972, there were 40 different sites attached to ARPAnet. The electronic traffic between these sites included small text files sent between individual users—a transfer called electronic mail or e-mail. The University of Utah was the first to control a remote computer over the network—a process called remote login or rlogin. Large text and data files were transferred between computers on ARPAnet using File Transfer Protocol (FTP). Thus, by 1972, the core technology was in place.
In 1972, the first International Conference on Computer Communications was held in Washington, D.C. Attended by representatives from around the world, the conference sought an agreement about communication protocols between different computers and networks. Vinton Cerf, who was involved in the establishment of the ARPAnet at UCLA, was named the first chairman of the InterNetwork Working Group, a group that was charged with creating a protocol that could be used by nearly any computer network in the world to communicate with any other network.
The year following the ICCC, ARPA, newly renamed Defense Advanced Research Projects Agency (DARPA), began a program called the Internetting Project to study how to link packet-switching networks together.
These two projects resulted in the development and introduction of the two basic Internet protocols. In 1974, Vinton Cerf and Robert Kahn released the Internet Protocol (IP) and the Transmission Control Protocol (TCP). These two protocols defined the way in which messages (files or commands) are passed among computer networks on the Internet.
TCP/IP—AN OVERVIEW
The work done by Kahn and Cerf continues to serve the Internet community. TCP/IP is the protocol of choice in most new networks established today. The approach used in TCP/IP is so straightforward that the original goal of creating a communications pathway among many different kinds of networks using their own internal protocols continues to be met.
It was, however, a curious, counter-intuitive event (given the times and the Cold War fears) that truly made the Internet broadly available: DARPA decided to release TCP/IP to the world, free of charge, with no restrictions. In other words, a core technology that solved the problem of computer-network reliability in times of war was suddenly released to the world.
The next part of the Internet story involves the development of a new concept in computer operating systems and a low-cost minicomputer.
Digital Equipment Corporation (DEC) was one of the early developers of the minicomputer, a breakthrough in relatively low-cost computers for the masses (as opposed to the large mainframes from IBM and Control Data that cost hundreds of thousands or even millions of dollars). DEC developed the PDP series of computers, followed in the early 1970s by the VAX family of computers. These moderately powerful computers could be afforded by many colleges, universities, and high-tech businesses. Originally, the VAX computers were only shipped with operating system software called VMS, but that was soon to change.
About the same time, researchers at AT&T Bell Labs were experimenting with a home-grown, multitasking, operating system that ran on DEC minicomputers: a system called UNIX.
UNIX was, from the beginning, an operating system that understood networking. In 1976, Mike Lesk at AT&T Bell Labs created a software package called the UNIX-to-UNIX Copy Program, or UUCP. With UUCP, any UNIX computer with a modem could call any other UNIX computer with a modem and transfer files. AT&T Bell Labs starting shipping UUCP with UNIX version 7 in 1977.
Here was a widely available and affordable computer that could run an operating system that actually had built-in support for networking. The UNIX/DEC combination spread like wildfire throughout industry and academia. Networking was no longer an esoteric act performed on expensive, government-sponsored computer facilities. All those slightly renegade UNIX users quickly understood and adopted the idea of networking.
UNIX was the original "open" system, and it promoted an anarchistic attitude toward computing. Clashes between traditional data processing organizations (with their rightful focus on limited access and security) were the antithesis of the UNIX approach. As much as anything else, UNIX was a game, and its users were global players.
AN UNDERGROUND NETWORK
FIGURE 2.3. Using UUCP to make a local call.
The wide distribution of DEC minicomputers running the UNIX operating system created a very large, casual network of computers running over the public telephone systems. This was the epitome of a decentralized, ungoverned network.
Toward the end of the 1970s, networks were starting to pop up everywhere, and they ran on all kinds of computers.
In 1977, the University of Wisconsin decided to create a network for science researchers. More than 100 researchers used Theorynet to trade e-mail messages with one another.
At this time, ARPAnet was serving a select number of research centers, but not all. Many centers, including the University of Wisconsin, were justifiably concerned that sites connected to the ARPAnet over its high-speed dedicated network were receiving an unfair advantage when compared to non-ARPAnet sites depending on slower telephone lines and UUCP. Wisconsin felt that there might be a real need for another network: one like ARPAnet but focused specifically on computer science. In 1979, a meeting was held between a number of researchers from various universities (including Wisconsin), DARPA, and the National Science Foundation.
That meeting in 1979 turned out to be the launch meeting for the creation of the Computer Science Research Network (CSnet), funded in large part by the National Science Foundation.
The story swings back now to Vinton Cerf. In 1980, Cerf suggested connecting ARPAnet and CSnet with a gateway, using the TCP/IP protocols he and Robert Kahn had developed. Cerf also suggested that CSnet could exist as a collection of several independent networks that shared a gateway to ARPAnet (see Figure 2.4).
FIGURE 2.4. The new CSnet linked to ARPAnet with a gateway.
It can be argued that this was the real birth of the Internet. Keep in mind that the Internet does not exist as a physical entity. You cannot reach your finger out and touch anything that can be called the Internet. The Internet is a collection of independent, free-standing networks that have come to an agreement about how to talk to one another. That is what Vinton Cerf envisioned when he suggested coupling CSnet to ARPAnet.
By 1982, researchers could dial in to CSnet to read and send e-mail both to sites within CSnet and to sites within ARPAnet. Thus was born the physical implementation of Internet.
While CSnet was being created and joined to ARPAnet, Steve Bellovin was busy at the University of North Carolina. Steve had the idea for a software application to create an electronic newspaper.
Actually, the newspaper paradigm doesn't quite fit unless your local newspaper has an editorial section as large as all the rest of the newspaper to which you can respond instantly. News on the Internet is definitely interactive, promoting letters to the editor, letters to authors, and even letters to innocent bystanders.
Steve's concept was turned into Release A of Usenet, developed by Steve Daniel and Tom Truscutt. This first release firmed up the concept of newsgroups and newsgroup hierarchies. The newsgroup hierarchies have since been expanded (and, in one notable case, sidestepped), but the fundamental operation remains the same.
Usenet's function was to provide a network that would enable any user to submit an article that would be routed to all computers on the network. It allowed any user to send a message that all other users on the network could have access to, or to all users that subscribed to one or more specific newsgroups. UUCP was called into action; very soon, a series of computers were calling one another, copying files back and forth. If a user posted an article, it was distributed to the host computer and sent out over the network, moving from host to host.
USENET IS ORGANIZED
By 1983, it seemed that networks were cropping up everywhere. Within the halls of the City University of New York, Bitnet emerged. (Bitnet is an acronym for the Because It's Time Network.) Bitnet is another source of news and opinion. Bitnet uses a mechanism called the Listserv. Under Bitnet operation, if you want to read the postings to a specific group, you subscribe to that discussion group by sending an e-mail subscription request to the appropriate Listserv. Newly posted articles are routed directly to all of the group's subscribers using e-mail.
Bitnet is waning in popularity and use, but there are now lists supported by programs such as listproc, majordomo, and L-Soft. Many former Bitnet sites are porting their lists to the Internet directly.
The use of Listserver software to support discussion groups migrated to other parts of the Internet, and the number of discussion groups is continuing to grow. As of this writing, there are more than 7,500 discussion lists available on the Internet.
In San Francisco, another important network was born: FidoNet. In 1983, Tom Jennings wrote a personal computer bulletin-board system called FidoBBS. The software rapidly grew in popularity, and soon there were Fido bulletin boards across the nation. The following year, Jennings released FidoNet, a networking package that could link all the different Fido bulletin boards using a modem and telephone line; FidoNet allows users to send e-mail to one another to create discussion groups just like Usenet and Bitnet.
By 1987, the UNIX-to-UNIX Copy (UUCP) software originally developed for the UNIX operating system was ported to the IBM PC and its clones; thus, FidoNet could share traffic with Usenet.
Because it is based on the PC, FidoNet is used world wide, linking all kinds of users to the larger Internet family.
In the latter half of the 1980s, a new concern worried the U.S. government, a fear that America's lead in the high-end computing platform business would be eroded by foreign competition. One of the outcomes of that fear was the creation of the National Science Foundation Network (NSFnet), which linked a handful of supercomputer centers across the United States. The purpose of NSFnet was to provide the highest quality computing services to researchers nationwide.
The NSFnet sites were linked by state-of-the-art transmission lines. Each of the sites in turn served as the central point for a local network or networks. The number of key sites grew over the years to more than 14, divided into two different regions: East and West.
While NSFnet was being formed, the existing networks were undergoing transformations, and new networks were being developed.
In 1983, the military portion of ARPAnet was spun off into its own network, Milnet (which soon disappeared from view). ARPAnet, the grandfather of networks, was slowly being supplanted by NSFnet; in 1990, ARPAnet was removed from service.
In 1989, Bitnet merged with CSnet, the Computer Science Research Network established a decade earlier. However, in two short years, CSnet closed down, shunted aside by NSFnet.
Although at this point it may seem that all existing networks were being consumed by NSFnet, nothing could be further from the truth. New, independent networks emerged, including services such as CompuServe, Prodigy, and America Online. Businesses, particularly those involved in research and product development, created huge networks, most of which are linked to the Internet.
At the other end of the scale, alternative, independent networks continue to come and go (there are no rules that say you cannot start your own network any time you like and link it into the vast Internet). Indeed, these small, local, or international networks benefit greatly from the larger, federally supported networks because their packets can "go along for the ride," using the NSFnet high-speed transmission lines to move data from one end of the continent to the other.
As you've seen, the Internet is an amalgam of over 45,000 different, independent networks with a system largely open to all.
Looking back, we see that it could have been far different. Computer networking was started with federal funding. The government could have stepped in at any time during the past 20 years and seriously restricted its development. Fortunately, that didn't happen. Industry as well could have prevented communications with the outside world. That too, for the most part, has not happened.
The result is an Internet that spans the globe. You can send e-mail to the South Pole, Fiji, Germany, and even to the land that had such a large and inadvertent role in the development of networks in the first place—Russia.
