It began in 1964, when a researcher at RAND considered the problem of communicating among networked computers when part of the network was out of commission. Paul Baran wrote a series of papers about "distributed" computing, in which he described a group of computers ("nodes") which were connected in such a way that damaged areas--even of significant size--could be routed around.

Network Routing Configurations. Source: Introduction to Distributed Communications Networks, Paul Baran

The question was how the network could be controlled, since any central authority would be an obvious target for an enemy missile. Baran's proposal was that the network would have no central authority, and--even more daring--would be designed from the beginning to operate while in tatters.

Put simply, the network would be assumed to be unreliable at all times.

Put simply, the network would be assumed to be unreliable at all times, and would be designed to transcend that unreliability. All of the nodes in the network would be equal in status to all other nodes, and each node would have its own authority to originate, pass, and receive messages. The messages themselves would be divided into separately-addressed packets. Each packet would begin at some specified source node and end at some other specified destination node, but each would wend its way through the network individually.

The route that any particular packet took was unimportant; only the final results were important. In essence, the packet would be tossed like a hot potato from node to node, approximately in the direction of its destination, until it arrived. The loss of a section--even a large section--of the network wouldn't affect the packet's journey; it would remain aloft, passed along by whichever nodes had happened to survive.

During the 1960's, the concept of such a decentralized packet-switching network was explored by several agencies. The National Physical Laboratory in Great Britain set up the first test network on these principles in 1968. Shortly thereafter, the Pentagon's Advanced Research Projects Agency decided to fund a larger, more ambitious project in the USA. The nodes of the network were the high-speed supercomputers at ARPA contract sites, networked so that researchers at any site could use resources at the other sites.

By December 1969, there were four nodes....In 1971, there were 15...by 1972, 37.

The first such node was installed at UCLA in 1969. By December 1969, there were four nodes on the nascent network, which was named ARPANET, after its Pentagon sponsor. The four computers could transmit data on dedicated high-speed lines, and could even be programmed remotely from the other nodes. ARPANET enabled scientists and researchers to share one another's computer facilities long-distance. In 1971 there were fifteen nodes in ARPANET; by 1972, thirty-seven nodes.

By the second year of operation, however, a somewhat unexpected pattern began to emerge: The main traffic on ARPANET was not long-distance computing, but rather news and personal messages. Researchers were using the network to collaborate on projects, to trade notes on work, and eventually, to...well...network. Not only were they using ARPANET for person-to-person communication, but they were very enthusiastic about this particular service; far more enthusiastic than they were about long-distance computation.

ARPANET grew throughout the '70's, its decentralized structure making expansion easy. Unlike "standard" networks, the ARPA structure could accommodate many different kinds of machine. As long as a machine could speak the packet-switching lingua franca of the new network, their variety, their content, and even their ownership were irrelevant.

ARPANET's original standard for communication was known as Network Control Protocol (NCP), but as time passed NCP was superceded by a higher-level, more sophisticated standard known as TCP/IP. Soon, TCP/IP was being used by other networks to link to ARPANET, and more and more organizations became part of this network-of-networks.

Today there are tens of millions of nodes in the Internet, scattered over forty-two countries.

ARPANET itself formally expired in 1989, a happy victim of its own incredible success. Its users scarcely noticed, for ARPANET's functions not only continued but steadily improved. In 1971, less than three decades ago, there were only four nodes in the ARPANET network. Today there are tens of millions of nodes in the Internet, scattered over forty-two countries, and the explosive growth continues. The robustness and simplicity of the packet-switched network has produced a revolution in the way that people communicate.

[ | | ]