|
|||||||||||||||||||||
|
|||||||||||||||||||||
 |
Looking Back On Three Decades Of Internet History |
||||||||||||||||||||
|
Creating an Electronic QuiltFor all the grandiosity of the vision, the Internet developed haltingly. To make Licklider's Galactic Net a reality, scientists and engineers had to design a communications system that operated like none that had come before.The obvious example to emulate might have been the telephone network, in which long-distance calls from one small town to another were routed like airline flights through one or more central hubs. But in the early 1960's, researchers began to realize that a computer network would be much less vulnerable to failure if it was more widely spread out -- less like the air travel system than like the network of back roads weaving together every municipality in the country. Each point is connected to its nearest neighbors by several redundant paths. If a connecting node between A and B fails, it is easy to find an alternative route. But there was an even more radical difference between the networks of old and the one that was about to take root. In an ordinary telephone system, two phones were linked by forming a temporary circuit, a dedicated physical channel through which the electrified voices flowed. The telegraph used a different technology called message switching: each telegram was given an electronic address, then sent into the Net, where it would be relayed, node by node, to its destination. Paul Baran at the RAND Corporation and Donald Watts Davies at the British National Physical Laboratory independently saw the advantages of taking this "store and forward" model a step further. A message from one computer to another would be chopped up into many little packets, which would be sent swarming through the network to find their way. Some would take this route, some would take that. Each would carry a destination label along with instructions for where the packet fit inside the overall message. No matter in what sequence the pieces arrived, they could be reassembled. In this system, called packet switching, there would be no need to tie up a circuit for a single transmission. And since the messages were broken into smaller fragments, the flow would be smoother. Finally, if a packet got corrupted, one could just resend it and not the whole transmission. Around the same time, Dr. Leonard Kleinrock, who would preside over the establishment of the first node at U.C.L.A., was using a mathematical tool called queuing theory to understand how data would flow in a packet-switched network. By 1966, with these and other basic ideas in place, Lawrence Roberts, recruited by ARPA from the Massachusetts Institute of Technology, mapped out the plan. Two years after the first transmission, the number of host computers grew to 23. The @ symbol was invented in 1972, and a year later 75 percent of the Arpanet traffic was E-mail. It was starting to look like the Net. Before long, other organizations, like the National Science Foundation, wanted their own networks. By the late 1970's, Dr. Robert Kahn and Dr. Vinton Cerf of ARPA were putting the finishing touches on the lingua franca, inelegantly called TCP/IP, that would weave the patches into the electronic quilt called the Internet.
data link, October 1999. |
||||||||||||||||||||
|
|
|||||||||||||||||||||
Looking Back ...
