ARPANET to Internet and the Protocol that Lead to Growth
The Internet was born from the ARPANET. This essay will attempt to assess the general differences between the two types of networks and how the creation of the ARPANET led to the birth of the Internet. There were many obstacles to overcome,hardware, software, protocols, communication and general acceptance to name a few, but each problem was solved and a network was built. The ARPANET was a closed network operating between fixed locations the Internet however, is available on a global scale. The rapid expansion of the Internet was assisted by the invention of the TCP/IP protocol and will be discussed in the second portion of this essay.
ARPA stands for Advanced Research Projects Agency (or more correctly Defence Advanced Research Projects Agency) a military organisation attached to the Pentagon that oversees computer research. In the early 1960's the agency was looking at defence strategies and how they could enhance their military capability and strategically protect and improve communication systems using computers.
There were several people at ARPA involved in developing ideas around computer networks namely Bob Taylor, J. C. R. Licklider and Larry Roberts. Coincidentally another government funded research agency was also working on networks and the idea of linking computers into a network can be attributed to Paul Baran and his colleagues at RAND and their documents under the umbrella title of "On Distributed Communications" (Hafner and Lyon 1996). Baran from RAND and Roberts from ARPA even described how networks may look and drew diagrams to show the various forms they may take. The initiative for such networks was their military potential, which is why the main source of funding for network research was the defence agency. However Barans vision went beyond the military aspects of networks and he stated"Is it time now to start thinking about a new and possibly non-existent public utility, a common user digital data communication plant designed specifically for the transmission of digital data among a large set of subscribers?"(P.Baran-Hauben). With defence agencies in charge of funding it seemed unlikely that anyone outside of the scientific and government funded communities would benefit from any future developments in networks.
Roberts presented his plan for an "ARPANET" in 1967 at a conference sponsored by the Association of Computing Machinery, he described the reasons for the network and how the network would operate using a subnet of "interface message processors" or IMPs. The IMPs would interconnect, send and receive data, check errors, verify messages and retransmit data in need. A new set of protocols would need to be established to allow communication between these computers, which would be sighted at various locations. The initial criteria for the IMPs and the network were set out as follows:
The contract to build the network was awarded to BBN(Bolt Beranek and Nueman Corp.) under the responsibility of Frank Heart an expert in building computer systems. Heart chose the Honeywell DDP-516 computer to be the base machine that network would run on because of its reputation for reliability and Input/Output (I/O) capabilities. At BBN Heart assembled a team of programmers and engineers to develop the network using the IMPs.
The IMP Team - Charles Babbage Institute 04/08/00 http://www.cbi.umn.edu/photo/cbi/imp.jpg
The team would have to decide which network operations would be managed by software, and which would be built into the hardware. Honeywell was to develop the I/O devices that would be needed to allow the IMPs to communicate. Communication between the IMPs would be via the normal telephone network using modems and lines supplied by AT&T. Each line would need to have special terminating equipment to allow high data transmission rates. (Hafner and Lyon 1996, University of Melbourne). Before the network could begin operating there were many technical and programming issues that had to be resolved. The prototype IMP delivered to BBN did not work and had to be debugged, software to allow dynamic routing had to be written and flow control programs had to be designed. Messages between the hosts were transmitted over the subnet by "packets" of data, the flow of data controlled by a special signal developed at BBN called "ready for next message". Eventually all these problems were overcome.
The ARPANET 1969: Atlas of Cyberspace 05/08/00 http://www.geog.ucl.ac.uk/casa/martin/atlas/historical.html
Four Universities were chosen as the initial sites for the network the first installation at the University of California Los Angeles (UCLA) on September 1, 1969. Additional sites were added at Stanford Research Institute (SRI) the University of California at Santa Barbara (UCSB), and the University of Utah. With the IMP providing a standard interface to the network the problem of connecting incompatible computers was simplified. Each IMP was then connected to the host computer at the site any problems of connectivity were left to the Universities to solve. By the end of 1969, the initial four-node configuration was complete. The ARPANET was expanded steadily and was able to operate with a wide variety of hardware and software systems. From the four sites that comprised the network in January 1970, the network expanded to thirteen by January 1971 and twenty-three by April 1972(Charles Babbage Institute).
ARPANET Expansion 1971: Atlas of Cyberspace 05/08/00 http://www.geog.ucl.ac.uk/casa/martin/atlas/historical.html
Once more sites were connected the ARPANET grew rapidly. Academics and Scientists began to communicate regularly which led to the development of e-mail programs which in turn led to greater usage. Although the ARPANET was not intended as a message system soon thousands of messages were being carried often leading to logjams. The APRANET was becoming the "backbone" for a great deal of net traffic but was also going to provide links to parallel networks such as CATNET, SATNET, MILNET, NSF, and PRNET that were being developed. It was realised that the host to host Network Control Protocol that had been specified for use on ARPANET would need to be replaced by a more independent protocol (Hafner and Lyon 1996).
The new protocol was twofold in its development. Transmission-control protocol (TCP) was the result of collaboration between Vint Cerf from UCLA and Bob Kahn at DARPA. After studying the problem in 1973 they released a paper in 1974 called "A Protocol for Packet Network Intercommunication". The paper described a new framework, messages would be encapsulated and de-capsulated in "datagrams" (TCP) it also introduced the notion of gateways that would only allow receiving hosts to read the contents of the datagrams (Hafner and Lyon 1996). TCP introduced built in reliability, if data sent was not acknowledged the host would assume the data message had failed and would send the message again. Reliability was shifted away from the network to the destination hosts so that the network only carried the data, which meant that once TCP was perfected anyone could build a network. TCP allowed communication across networks as long as gateway computers were installed that computer could then interpret and route the data to any other network. By 1975, Yogen Dalal had turned the TCP protocol into a full set of specifications and these were implemented in three places BBN, Stanford University and University College, London.
In 1977 Cerf and Kahn among others had demonstrated the first three-network system ARPANET, SATNET and packet radio that used a satellite link to London (Hafner and Lyon). It was during these demonstrations that the idea of separating the part of TCP that dealt with routing packets and create a new protocol IP (Internet Protocol). TCP would be responsible turning messages into datagrams and re-organising them in the right order, detecting errors and resending lost messages. IP would be responsible for routing individual datagrams (Hafner and Lyon). With the separation of the protocols it was possible to build cheap and fast gateways and so speed the growth of networking. TCP officially became TCP/IP in 1978. The TCP/IP protocol was challenged by a protocol called OSI, which was sponsored by the International Organisation for Standards. There was much discussion about which protocol should be used but in the end TCP/IP was chosen. In 1983 the ARPANET switched from NCP to TCP/IP.
The ARPANET Internet: Atlas of Cyberspace 05/08/00 http://www.geog.ucl.ac.uk/casa/martin/atlas/historical.html
The ARPANET essentially started as a military and academic network to support defence strategies and was restricted, operating between contracted defence companies and University campuses. However once development began many new features and protocols were added so that the ARPANET went beyond the original scope of it's inventors early intentions and this eventually led to the development of the Internet. New protocols were needed so that expansion could continue. One of the most important was TCP/IP, which meant that all the new networks could communicate and transmit data to each other.
However the development of the Internet continued and new protocols were introduced, such as HTTP developed at CERN by Tim Berners-Lee and which led to the development of the World Wide Web. Scientists and programmers have continued to introduce new applications that make the Internet easier to use and navigate. The Internet is now available to everyone who owns a computer and has a phone line connected to a network.
BACK to TOP