ARPANET

The Advanced Research Projects Agency Network (ARPANET) was not the Internet itself, but it was the first wide-area packet-switched network with distributed control. It was also one of the first computer networks to use the TCP/IP protocol suite. Both of these technologies became the technical foundation of the Internet. The ARPANET was set up by the Advanced Research Projects Agency (now DARPA) of the United States Department of Defense.

Bob Taylor started the ARPANET project in 1966. He was building on the ideas of J. C. R. Licklider, and the goal was to allow resource sharing between computers that were far apart. Taylor chose Larry Roberts as the program manager. Roberts made important decisions about the request for proposal for building the network, and he used the packet switching ideas of Donald Davies. He also asked for advice from Paul Baran. In 1969, the contract to build the Interface Message Processors (IMPs) went to Bolt Beranek & Newman (BBN), and the design was led by Bob Kahn, who created the first protocol for the network.

The first computers were connected in 1969. The Network Control Program was put in place in 1970, and the network was declared operational in 1971. Soon after, new software allowed people to use remote login and file transfer, which provided an early form of email. The network grew quickly, and in 1975 its control passed to the Defense Communications Agency.

Later, Bob Kahn and Vint Cerf worked together to create the Transmission Control Program for internetworking, which is the joining of networks. This work allowed separate networks to be connected into a network of networks. It drew on ideas from the French CYCLADES project led by Louis Pouzin and from Bob Metcalfe at Xerox Parc. TCP/IP version 4 was put into production use on the ARPANET on 1 January 1983. The ARPANET was formally shut down in 1990, after partnerships with partnerships with the telecommunication and computer industry helped the wider network, known as the Internet, grow and become commercial.

Inspiration

Early voice and data communications used circuit switching, as in the traditional telephone network, where each call got its own dedicated connection between the two stations. In the early 1960s, Paul Baran at the RAND Corporation challenged this model with ideas for systems that could keep working even if parts were destroyed. Donald Davies at the United Kingdom's National Physical Laboratory (NPL) arrived at a similar concept in 1965.

In April 1963, computer scientist J. C. R. Licklider of Bolt Beranek and Newman (BBN) wrote about the idea of an "Intergalactic Computer Network" that would let computer users communicate widely. Later that year, he became head of programs at the Advanced Research Projects Agency (ARPA). Ivan Sutherland and Bob Taylor continued developing the idea, so that ARPA-funded researchers could share computers and quickly distribute new software. Taylor's frustration with using three separate computer terminals led to the idea that became the ARPANET. Davies' work also drew attention from ARPANET developers in October 1967. In August 1968, he gave the first public presentation on packet switching, a term he coined, and used it in the NPL network. The NPL network and the ARPANET were the first two networks in the world to use packet switching.

Creation

In February 1966, Bob Taylor convinced ARPA's Director Charles M. Herzfeld to fund a network project, and Herzfeld moved one million dollars to Taylor's budget from a ballistic missile defense program. In January 1967, Taylor hired Larry Roberts to work on the ARPANET in the ARPA Information Processing Techniques Office. Roberts asked Frank Westervelt to study message size and contents for the network. At an April 1967 design session, technical standards were discussed. Wesley Clark then suggested using minicomputers as an interface to create a message switching network, and Roberts adopted this idea and named them Interface Message Processors (IMPs). The plan was presented in October 1967 at the first Symposium on Operating Systems Principles. Ideas about packet switching from Donald Davies, presented by a colleague, Roger Scantlebury, and from Paul Baran also reached the ARPA investigators there. By mid-1968, Roberts and Barry Wessler had written a final version of the IMP specification, based on a report from Stanford Research Institute (SRI). After approval, a Request for Quotation was issued, and in January 1969 ARPA awarded the contract to build the network to BBN.

The BBN team, led by Frank Heart for the IMPs and Bob Kahn for the network's theoretical design, included Dave Walden, Severo Ornstein, William Crowther, and others. Their network followed Roberts' plan and used small IMP computers, similar to the later idea of routers, to connect local resources. The IMPs were linked by leased lines through modems at an initial speed of 50 kbit/s, and host computers connected to them through custom serial communication interfaces. The first IMPs used a rugged computer version of the Honeywell DDP-516, with 24 KB of expandable magnetic-core memory and a direct memory access unit. The first network connected one computer in Utah with three in California, and later the Department of Defense allowed universities to join for sharing hardware and software resources over early Digital Signal 0 leased telephone lines.

Implementation

The first four computers on the network were used as a testing ground for building and fixing the 1822 protocol, which was a major task. They were connected electronically in 1969, but people could not yet run network programs until the Network Control Program was added in 1970. This allowed the first two ways for computers to work together: Telnet for remote login and FTP for sending files. The network was declared working in 1971, and traffic began to grow once email was set up at most sites by about 1973.

The first four sites were UCLA, ARC, UCSB, and the University of Utah School of Computing. At UCLA, Leonard Kleinrock studied how well the network performed and tested his ideas about message delay. On 29 October 1969, Charley Kline tried to log in from UCLA to a computer at the Stanford Research Institute. The system crashed after he typed two characters, but after an adjustment he logged in successfully, so the first two characters sent over the ARPANET were "lo". By 5 December 1969, the four-site network was in place. Elizabeth Feinler created the first Resource Handbook for ARPANET in 1969, which led to a directory that made it possible to navigate the network.

By 1971, the network was growing, and by 1973 it had reached more sites. In 1972, the ARPANET was demonstrated at the International Conference on Computer Communications (ICCC). Networks were also studied through queueing theory and mathematical models developed by Kleinrock, which helped the ARPANET expand rapidly in the early 1970s.

Networking evolution

The ARPANET's message processors, called IMPs, were built by BBN and could pass messages between one another, which helped them route data and send it reliably. In 1969, host computers began talking to each other using the 1822 protocol. Because the 1822 protocol could not handle many connections at once, a Network Control Program (NCP) was developed under Steve Crocker, with Jon Postel among the researchers. NCP let application software run across the network using communication protocols, an early step toward the OSI model. Computers at MIT also took part in these early networking efforts.

Starting in 1972, Louis Pouzin and Hubert Zimmermann in France worked on a simpler way of linking networks, called a catenet. Bob Kahn and Vint Cerf led research that produced the Transmission Control Program in 1974, and the International Network Working Group was formed. Testing began the next year, and the software was later redesigned as a modular Version 4 in 1978. It was installed on the ARPANET in January 1983, replacing NCP. By 1989, the Internet protocol suite was complete, which helped lay the foundation for the Internet.

Operation

ARPA was meant to pay for advanced research. The ARPANET was a research project that focused more on communication than on the people who used it. In the summer of 1975, operational control of the ARPANET passed to the Defense Communications Agency. Around that time, the first ARPANET encryption devices were set up to protect secret messages. A report written in 1978 and published in 1981 by BBN and DARPA concludes that the ARPANET program had a strong and direct effect on computer science, the field from which the network itself came.

In 1981, the National Science Foundation (NSF) funded the Computer Science Network (CSNET), which expanded access to the ARPANET.

NCP gave computers on ARPANET a standard set of network services that many programs could share. This led to application protocols, which are sets of rules for specific tasks, that worked mostly on their own and could improve without changing the underlying network. Programs such as TELNET, used for remote time-sharing access, and File Transfer Protocol (FTP) were later adapted to run over the TCP/IP protocol suite. Telnet was developed in 1969, and Abhay Bhushan wrote the first FTP specification, published as RFC in 1971. By 1973, FTP had been defined and used to transfer files over the ARPANET.

In 1971, Ray Tomlinson of BBN sent the first network e-mail. A study by ARPA in 1973 found that three-quarters of ARPANET traffic was email, and email stayed a very large part of the traffic. In the 1980s, FTP was replaced for email by the Simple Mail Transfer Protocol, and later by POP and IMAP. The Network Voice Protocol (NVP) was defined and used in 1977, but because of technical problems, conference calls over the ARPANET never worked well. Packet voice, known today as Voice over Internet Protocol, was still decades away.

Security

The Purdy Polynomial hash algorithm was developed for the ARPANET in 1971 to protect passwords. It was made at the request of Larry Roberts, who was head of ARPA at that time.

Later, Digital Equipment Corporation (DEC) used the algorithm to protect passwords in its VMS operating system, and it is still being used for this purpose.

Rules and etiquette

Because the ARPANET was paid for by the government, some kinds of use were discouraged or not allowed.

Leonard Kleinrock claims to have committed the first illegal act on the Internet. After a meeting in England in 1973, he sent a request for the return of his electric razor. At that time, using the ARPANET for personal reasons was against the law.

In 1978, Gary Thuerk of Digital Equipment Corporation (DEC) broke the network's rules by sending the first mass email to about 400 possible clients over the ARPANET. He claims this brought in $13 million in sales of DEC products and showed how useful email marketing could be. A 1982 handbook on computing at MIT's AI Lab said that using the ARPANET for anything not directly supporting government work was illegal. Personal messages, such as arranging a meeting or saying hello, were generally not seen as harmful. Sending emails for profit or politics was described as harmful to others and illegal, and it could get MIT into serious trouble.

Decommissioning

In 1985, the NSF paid for national supercomputing centers at several universities. In 1986, it also provided network access and connections through the NSFNET project. NSFNET became the Internet backbone for government agencies and universities.

The ARPANET project was officially shut down in 1990. Its original machines were phased out as NSFNET came into use, though some were still running as late as July 1990. After the ARPANET was shut down on 28 February 1990, Vinton Cerf wrote a poem called "Requiem of the ARPANET."

It was the first, and being first, was best, but now we lay it down to ever rest. Now pause with me a moment, shed some tears. For auld lang syne, for love, for years and years of faithful service, duty done, I weep. Lay down thy packet, now, O friend, and sleep.

— Vinton Cerf

Legacy

The technology and ideas developed through the ARPANET helped shape modern computer networking, including the Internet. Packet switching, decentralized networks, and communication protocols laid the foundation for a global network that changed how people communicate, share information, and do research together. Many other research projects were related to the ARPANET. Some influenced its design, some were side projects, and some grew out of it. Senator Al Gore wrote the High Performance Computing and Communication Act of 1991, often called "The Gore Bill," after hearing a 1988 concept for a National Research Network sent to Congress by a group chaired by Leonard Kleinrock. The bill passed on 9 December 1991 and led to the National Information Infrastructure, which Gore called the information superhighway. The ARPANET project was honored with two IEEE Milestones, both dedicated in 2009. On 17 May 2011, Arlington County, Virginia, put up two historical markers about the internet's roots at 1400 Wilson Boulevard, in its Rosslyn neighborhood, the agency's home from 1970 to 1975.

Charles Herzfeld, ARPA Director from 1965 to 1967, said the ARPANET was not started to create a command and control system that would survive a nuclear attack, even though many now claim it was. The network came from frustration that there were few large research computers in the country and that many researchers who should have access to them were far away. The ARPANET used distributed computation and frequent recalculation of routing tables, which made it more able to survive large interruptions. The Internet Society agrees with Herzfeld, noting that the nuclear-war idea came from an earlier RAND study of secure communication. Paul Baran, who first proposed a theoretical model for packet switching, conducted that RAND study. Although the ARPANET did not share his goal, Baran said his work helped its development, and minutes from Elmer Shapiro of Stanford Research Institute suggest a version of his routing method might be used. Later, in the 1970s, ARPA did stress "command and control." Stephen J. Lukasik, who was deputy director from 1967 to 1970 and Director of DARPA from 1970 to 1975, explained that the goal was to use new computer technologies to meet military command and control needs against nuclear threats and to improve military decision making.