ARPANET

Logical map of ARPANET, March 1977.

ARPANET was a computer network commissioned by the United States Department of Defense (DOD) to be used as a means of communication between different academic and state institutions. The first node was created at the University of California at Los Angeles (UCLA), and it was the backbone of the Internet until 1990, after completing the transition to the TCP/IP protocol model, which began in 1983.

ARPANET stands for Aadvanced Research Projects Agency Network, that is, the Advanced Research Projects Agency Network, now known as the Defense Advanced Research Projects Agency.

Origins

The concept of a computer network, capable of communicating users on different computers, was formulated in April 1963 by Joseph C. R. Licklider, of Bolt, Beranek and Newman (BBN), in a series of notes discussing the idea of "galactic network". In October of the same year, Licklider was summoned by the ARPA (Advanced Projects Research Agency), a body belonging to the DOD. In this forum, he convinced Ivan Sutherland and Robert "Bob" Taylor of the importance of the concept before he left the Agency, and before any work was started. Meanwhile, at the RAND Corporation since 1958, Paul Baran was working on a secure communications network capable of surviving a nuclear weapons attack, for military purposes. His results were published starting in 1960, and they described two key ideas:

• The use of a decentralized network with multiple paths between two points;
• The division of full messages into fragments that would follow different paths.

The network would be able to respond to its own failures. The final summary of this scheme was presented in 1962 and published in 1964.

At the same time, Leonard Kleinrock was already working on the concept of storing and forwarding messages in his doctoral thesis at the Massachusetts Institute of Technology (MIT). This included an important analysis of queuing theory applied to communication networks (published as a book in 1964). His work did not yet include the idea of fragmentation in data packets.

Lastly, Donald Davies of the National Physical Laboratory (NPL) began connecting all these concepts in 1965, after attending a conference in the UK on time multiplexing. His research took place independently of Baran's work, which he was not aware of until 1966. Coincidentally, it was Davies who began using the term 'package'.

In this state of affairs, four independent research centers (DARPA, RAND, MIT and UK NPL) would be the first experimental ARPANET nodes.

While all this was going on, ARPA and Taylor were still interested in creating a computer network. At the end of 1967, Taylor contacted Lawrence G. Roberts (of the Lincoln Laboratory at MIT) with the aim of leading the project to create the new network. Roberts already knew Davies from the aforementioned conference on multiplexing. Roberts' original concept was to use the technique of time multiplexing, linking machines directly to telephone wires. At one of the first meetings (1967), many participants were not willing to let their computers have to manage telephone lines. One of these participants, Wesley A. Clark, had the idea of using separate small computers just to manage the communications links. This idea allowed the main computers to be unloaded, in addition to isolating the network from the different nature of each computer. On this basis the initial design of the ARPANET began. Roberts presented his first plan at a symposium in 1967. Roger Scantlebury, a Davies collaborator, was present at this same event. Thanks to this meeting they discussed the idea of packet switching, and it allowed Roberts to learn about Baran's work.

Birth of the ARPANET

In the summer of 1968, a complete plan already existed and was approved by ARPA, so a competition was held with 140 potential suppliers. However, only twelve of them submitted proposals. In 1969, the contract was awarded to the company Bolt, Beranek and Newman (known as BBN Technologies or simply BBN) with whom Licklider, creator of the "galactic web" concept, had worked. On October 29, 1969, the first message was transmitted via ARPANET and in less than a month (November 21) the first link between the University of California, Los Angeles and the Stanford Research Institute was established.

BBN's offer quickly followed Roberts' plan. The small computers were called "interface message processors" (IMPs). These implemented the technique of store and forward, and used a telephone modem to connect to other computers (at a speed of 50 kbits per second). The mainframe computers were connected to the IMPs via custom serial ports.

IMPs were initially implemented using Honeywell DDP-516 computers. They had 24KB of main memory with the capacity to connect a maximum of four mainframe computers, and communicate with six other remote IMPs.

BBN had all the necessary hardware and software available in just nine months.^{[citation needed]}

First deployment

The initial ARPANET consisted of four IMPs installed at:

• The UCLA, where Kleinrock created the Network Measurement Center. A SDS Sigma 7 computer was the first to connect.
• The Augmentation Research Center, at the Stanford Research Institute, where Doug Engelbart created the new National Language Services (NLS), an emerging hypertext system. A SDS 940 computer was the first to connect.
• University of California, with an IBM 360 computer.
• The U.S. Department of Graphics, where Ivan Sutherland moved. With a PDP-10 initially connected.

The first ARPANET link was established on November 21, 1969 between UCLA and Stanford. By December 5 of the same year, a 4-node network had been formed, adding the University of Utah and the University of California at Santa Barbara. In March 1970, ARPANET crossed to the East Coast, when BBN itself joined. to network. In 1971, there were already 24 connected computers, belonging to universities and research centers. This number grew to 213 computers in 1981, with a new addition every 20 days on average, and reached 500 connected computers in 1983.

ARPANET TimeLine - en

Developed software and prominent people

The Network Control Program (NCP) protocol was the basis of communications between systems belonging to the ARPANET until 1981, when TCP/IP was designed to allow better network growth. It was first deployed around 1970.

In 1972, Ray Tomlinson of BBN invented email.

In 1973, the File Transfer Protocol (FTP) was already defined and implemented, facilitating the movement of files on the ARPANET. By then, 75% of traffic was already due to email success.

• Will Crowther: ARPAnet programmer

• Dough Engelbart: ARPAnet Administrator at Stanford; mouse inventor.

• Frank Herat: Administrator of ARPAnet's development team.

• Bob Kahn: ARPAnet Developer. TCP/IP coinventor.
• Severo Ornstein: Systems Engineer at ARPAnet.
• Jonathan Postel: RFC Editor at ARPAnet.
• Ivan Sutherland: Director of the Office of Information Techniques in ARPA.
• Dave Walden: Systems Engineer at ARPAnet.

A protocol for voice transmission was also specified (RFC 741), which was implemented but failed for technical reasons. This would not see the light until several decades later.

The “myth” of nuclear attacks

The belief that the ARPANET was built to survive nuclear attacks is still very popular. However, some believe that this was not the reason. While it is true that the ARPANET was designed to survive network failures, they argue that the real reason for this was that the switching nodes were unreliable, as attested in the following quote:

Following a RAND study, the false rumor that ARPANET was designed to resist a nuclear attack was spread. This was never true, only a study by RAND, not related to ARPANET, considered nuclear war in the safe transmission of voice communications. However, subsequent work emphasized the robustness and survival capacity of large portions of the underlying networks.

Notwithstanding the foregoing, the majority opinion currently connects the initial effort of the RAND Corporation, with those of MIT and UCLA, which can be evidenced in this quote:

During the '60s, this intriguing concept of decentralized and bomb-proof package switching network walked smoothly between the RAND, the MIT (Masachussets Institute of Technology) and UCLA (University of California in Los Angeles). The National Physical Laboratory of Great Britain prepared the first test network based on these principles in 1968. Shortly afterwards, the Pentagon's Advanced Research Project Agency (ARPA) decided to finance a more ambitious and larger project in the United States. The nodes of the network were to be high-speed supercomputers (or what was called that at the time). They were unusual and highly valued machines, and they were in need of a good network network for national research and development projects.

However, ARPA was a program financed by the Pentagon, which shows that there were military interests in the creation of the ARPANET. Likewise, the works of Paul Baran, pioneers in the creation of the network, also testify that part of the Pentagon's interests was to achieve efficient systems in possible defenses that the country had to carry out against various attacks, including nuclear ones.