ARPANET

ARPANET (Advanced Research Project Agency Network) was the original and first world-wide packet-switching network, originally created by a small team of research experts funded by DARPA in the United States Department of Defense.

Historical Background

The Creation of ARPA/DARPA

During the Cold War era, one of the primary concerns for the United States government was to find ways of protecting its military communications system, as well as ways to make its command and control network able to survive a nuclear attack from the Soviet Union. As a result of the Soviet Union launching Sputnik in 1957, in 1958 US President Dwight Eisenhower instructed the creation of the Advanced Research Project Agency (ARPA), which later became known as DARPA.

Paul Baran's Packet Switching: The Basis of ARPANET

The basis of the academic research for ARPANET started with Paul Baran, a researcher from RAND, Inc., who would one day be considered one of the fathers of the Internet. In 1962, Baran suggested that a more robust communications network using redundancy and digital technology be designed. Although his idea was dismissed by many, Baran continued to work on developing the idea with colleagues at RAND. He envisioned a method of distributed communications wherein a network of unmanned nodes would serve as switches to route information from one node to another until the information reached its final destination. He later developed packet switching, a method of dividing information into "message blocks" before sending them out to a network. Each message block was to be sent separately and then rejoined as a whole when reaching the final destination.

J.C.R. Licklider Leads IPTO

During the same year, DARPA commissioned Dr. J.C.R. Licklider from Bolt, Beranek and Newman to lead the Information Processing Techniques Office, which was mandated further to develop the Semi-Automatic Ground Environment (SAGE) program and build the first wide-area computer network for the cross country radar defense system, interconnecting the three major defense sites of the United States in a manner that was capable of surviving a nuclear attack.

Licklider envisioned the project based on Baran's method of packet switching. He also visualized a universal network that would allow people to communicate with each other using the computer. He referred to the first group of computer specialists working on the research project as the "Intergalactic Network." Licklider's interest in connecting the community through a computer network resulted in the creation of the ARPANET. He believed in the "promise offered by the computer as a communication medium between people, not as an arithmetic engine." In 1964, Licklider resigned from IPTO as Director and went to work for IBM.

Ivan Sutherland became IPTO Director

Licklider was succeeded by Ivan Sutherland, who developed the Sketchpad program which enabled computer displays to be saved and modified in memory. He also developed computer graphics. Inspired by his predecessors' vision of a universal network, in 1965 Sutherland awarded a contract to Lawrence Roberts from MIT to develop the computer networking technology. Roberts worked with Thomas Marril and together they were able to implement the first packet dial-up telephone connection exchange between a TX-2 computer at MIT and a Q-32 computer in California.

Robert Taylor Approval of ARPANET

In 1966, Robert Taylor succeeded Sutherland as IPTO Director. He lobbied for additional funding to be able to carry out a research project to achieve a distributed communication network. A one million dollar funding was promised to him by DARPA then-Chief Charlie Hertzfeld, and Roberts was hired by IPTO as Chief Scientist. On June 21, 1968, Taylor approved the Resource Sharing Computer Network Report, a plan to create ARPANET, which was prepared by Roberts. ARPANET was officially developed on August 30, 1969. Lawrence Roberts became Director of IPTO the following month.

Development of ARPANET

Network Working Group

In 1968, computer experts from four of the research laboratories receiving funding from IPTO were called together to identify and solve the technical problems associated with the development and completion of the ARPANET. Elmer Shapiro from the Stanford Research Institute (SRI) organized the first meeting, as instructed by ARPA, to discuss host-to-host problems. The meeting was attended by Steve Crocker, who represented UCLA, Steve Carr, from the University of Utah, Jeff Rulifson from SRI, and Ron Stoughton from UCSB. The group decided to meet regularly and became known as the Network Working Group. Crocker describes what transpired during the first meeting with the following statement:

"With no specific service definition in place for what the IMPs [Interface Message Processor] were providing to the hosts, there wasn't any clear idea of what work the hosts had to do. Only later did we articulate the notion of building a layered set of protocols with general transport services on the bottom and multiple application-specific protocols on the top. More precisely, we understood quite early that we wanted quite a bit of generality, but we didn't have a clear idea of how to achieve it. We struggled between a grand design and getting something working quickly."

The NWG's initial advance protocol developments were DEL (Decode-Encode-Language) and NIL (Network Interchange Language), which were intended to instruct how to understand messages received from a sender. In 1969, Bolt, Beranek and Newman (BBN) submitted a detailed report on the host-IMP interface specifications from the IMP side, which provided the NWG a basic starting point to develop the ARPANET, analyze its cost, performance, and reliability.

NWG Started RFC

A Request for Comments (RFC) was started by the NWG when they realized that the discussions during their meetings need to be recorded. Crocker organized the notes and noted the discussions in their meetings with the label Request for Comments. According to Crocker, the NWG had a vision for inter-computer communication but struggled to create a detailed protocol design. The RFCs served as a document to give status updates to the members of the NWG. The RFCs were open and provided an exchange of ideas and information between the computer scientists working on the ARPANET.

The ARPANET was born

The realization of Licklider's vision for the ARPANET as a universal communication network became evident in September 1969 when NWG member Leonard Kleinrock, head of the Network Measurement Center at UCLA and his team, which included Vinton Cerf, Steve Crocker, Bill Naylor, Jon Postel, and Mike Wingfield, connected one of the center's SDS Sigma 7 computers to an IMP. The team from UCLA was able to successfully exchange the message, "Do it to it, Truett" with BBN hardware designer Ben Barker. On that day, ARPANET/the Internet was born.

The second node connected to the ARPANET was the NLS System at SRI, which was developed by Douglas Engelbart. This was the first full ARPANET network connection using an SDS-940 computer with the Genie Operating System and a 50 kbps line from AT&T. However, the first test did not work properly and the system crashed. The second test worked fine according to Kleinrock.

The third connection added to the ARPANET was an IBM 360/75 computer using the OS/MVT operating system from the Culler-Fried Interactive Mathematics Centre at the University of California at Santa Barbara, followed by the DED PDP-q0 computer using a Tenex Operating System from the University of Utah Graphix Department. The first four nodes connected to the ARPANET became operational.

The Network Control Program

Despite the great successes with ARPANET's operation, in 1969, Lawrence Roberts, IPTO's Senior Computer Scientist, met with the NWG in Utah and emphasized the need for further achievement. He guided the group to develop a viable network protocol. In 1971, the group was able to create the Network Protocol Program (NCP), which became the standard networking program for the ARPANET. NCP's main function was to establish, break, and switch connections, and to control the flow of communication between different host computer systems. By the end of 1971, there were already 15 sites connected to the NCP. The NCP became the basis for Robert Kahn and Vinton Cerf to create the modern Internet Protocol Suite.

The first 15 sites connected to NCP of ARPANET were:

1. Bolt Baranek and Newman (BBN)
2. Carnegie Mellon University
3. Case Western Reserve University
4. Harvard University
5. Lincoln Laboratories
6. Massachusetts Institute of Technology (MIT)
7. NASA at AMES
8. RAND Corporation
9. Stanford Research Institute (SRI)
10. Stanford University
11. System Development Corporation
12. University of California at Los Angeles (UCLA)
13. University of California of Santa Barbara (UCSB)
14. University of Illinois at Urbana
15. University of Utah

First International Connections to ARPANET

In 1973, the first two international sites connected to ARPANET were the University College of London in England and the Royal Radar Establishment in Norway. By that time there were already 37 connected locations within the United States, including a satellite link from California to Hawaii.

MILNET Splits from ARPANET

The Military Network (MILNET), which was part of the ARPANET, served as the backbone for the U.S. Department of Defense's unclassified e-mail and Internet traffic, with a total of 65 sites connected to the ARPANET. In 1983, MILNET split from the ARPANET because of security reasons, after which only a few small gateways remained connected to the ARPANET for e-mail exchange purposes only.

ARPANET Became Global

In 1985, connection to the ARPANET became global across the United States, Europe, and Australia. There were already 2,000 hosts connected to the Internet Protocol Suite, which became the standard of inter-networking within ARPANET. Jon Postel, who succeeded Steve Crocker as Chairman of the NWG, assigned IP addresses to the networks connected to ARPANET. Postel, together with Paul Mockapetris and Craig Partridge from USC/ISI and BBN respectively, developed the Domain Name System (DNS). By the end of 1989, connections to the Internet reached around 160,000.

ARPANET Retires

In 1990, the ARPANET officially shut down. By that time there were approximately 300,000 sites connected, and these connections were all moved to National Science Foundation Network (NSFNET), which took over the remaining responsibilities of the ARPANET research network. The network, now a part of NSFNET, continued to spread among academic and research institutions, and soon there was pressure to open the network for commercial use. This was prohibited by NSFNET management to maintain bandwidth usage for research purposes. In response to the demand, several parallel networks began to appear. In 1991, NSFNET finally recognized that the network was growing beyond its original research purposes and amended its usage policy to allow for commercial use. On April 30, 1995, the NSFNET was officially dissolved, paving the way for the growth of the modern Internet.

Defense Advanced Research Projects Agency

Request For Comments

Policy Development Process