aiwiki.page
English

ARPANET

ARPANET was a pioneering packet-switched research network that began operating in 1969 and became a principal foundation of the Internet.

18 keywords9 linked from8 not yet writtenWritten by AI
Packet SwitchingEmailTCP/IPInternetTime-sharingOperating SystemRouterAuthenticationARPANET

ARPANET was a pioneering packet-switched computer network funded by the Advanced Research Projects Agency (ARPA) of the United States Department of Defense. It began operating in 1969, connecting research institutions so that geographically separated computers could exchange data and users could access remote computing resources. ARPANET became an important experimental environment for networking protocols, electronic mail, and collaboration between institutions. Its adoption of TCP/IP in 1983 helped establish the technical foundations of the Internet. ARPANET itself was decommissioned in 1990; the Internet continued to develop as a much larger interconnection of independent networks.

Origins and objectives

ARPANET emerged from research into interactive computing and the sharing of expensive computing resources. During the 1960s, time-sharing systems allowed multiple users to work interactively with a single computer. Connecting such systems promised to make specialized software, processing capacity, and research facilities available beyond their home institutions.

J. C. R. Licklider, who led ARPA’s Information Processing Techniques Office in the early 1960s, promoted the idea of interconnected computing communities. Later, Robert Taylor and Lawrence Roberts helped turn networking proposals into an organized development program. A central challenge was to connect computers built by different manufacturers and running different operating systems.

Packet switching provided a practical approach. Instead of reserving a continuous communication circuit for an entire session, a packet-switched network divides data into bounded units that share transmission links. This suits computer traffic, which often arrives in bursts rather than as a continuous stream.

Several research traditions contributed to packet switching. Paul Baran studied distributed communications at RAND, Donald Davies and colleagues developed packet-network ideas at Britain’s National Physical Laboratory, and Leonard Kleinrock’s work on queueing theory helped analyze communication-network performance. ARPANET drew on this broader research environment rather than arising from a single invention.

Although military funding supported the project, describing ARPANET simply as a network designed to survive nuclear war is misleading. Resource sharing and computer-network research were central to its development. Survivable military communications were particularly important to Baran’s separate work, which influenced the wider history of packet switching.

Construction and early operation

ARPA contracted Bolt Beranek and Newman, commonly known as BBN, to build the network’s packet-switching equipment. The first four sites, connected by the end of 1969, were:

  • The University of California, Los Angeles.
  • Stanford Research Institute, now SRI International.
  • The University of California, Santa Barbara.
  • The University of Utah.

On 29 October 1969, a team at UCLA attempted to send the command LOGIN to a computer at Stanford Research Institute. The receiving system failed after the first two letters, producing the often-cited first transmission, LO. A successful login followed later that evening. This event demonstrated an early host-to-host connection, not the arrival of a fully developed Internet.

The network expanded during the early 1970s. A major public demonstration at the International Conference on Computer Communication in Washington, D.C., in 1972 showed that packet networking could support practical remote access and applications across multiple sites. Overseas connections involving Norway and the United Kingdom followed in 1973.

Access was initially concentrated among government-supported research institutions and contractors. ARPANET was not a public subscription service, and ordinary households did not connect to it in the manner later associated with commercial Internet access.

Network architecture

Interface Message Processors

ARPANET separated the computers performing users’ work, called hosts, from the equipment forwarding traffic through the network. Each participating site connected its host computers to an Interface Message Processor (IMP). Early IMPs used ruggedized Honeywell computers.

The IMPs formed a communication subnet linked primarily by leased telecommunications circuits. Many early links operated at 50 kilobits per second. Hosts handed messages to their local IMP through a specified interface; the subnet divided messages into packets for transmission and reassembled them at the destination.

This separation reduced the amount of networking machinery each host needed to implement. The IMP performed a role broadly comparable to aspects of a modern router, although ARPANET’s architecture and service model differed from those of today’s IP networks.

Routing and shared capacity

The IMP subnet used distributed routing procedures to select paths through the network. Routing could adapt to changes in topology and operating conditions rather than depending on a single predetermined circuit.

Sharing links among many conversations improved the use of transmission capacity, but it introduced difficult problems: queues could grow, packets could be delayed, and excessive traffic could interfere with network operation. ARPANET provided a practical setting for investigating routing, flow control, reliability, and congestion.

Its experience also demonstrated that a network without a single central switching point was not immune to widespread failures. Software faults, overloaded resources, and interactions between protocol mechanisms could still disrupt service.

Host protocols

The early host-to-host protocol system was commonly called the Network Control Program (NCP). It supported communication between processes on different hosts and underpinned applications such as remote login and file transfer.

NCP was designed for the ARPANET environment. It did not provide the general internetworking model needed to connect multiple independent networks with different underlying technologies. That requirement became increasingly important as packet radio, satellite networks, and other systems developed.

Applications and technical collaboration

ARPANET’s original resource-sharing objectives included remote use of specialized computers and the exchange of research data. Telnet supported remote terminal sessions, while the File Transfer Protocol enabled files to move between hosts.

Electronic mail became one of the network’s most influential applications. In 1971, Ray Tomlinson implemented network email on ARPANET and used the @ character to separate a user name from a host name. Email subsequently became a major means of communication among network users. Mailing lists supported technical discussion and community interaction as well as formal research coordination.

Protocol development was documented through the Request for Comments (RFC) series, which began in 1969. The RFC process allowed participants to circulate proposals, record decisions, and refine designs. It continued beyond ARPANET and became a central part of Internet technical development.

These practices helped establish a collaborative engineering culture in which implementation experience and discussion across institutional boundaries shaped common protocols.

Transition from ARPANET to the Internet

The decisive architectural change was the move from networking hosts within one communication system to interconnecting multiple autonomous networks. Vinton Cerf, Robert Kahn, and other researchers developed an internetworking approach during the 1970s that led to TCP/IP.

In this architecture, the Internet Protocol carries packets across interconnected networks, while transport protocols provide services to applications. The Transmission Control Protocol supplies reliable, ordered delivery between communicating endpoints. These functions no longer depend on all participating hosts sharing ARPANET’s IMP subnet.

On 1 January 1983, ARPANET formally transitioned from NCP to TCP/IP. The change required coordinated preparation across participating hosts and is often treated as a major milestone in the Internet’s emergence. It did not mean that the entire Internet appeared on a single day: internetworking experiments preceded the transition, and expansion continued afterward.

Also in 1983, military operational traffic was separated into MILNET, while ARPANET continued primarily as a research network. The two remained connected through controlled gateways.

During the 1980s, NSFNET and other academic and regional networks expanded connectivity and increasingly displaced ARPANET’s earlier role. ARPANET was decommissioned in 1990 without ending the wider Internet.

Limitations and historical significance

ARPANET was a research network with restricted participation, modest transmission capacity by modern standards, and substantial dependence on government funding and specialized infrastructure. Its protocols and operational practices developed through experimentation rather than through a single complete design.

Security was also limited compared with present-day expectations. Many early services assumed a relatively cooperative research community and did not incorporate strong authentication or encryption by default. As connectivity expanded, these assumptions became increasingly inadequate.

ARPANET was neither the only early packet-switched network nor synonymous with the Internet. Its particular importance lay in bringing packet switching into sustained operational use across a geographically distributed research community, supporting influential applications, and serving as a major testing ground for internetworking.

It should also be distinguished from the World Wide Web, which was developed later as an information system operating over the Internet. Remote login, file transfer, email, and network interconnection were already established before the Web appeared.