Internet Guide Logo

(iG) The Internet Guide

Menu: ARPANET - NSFNET - TCP/IP - Web - Email - ISPs - DNS
bullet Introduction

The Internet is a loosely-organised collaboration of computer networks that voluntarily communicate and interconnect via open procedures and protocols defined by Internet Standards. The Internet Protocol Suite (TCP/IP) consists of the core communications protocols that are used by these computer networks to interconnect, and become Interconnected Networks. During the development of TCP/IP (1974), Carl Sunshine, Vinton Cerf and Yogen Dalal coined the term 'Internet' in RFC 675. The most important protocol of the Internet Protocol Suite is the Internet Protocol (IP): it creates a datagram format and numbering (address) system that enables packet-switching networks to interconnect. No single individual is credited with inventing the Internet, but a small group of individuals, named the 'Internet Pioneers', are credited with designing the core technical architecture of the Internet in the 1970s and 1980s. While there has been other terms used to describe the Internet -- Cyberspace and the Information Superhighway -- the Internet is the name that has stuck. In 1995, the Federal Networking Council -- a group composed of members from the DoD, NSF and NASA (US organisations) -- passed a resolution that defined the Internet as:

Federal Networking council statement

bullet Developing and managing the Internet

bullet Introduction

From 1966-1992, the U.S. government funded the development of the Internet through a variety of federal agencies. The first wide area network, funded by ARPA, was ARPANET. During the 1980s more federal computer networks were created -- NSFNET, MILNET, EFNET -- and when these federal networks interconnected in the 1980s, they became known as the Internet: Interconnected Networks. The software (protocols) used by these networks was named TCP/IP and its development was funded by DARPA (agency of the U.S. Department of Defense) throughout the 1970s and 1980s. TCP/IP became a defense standard in 1982, and was implemented on the majority of federally funded computer networks in the 1980s.

DARPA decided that a central authority was needed to oversee the development of TCP/IP: in 1983, DARPA created the Internet Activities Board (IAB). By 1987, the IAB had created the Internet Engineering Task Force (IETF) and the Internet Research Task Force (IRTF), to perform the research and engineering of TCP/IP. The protocols of the Internet produced information resources that needed to be managed: the most obvious of which was Internet numbers. Every device, which connected to the Internet, was assigned an Internet Protocol (IP) number, and the way in which these numbers was assigned needed management. Groups, like IANA and NIC, were created to conduct this management.

Due to the success and popularity of federally funded computer networks, like NSFNET, commercially funded U.S. networks wanted to connect to the Internet. At the same time, more and more international networks, like NORDUNET, had connected to the Internet. By the late 1980s, it became obvious that a federally funded Internet would not work for two reasons:

1) The political issues created by allowing commercial companies to profit from being connected to a federally funded Internet backbone.
2) The demand from international and commercial networks wishing to interconnect, and have a 'voice' in the development of the Internet.

In the early 1990s, the U.S. government decided that the Internet backbone network should consist of autonomous commercial computer networks, and the development and management of its protocols and information resources should be administered by international multi stakeholder non-profit organisations.

bullet Internet protocol development

In 1992, the Internet Society (ISOC) was created -- as a non-profit organisation -- to provide leadership for Internet: development, governance, policy and technology. The Internet Society's membership is international in scope, and its development process is open and aims to be of "benefit to all people throughout the world". The development of Internet Standards is outlined in RFC 1336:

How organisations manage the resources of the Internet in RFC 1336

In 1992, the Internet Society replaced DARPA and NSF (National Science Foundation) by providing leadership and a legal standing for the Internet Architecture Board (IAB) and the Internet Engineering Task Force (IETF). In 1983, the IAB was created by DARPA to manage the evolution of TCP/IP, but it had been operating with 'little in the way' of legal protection from vendors. The core protocols (TCP and IP) of the Internet evolved to become the Internet protocol suite, and it is currently managed by the following organisations (listed below in their hierarchy).

  1. Internet Society (ISOC)
  2. Internet Architecture Board (IAB)
  3. Internet Engineering Steering Group (IESG)
  4. Internet Engineering Task Force (IETF)
  5. Internet Research Steering Group (IRSG)
  6. Internet Research Task Force (IRTF)

The Internet Society (ISOC) provides overall leadership, policy and potentially finance. The Internet Architecture Board (IAB) provides the planning and oversight for the organisations who conduct the research and technical development of the Internet. Internet standards are documented in Request for Comments (RFC) documents. RFC documents predate the Internet: it was a document system created during the development of ARPANET. The Internet Society (ISOC) has ensured that the term 'Internet' is free to use and is not trademarked. The Internet Society administrates an award scheme named the Internet Hall of Fame.

bullet Internet namespace management

Alongside the development of Internet Standards (protocols), another resource that needs to be managed is Internet numbers and names. Internet Protocol (IP) numbers, and the Domain Name System (DNS), are the 'heart' of the address system of the Internet, and are referred to as the Internet namespace. Each device that connects to the Internet is assigned an IP address: which is a 32-bit number (IPv4) or 128-bit number (IPv6). One problem with using IP numbers to locate the host of a resource, is the difficulty of remembering them, therefore, domain names were invented -- which are tied to IP addresses -- which makes it easier for users to remember the location of an Internet resource.

The IP and DNS namespace are managed by ICANN, who in turn, has allocated IANA (a department of ICANN), the job of assigning IP number blocks and managing the data in the DNS root zone (highest level). IANA gives IP blocks to five Regional Internet Registries (RIR) -- ARIN, LACNIC, AFRINIC, RIPE NCC, APNIC -- who assign smaller blocks of IP numbers to network operators within their region. ICANN also accredits Internet registries and registrars to manage Top Level Domains in the Domain Name System (DNS) -- the level directly below the DNS root zone -- and dictates the rules by which domain names can be registered by individuals and businesses.

ICANN were originally under-contract to the United States Department of Commerce (DOC), but, on the 1st of October 2016, were freed from U.S. government oversight -- ICANN will still be based in California. The removal of DOC oversight marks a significant transition: an important aspect of the Internet was effectively governed by the U.S. government, but will now be in 'the hands' of an international multi-stakeholder arrangement.

Russia, Iran, China, India, France and Brazil, have all been historical critics of U.S. oversight of ICANN. This criticism was further compounded by the Edward Snowden leaks (2013), which alleged that DNS data was being spied on by the National Security Agency (NSA). The Snowden leaks led for 'calls' that an organisation, like the International Telecommunications Union (affiliate union of the United Nations), should manage the Internet namespace. Internet governance has previously discussed by the UN's Working Group on Internet Governance (WGIG) from 2003-2005, and NETmundial Initiative (NMI) in 2014 -- both of these platforms discussed ICANN and Internet namespace management.

bullet Internet History

bullet Origins

The origins of how the Internet came to be are complex, while ARPANET is identified as the computer network that evolved into the Internet, the scientists who inspired the scientists who built ARPANET are harder to identify. In 1948, Claude Shannon wrote an article titled A Mathematical Theory of Communication, he is credited as being the 'father' of information theory; while at MIT, he taught Ivan Sutherland, who would later install one of the first ARPANET nodes. Vannevar Bush and his proposed Memex automatic information management system -- outlined in an Atlantic Daily article titled As We May Think in 1945 -- is often credited with inspiring budding American research scientists. In the 1960s, Marshall McLuhan envisioned a 'global village' which would be achieved through the use of advanced electronics.

The man who is probably cited the most by the scientists who built ARPANET -- as inspiring them to build the network -- is Joseph "Lick" Licklider. While working at MIT in the 1950s, Licklider was involved in the SAGE program: a radar defense system that linked radar site across the United States through a phone line network. Early papers that Licklider published included Man-Computer Symbiosis and On-Line Man Computer Communication (co-authored with Welden Clark). Licklider also published the book Libraries of the Future and a later paper named The Computer as a Communication Device (co-authored with Bob Taylor). Probably the most important memo / paper Licklider wrote, and certainly the one cited the most, is Members and Affiliates of the Intergalactic Computer Network, which outlined his idea of a 'Galactic Network'. It should also be noted that earlier work by Norbert Wiener -- the inventor of cybernetics -- influenced the later work of those scientists who participated in the SAGE program; such as Licklider.

What is without question, is that the origins of the Internet are intertwined with the development of packet switching: packet switching is a method of transmitting data -- in blocks of data called packets -- across digital networks. Paul Baran and Donald Davies independently developed the theory of packet switching: Baran published his paper before (1964) Davies (1966). The three papers listed below influenced or outlined the development of packet switching.

  1. Donald Davies: working at the National Physical Laboratory published a paper A Digital Communications Network For Computers.
  2. Paul Baran: whilst working at the RAND Corporation (United States) published a paper On Distributed Communications Networks.
  3. Leonard Kleinrock: published a paper Information Flow in Large Communication Nets.

It is generally accepted that it was the work of Donald Davies that provided the term "packet", and he oversaw one of the first working examples of packet switching at the National Physical Laboratory (United Kingdom) in 1968; this project created the NPL Mark I computer network. Another influencial, early computer network, was CYCLADES (1973): which invented the use of datagrams (best effort unreliable data transmission). Some other early computer networks that may have had a minor influence upon the design of the Internet include: Societe lnternationale de Telecommunications Aeronautiques (SITAnet), European Informatics Network (EIN), TYMNET, and RCP (Reseau a Commutation par Paquets). However, the most influencial packet switching network was ARPANET: many of the scientists who built APRANET were involved in the design of the Internet. ARPANET is generally regarded as the forerunner to the Internet.

bullet ARPANET era

In 1962, Joseph Licklider, having previously worked at MIT on the SAGE program, was hired as the director of the Information Processing Techniques Office (IPTO) at the Advanced Research Projects Agency (ARPA). While Licklider's stint as the IPTO director was short (two years), his replacements, Ivan Sutherland in 1964, and Bob Taylor in 1966, were both persuaded by Licklider of the importance of a wide area computer network. In 1965, Ivan Sutherland contracted Thomas Marill and Lawrence Roberts to build an experimental computer connect: this experiment proved a success and would lead to more expansive experiments. In 1966, Robert Taylor, building upon on the Marill/Roberts experiment, created a blueprint for a 'wide area' computer network and persuaded Lawrence Roberts to join DARPA and help him build it. Robert Taylor received $1 million in funding from ARPA's director, Charles Herzfeld, to build ARPANET (1966-1968).

ARPANET was one of the first computer networks to use packet switching and was launched in 1969; with procurement initiated in the summer of 1968. The physical hardware of ARPANET was built by Bolt, Beranek and Newman technologies; who developed the Interface Message Processor (IMP) (router) for ARPANET. This technology was outlined in the first ever RFC document (RFC 1). The first IMP was delivered to UCLA, and the team involved in its installation included: Vint Cerf, Bill Naylor, Jon Postel, Steve Crocker, and Mike Wingfield. The first message sent was between an IMP and SDS Sigma 7; which created the first ARPANET node.

The first ARPANET connection, made between two nodes, consisted of nodes located at: SRI International (overseen by Douglas Engelbart) and UCLA (overseen by Leonard Kleinrock). Leonard Kleinrock had conducted early research into a mathematical study called 'queueing theory': this theory proved important in the design of packet switching networks. The first message sent on ARPANET was sent on the 29th of October 1969, and ARPANET initially operated with four nodes - each was installed with an identical Interface Message Processor (IMP). The four nodes were located at the following universities/research institutions: UCLA, SRI, UCSB and the University of Utah. By 1972, there was over 20 IMPs connected to ARPANET. In 1973, ARPANET was connected to its first two international nodes: NORSAR in Norway, and the University College of London (United Kingdom).

ARPANET initially used the NCP (Network Control Program) to transport data packets across the network. NCP was devised by Larry Roberts, Stephen D. Crocker, and a team of graduate students. Stephen D. Crocker said that NCP was not created according to a 'grand plan' but largely occurred accidentally. It was during the design process of the ARPANET protocols, that it was concluded that an official document system was needed: that document system was named Request for Comments (RFC). The Defense Communication Agency (DCA) took control of ARPANET in 1975: by that time ARPANET was fully operational, and ARPA's remit was the development of systems, not the management of them. In the 1980s, the DCA would split ARPANET into a military and an academic research network.

DARPA (ARPA was renamed) continued to fund the development of computer network protocols -- which would improve the efficiency of adding new nodes -- and TCP/IP was the eventual result: becoming a defence standard in 1980 and applied to ARPANET on the 1st of January 1983; a day commonly referred to as 'the day the Internet was born'. ARPANET was phased out from 1985-1989: due to the creation of NSFNET and the Federal Internet Exchange (FIX).

bullet CSNET era

The National Science Foundation helped to fund the development of CSNET (Computer Science Network); beginning in 1979 and finalising the project by 1981. The purpose of CSNET was to expand network access to science departments at U.S. educational institutions. Due to the Defense Communication Agency (DCA) desire to phase out its support for the ARPANET network, it was clear that universities in the U.S. would have to fund a computer network through another means. Side note: the computer network JANET would connect universities in the United Kingdom. Most universities in the U.S. had not been connected to ARPANET because they had not been involved in research related to the Department of Defense; who had funded its development.

Larry Landweber, a Professor Emeritus of Computer Science at the University of Wisconsin-Madison, proposed the creation of CSNET, which would connect U.S. universities who were not connected to ARPANET. Seed funding for the project was provided by the National Science Foundation, with additional support provided by a range of U.S. government departments. Landweber had acquired millions of dollars to fund a new network by 1980. The first universities connected to the network were: Wisconsin-Madison, Princeton, Purdue, and Delaware. By the end of 1983, there was over seventy sites connected to the network. By 1984, Vint Cerf had organised a gateway between CSNET and the ARPANET -- TCP/IP being the protocol used to facilitate it -- and by 1985 a number of international sites had connected to CSNET.

In 1988, CSNET merged with BITNET, but, due to the success of NSFNET, the network was shutdown in 1991.

bullet NSFNET era

In 1985, the National Science Foundation funded the creation of NSFNET: this network would expand upon the capabilities of CSNET, and would use TCP/IP. Dennis Jennings was hired by the NSF to build NSFNET in 1985. NSFNET was operational in 1986, with six supercomputer centers forming its backbone. NSFNET was based on a three-tier model: 1) backbone; 2) regional networks; 3) campus networks. This three-tier model was important, as it would evolve to become the modern day Internet. From 1987-1991, the speed of the NSFNET backbone was increased from 56 K-bit/sec to a 45-Mbit/s. In 1987, NSF had given Merit, IBM, and MCI a five-year contract to upgrade the infrastructure of the NSF backbone. NSFNET was connected to regional networks, like MIDnet, and federal agency networks, like MILNET, and by 1992 it was connected to over 5000 computer networks. By 1988-1990, it was clear that the NSFNET backbone had become the Internet's backbone, and had become a 'network of networks'.

NSFNET was funded by the U.S. government, and commercial use of its 'backbone' network infrastructure was prohibited; at a later stage commercial companies accessed the network under an 'acceptable use' policy. However, NSFNET was far from being 'open' to commercial organisations, and in 1991, the Commercial Internet Exchange (CIX) was created by three networks (using TCP/IP) to freely exchange commercial network traffic. In 1991, Merit, IBM, and MCI created a commercial Internet Service Provider, named ANS CO+RE, that was able to use the NSFNET backbone infrastructure: due to its position of having designed the network infrastructure, and its apparent 'cosy' relationship with the NSF. ANS CO+RE refused to interconnect with the Commercial Internet Exchange (CIX).

The controversy of ANS CO+RE -- where a federally funded project was being using for commercial profit -- led to the High Performance Computing Act of 1991 (U.S.), and the transition of the Internet from a NSFNET backbone to a backbone controlled by commercial companies. In 1991, Network Access Points (NAPs) were created to replace the NSFNET backbone; NAPs would exchange data between commercial backbone networks. The four original NAPs were: Ameritech, MFS Datanet, Sprint, and Pacific Bell. Network Access Points were a strategic 'stepping stone', and were eventually replaced by Internet Exchange Points. NSFNET was decommissioned on the 30th of April 1995, and ushered in the Internet era of commecial networks.

bullet Internet era

The networks that form the Internet differ in size, and are categorised as the following:

  1. Tier One: typically long-distance international telecom companies who are viewed as 'backbone' networks.
  2. Tier Two: typically large regional Internet Service Providers.
  3. Tier Three: small ISPs who pay for upstream transit from Tier Two networks.

Tier one, two and three networks create 'interconnect agreements' with one another: these agreements decide how data is transferred and routed on the Internet. There are two types of interconnect agreement: 1) Transit agreement: data transfer is paid for, 2) Peering agreement: data transfer is free. Tier one networks tend to be telecom companies who own national and international fiber optic trunk lines. The data routes of the Internet are similar to the thoroughfares of a country: motorways are the primary route that cars travel upon, and likewise, tier one networks are the primary route that data travels upon the Internet. Tier one networks typically connect to the entirety of the Internet via peering agreements, and do not pay for their access (transit of data). Some recognised 'tier one' networks are: AT&T, Verizon, and Sprint.

Tier two networks are networks that have to pay for upstream transit of data (access to tier one networks). Tier two networks also use Internet Exchange Points (IXP) (peering points) to reduce their upstream transit costs and faults, and to improve latency and bandwidth. Internet Exchange Points (IXP) use the Border Gateway Protocol (BGP) to exchange data across networks. Some examples of tier two networks are: BT, Deutsche Telekom and France Telecom. Finally, we have tier three networks: these networks pay for upstream data transit from tier two networks. Tier three networks can be 'standalone' single-homed networks, or can be multihomed networks. Multihoming is where a tier three network purchases upstream access from multiple 'tier two' networks: this helps to improve the reliability of their service: in the scenario of a (SPOF) single point of failure.

Internet Service Providers and Internet Exchange Points (IXP) typically place their most important network equipment at colocation centres: these are facilities that are secure and provide protection against a multitude of natural and man-made threats. In the UK, Telehouse (London) and Telecity (Manchester), are two of the most important colocation centres in the UK and Europe. BT is classified as a tier two network, because its believed they purchase some upstream access from Sprint (overseas transit). While BT purchase some upstream access, they are a dominant 'player' in the UK, and many UK ISPs rent BT Central Pipes -- Plusnet is one example of a large UK ISP who rents BT Central Pipes. Some of BT's Central Pipes are located at Telehouse North and East (London). Alongside renting or placing equipment at colocation centres, ISPs will also use edge and gateway routers to manage and route traffic through their network.

At the bottom of the hierarchy is the end-user, also referred to as a subscriber (the 'local loop' of a telephone line is also referred to as the subscriber line): subscribers have to pay a fee to access the above networks and transmit data across them. End-users can be subscribers to either tier 1, 2 or 3 networks; most end-users will have no idea which category of network they are subscribing to. The first location that an end-user will connect to an Internet network, will usually be at the local telephone exchange -- there is roughly 5,500 local telephone exchanges in the UK. Internet Service Providers install equipment, such as a DSLAM (Digital Subscriber Line Access Multiplexer), to communicate across the 'local loop' -- telephone line between the exchange and customers premises -- with an end-users router/modem. Related: Happy Birthday, the Internet becomes 50 years old

bullet Internet History: UK and Continental Europe

The development of the protocols (TCP/IP) of the Internet was funded by the U.S. Department of Defense, therefore, a certain level of scepticism and mistrust existed within European governments about their own networks adopting TCP/IP. Europeans tended to favour networking standards they had more influence upon. This scepticism potentially has some merit: the recent Edward Snowden leaks disclosed vast court-approved NSA spying powers; VeriSign has stated it complied with "lawful orders" from U.S. courts to redirect DNS data to a government IP address.

When the North American ARPANET computer network was being developed in the early 1970s, Donald Davies, as head of the UK's NPL Network, declined an offer from Larry Roberts (ARPANET) to interconnect their networks. This was due to the European computer network initiative, the UK at the time wanted to join the European Union, and Davies had to focus on European research projects. However, in 1971, ARPANET did manage to connect to a UK node: UCL (University College of London), at a speed of 9.6 Kbps. By 1975, the UCL link was being used by a number of Ministry of Defense (UK) installations, and it was also used to test TCP/IP in the mid-1970's: sending data packets via radio from a car travelling across the Golden Gate Bridge in San Francisco to a terminal at the Royal Radar Establishment in Malvern. UCL scholars, Paal Spilling and Peter Kirstein, played an important role in the development of TCP/IP, and Kirstein is often referred to as a 'founding father of the European Internet'. Kirstein has commented that a UK government agency asked him to stop working on TCP/IP and focus on an international network standard, but he refused.

In the 1970s, European efforts to connect to ARPANET were less successful than the UK's: with French and German government agencies preferring to fund the development of a European network standard rather than relying on a U.S. dominated one. The CYCLADES packet switching computer network was developed in the early 1970s; which was a French alternative to ARPANET. CYCLADES helped to influence the development of the OSI networking model; an alternative to TCP/IP. However, by the early 1980s, some European networks were creating gateways to U.S. networks: such as the gateway between the European Academic and Research Network (EARN) and BITNET. In 1983, the CERN Networking Group planned to build a new computer network, and it eventually decided to adopt TCP/IP instead of the ISO networking standard. By the end of the 1980s, the CERN network would provide nearly 80% of Europe's Internet capacity; according to the head of its external networking group, François Flückiger.

The Réseaux Associés pour la Recherche Européenne (RARE) -- later renamed (TERENA) -- was founded in 1986 to promote European network research and OSI protocols. In 1988 and 1989, members of the CERN Networking Group were present at a meeting of the Coordinating Committee for Intercontinental Research Network (CCIRN). Vint Cerf was also present at these meetings, and he encouraged Europeans to create an organisation that would be responsible for allocating Internet IP blocks in Europe. The result was the creation of RIPE (Réseaux IP Européens). By the late 1980s, more European networks, such as the European Network (EUnet), began to switch from X.25 protocol to TPC/IP; this era is sometimes referred to as the era of the "protocol wars". In 1991, TCP/IP became the dominant networking model, when the EBONE (standing for European Backbone) consortium was established to transition European networks from OSI to IP.

In 1988, Telehouse Europe was founded, and it opened the Telehouse Docklands datacentre in London: Telehouse Docklands became the primary hub for the Internet in the United Kingdom. Telehouse Europe has expanded across London and Europe, and it presently describes itself as a "Backbone for the Global Internet Network". Internet Exchange Points (IXP), like LINX, operate out of Telehouse Docklands -- they enable Internet Service Providers to freely transit data across each others networks -- and are a key data route for the Internet in Europe.

bullet Using the Internet

The technology that can access the Internet is expanding, and the speed at which people can send data across the Internet is growing year-on-year. The most common access technology at present is broadband: digital subscriber line networks, fibre optic networks, and mobile networks. Most Internet users access the Internet through a wireless home network. A range of ethics has been outlined in RFC documents for Internet use. The amount of services on the Internet has slowly expanded since the early 1990s, but the majority are based around a client-server application: web browsers, instant chat applications, email clients, and file downloads (FTP) clients. Security and privacy has been a constant issue since the general public started accessing the Internet, threats include: unauthorised system access; viruses; malware; spyware; surveillance by web companies and governments; and WiFi vulnerabilities. This had led to the growth of VPNs to protect users when using financial services online (banking/shopping) and overcoming censorship / geographical content blocks. One of the latest developments, in relation to the use of the Internet, is the Internet of things (IoT), which refers to the current inter-connection of everyday 'dumb' items to the Internet, with the aim of improving the convenience and efficiency of everyday mundane activities. Internet content is usually found by using a web search engine, that crawls the web and the Internet for new information.