Table of Contents Introduction
TCP/IP Tutorial and Technical Overview

1.2 Internetworks

Those original designers, funded by the Defense Advanced Research Projects Agency (DARPA), of the ARPANET protocol suite introduced fundamental concepts such as layering and virtualizing in the world of networking, well before ISO even took an interest in networking.

The official organism of those researchers was the ARPANET Network Working Group, which had its last general meeting in October 1971. DARPA has continued its research for an internetworking protocol suite, from the early NCP (Network Control Program) host-to-host protocol to the TCP/IP protocol suite, which took its current form around 1978. At that time, DARPA was well known for its pioneering of packet-switching over radio networks and satellite channels. The first real implementations of the Internet were found around 1980 when DARPA started converting the machines of its research network (ARPANET) to use the new TCP/IP protocols. In 1983, the transition was completed and DARPA demanded that all computers willing to connect to its ARPANET use TCP/IP.

DARPA also contracted Bolt, Beranek, and Newman (BBN) to develop an implementation of the TCP/IP protocols for Berkeley UNIX on the VAX and funded the University of California at Berkeley to distribute that code free of charge with their UNIX operating system. The first release of the Berkeley System Distribution to include the TCP/IP protocol set was made available in 1983 (4.2BSD). From that point on, TCP/IP has been rapidly spreading among universities and research centers and has become the standard communications subsystem for all UNIX connectivity. The second release (4.3BSD) was distributed in 1986, with updates in 1988 (4.3BSD Tahoe) and 1990 (4.3BSD Reno). 4.4BSD was released in 1993. Due to funding constraints, 4.4BSD will be the last release of the BSD by the Computer Systems Research Group of the University of California at Berkeley.

As TCP/IP internetworking spread rapidly, new wide area networks were created in the US and connected to ARPANET. In turn, other networks in the rest of the world, not necessarily based on the TCP/IP protocols were added to the set of interconnected networks. The result is what is described as The Internet. Some examples of the different networks which have played key roles in this development are described in the next sections.

1.2.1 The Internet

What exactly is the Internet? First, the word internet (also internetwork) is simply a contraction of the phrase interconnected network. However when written with a capital ``I'' the Internet refers to a worldwide set of interconnected networks, so the Internet is an internet, but the reverse does not follow. The Internet is sometimes called the connected Internet.

The Internet consists of the following groups of networks (see the following sections for more information on some of these):

In many cases, particularly for commercial, military and government networks, traffic between these networks and the rest of the Internet is restricted (see also Firewalls). This leads to the question: how do I know if I am connected to the Internet? One workable approach is to ask the question: can I ping the host Ping, described in Ping, is a program used for determining the reachability of hosts on internets, implemented on every TCP/IP platform. If the answer to this question is ``no'', then you are not connected. This definition does not necessarily mean that you are completely cut off from the Internet: many systems which would fail this test have, for example, electronic mail gateways to the Internet.


Sometimes referred to as the ``grand-daddy of packet networks'', the ARPANET was built by DARPA (which was called ARPA at that time) in the late 60s to accommodate research equipment on packet-switching technology and to allow resource sharing for the Department of Defense's contractors. The network interconnected research centers, some military bases and government locations. It soon became popular with researchers for collaboration through electronic mail and other services. It was developed into a research utility run by the Defense Communications Agency (DCA) by the end of 1975 and split in 1983 into MILNET for interconnection of military sites and ARPANET for interconnection of research sites. This formed the beginning of the ``capital I'' Internet.

In 1974, the ARPANET was based on 56 Kbps leased lines which interconnected packet-switching nodes (PSN) scattered across the continental US and western Europe. These were minicomputers running a protocol known as 1822 (after the number of a report describing it) and dedicated to the packet-switching task. Each PSN had at least two connections to other PSNs (to allow alternate routing in case of circuit failure) and up to 22 ports for user computer (host) connections. These 1822 systems offered reliable, flow-controlled delivery of a packet to a destination node. This is the reason why the original NCP protocol was a rather simple protocol. It was replaced by the TCP/IP protocols, which do not assume reliability of the underlying network hardware and can be used on other-than-1822 networks. This 1822 protocol did not become an industry standard, so DARPA decided later to replace the 1822 packet switching technology with the CCITT X.25 standard.

Data traffic rapidly exceeded the capacity of the 56 Kbps lines that made up the network, which were no longer able to support the necessary throughput. Today the ARPANET has been replaced by new technologies in its role of backbone on the research side of the connected Internet (see NSFNET later in this chapter), whereas MILNET continues to form the backbone of the military side.

1.2.3 NSFNET

NSFNET, the National Science Foundation Network, is a three-level internetwork in the United States consisting of:

First Backbone
Originally established by the National Science Foundation (NSF) as a communications network for researchers and scientists to access the NSF supercomputers, the first NSFNET backbone used six DEC LSI/11 microcomputers as packet switches, interconnected by 56 Kbps leased lines. A primary interconnection between the NSFNET backbone and the ARPANET existed at Carnegie Mellon, which allowed routing of datagrams between users connected to each of those networks.
Second Backbone
The need for a new backbone appeared in 1987, when the first one became overloaded within a few months (estimated growth at that time was 100% per year). The NSF and MERIT, Inc., a computer network consortium of eight state-supported universities in Michigan, agreed to develop and manage a new, higher-speed backbone with larger transmission and switching capacities. To manage it they defined the Information Services (IS) which is comprised of an Information Center and a Technical Support Group. The Information Center is responsible for information dissemination, information resource management and electronic communication. The Technical Support Group provides support directly to the field. The purpose of this is to provide an integrated information system with easy-to-use-and-manage interfaces accessible from any point in the network supported by a full set of training services.

Merit and NSF conducted this project in partnership with IBM and MCI. IBM provided the software, packet-switching and network-management equipment, while MCI provided the long-distance transport facilities. Installed in 1988, the new network initially used 448 Kbps leased circuits to interconnect 13 nodal switching systems (NSS) supplied by IBM. Each NSS was composed of nine IBM RT systems (running an IBM version of 4.3BSD UNIX) loosely coupled via two IBM Token-Ring Networks (for redundancy). One Integrated Digital Network Exchange (IDNX) supplied by IBM was installed at each of the 13 locations, to provide:

Third Backbone
In 1989, the NSFNET backbone circuits topology was reconfigured after traffic measurements and the speed of the leased lines increased to T1 (1.544 Mbps) using primarily fiber optics.

Due to the constantly increasing need for improved packet switching and transmission capacities, three NSSs were added to the backbone and the link speed was upgraded. The migration of the NSFNET backbone from T1 to T3 (45Mbps) was completed in late 1992. Advanced Network & Services, Inc. (a company formed by IBM, MCI, and Merit, Inc.) currently provides and manages the NSFNET.

The migration to gigabit levels has already started and will continue through the late 1990s. For an overview, please see Future.

In April 1995 the US government intends to remove its funding of NFSNET. This is in part a reaction to the growing commercial use of NSFNET. For more on the commercial use of the Internet plus the US government plans refer to Commercial Use of the Internet and Information Super Highway respectively.

1.2.4 EBONE

EBONE, the Pan-European Multi-Protocol Backbone plays the same role in Europe for Internet traffic as the NSFNET backbone does in the United States. EBONE has kilobit and megabit links between five major centers.

1.2.5 CREN

Completed in October 1989, the merger of the two already well known networks CSNET (the Computer+Science Network) and BITNET (the Because It's Time Network) formed the Corporation for Research and Educational Networking. CREN draws on both of its historical CSNET and BITNET service families to provide a rich variety of network connection options:

Is the original CSNET network service and provides a store-and-forward electronic mail service using dial-up telephone lines (1200/2400 bps). It allows CSNET service members to exchange messages with other CREN members and with other major mail networks, including NSFNET, MILNET, etc.
Is a CSNET full-service Internet-connected network that uses TCP/IP protocols on top of X.25. It is the common method for international members to connect to CSNET, since they can use their local X.25 Public Data Network to reach Telenet in the US. It provides file transfer, remote login, and immediate electronic mail service between X.25Net hosts.
Dial-up IP
Is an implementation of SLIP (Serial Line IP) that allows sites using the switched telephone network (9600 bps) to send IP packets, by means of a central server, to the Internet. Users of this method thus have the same services as X.25Net sites.
Leased Line IP
Used by many CREN members to connect to CREN. A variety of link speeds are supported up to T1 rates.
Traditionally runs over leased lines at 9600 bps and provides interactive messages, unsolicited file transfer and electronic mail.
RSCS over IP
Allows BITNET service hubs to relax the dedicated RSCS bisync line in preference to an Internet IP route, if one exists.

1.2.6 Cypress

Cypress is a leased-line network that provides a low-cost, protocol-independent, packet switching system used primarily for interconnecting small sites to the TCP/IP Internet. Originally established as part of a joint research project with CSNET, it is now independent of CSNET.

There are no restrictions on its use, outside those imposed by other networks. Thus commercial traffic can pass between two industrial sites across Cypress. Industrial sites cannot pass commercial traffic onto the Internet due to restrictions imposed by government agencies that control the backbone networks (for example NSFNET).

1.2.7 DRI

The Terrestrial Wideband Network is a wide area network whose purpose is to provide a platform for research on high-speed networking protocols and applications (this role was formerly played by the ARPANET). This system includes both connection-oriented as well as connectionless service, broadcast and multicast service and real-time conferencing.

The Terrestrial Wideband Network was built and deployed by BBN Systems and Technologies Corporation during the first half of 1989 as part of the initial phase of the Defense Research Internet (DRI). Its main purpose was to carry cross-country Internet datagram traffic associated with DARPA-funded projects. It was composed of Internet gateways and Terrestrial Wideband Network packet switches (WPS) which communicated with each other using the Host Access Protocol (HAP) specified in RFC 1221. Wideband Monitoring Protocol (WB-MON) was used between the WPSs and the monitoring center. The backbone also supported a research environment for multimedia conferencing and voice/video conferencing using gateways that used a real-time connection-oriented protocol (ST-II - Stream Protocol - RFC 1190) over a connectionless network.

1.2.8 European Academic Research Network (EARN)

EARN, started in 1983, was the first and largest network serving academic and research institutions in Europe, the Middle East and Africa. EARN was initially started with help from IBM. It evolved to become a non-profit non-commercial traffic-based network serving academic and research institutions.

1.2.9 Réseaux Associés pour la Recherche Européenne (RARE)

RARE, founded in 1986, is the association of European networking organizations and their users. The association has 20 voting Full National Members (all of which are European countries), several Associate National Members (some European and Asian countries), International Members (for example EARN) and Liason Members (for example CREN).

It supports the principles of open systems as defined by the ISO as well as a number of mainly European groups, such as the European Workshop for Open Systems (EWOS) and the European Telecommunications Standards Institute (ETSI).

For more details, please refer to Réseaux Associés pour la Recherche Européenne (RARE).

1.2.10 Réseaux IP Européens (RIPE)

Réseaux IP Européens (RIPE) coordinates TCP/IP networking for the research community in Europe. It operates under the auspices of RARE. RIPE has been functioning since 1989. By the early 1990s more than 60 organizations were participating in the work. The objective of RIPE is to ensure the necessary administrative and technical coordination to allow the operation of a pan-European IP network. Note that RIPE does not operate a network of its own. RIPE is the IP activity of RARE.

One of the activities of RIPE is to maintain a database of European IP networks, DNS domains and their contact persons. The content of this database is considered to be public information. The database can be accessed either via a WHOIS server on host (TCP port 43) or via anonymous FTP to

The RIPE NCC (Network Coordination Center) can be contacted via:
Kruislaan 409
NL-1098 SJ Amsterdam
The Netherlands
Phone: +31 20 592 5065
Fax: +31 20 592 5155

1.2.11 Japanese Internet

Japan has many different networks. The following are some major ones.

For more details, please refer to [LaQuey] and [Malamud] listed in Bibliography.

1.2.12 Commercial Use of the Internet

In recent years the Internet has grown in size and range at a greater rate than anyone could have predicted. In particular, more than half of the hosts now connected to the Internet are of a commercial nature. This is an area of potential and actual conflict with the initial aims of the Internet, which were to foster open communications between academic and research institutions. However, the continued growth in commercial use of the Internet is inevitable so it will be helpful to explain how this evolution is taking place.

One important initiative to consider is that of the Acceptable Use Policy (AUP). The first of these policies was introduced in 1992 and applies to the use of NSFNET. A copy of this can be obtained at At the heart of this AUP is a commitment "to support open research and education". Under "Unacceptable Uses" is a prohibition of "use for for-profit activities", unless covered by the General Principle or as a specifically acceptable use. However, in spite of this apparently restrictive stance the NSFNET has increasingly been used for a broad range of activities, including many of a commercial nature.

Apart from the NSFNET AUP, many of the internets that connect to NSFNET maintain their own AUPs. Some of these are relatively restrictive in their treatment of commercial activities while others are relatively liberal. The main thing to say is that AUPs will need to evolve as the inevitable growth in commercial use continues.

Let us now focus on the Internet service providers who have been most active in introducing commercial uses to the Internet. Two worth mentioning are PSINet and UUNET, which began in the late 80s to offer Internet access to both businesses and individuals. The California-based CERFnet provides services that are free of any AUP. An organization to interconnect PSINet, UUNET and CERFnet was formed soon after, called the Commercial Internet Exchange (CIX). To date CIX has more than 20 members connecting member internets in an AUP-free environment. At about the same time that CIX was conceived, a non-profit company, Advance Network and Services (ANS), was formed by IBM, MCI and Merit, Inc. to operate T1 backbone connections for NSFNET. This group has been active in increasing the commercial presence on the Internet.

ANS also formed a commercially oriented subsidiary called ANS CO+RE to provide linkage between commercial customers and the research and education domains. ANS CO+RE provides AUP-free access to NSFNET as well as being linked to CIX.

1.2.13 Information Super Highway

One recent and important initiative has been the creation of the US Advisory Council on the National Information Infrastructure headed by Al Gore. In essence, this initiative makes the creation of a "network of networks" a national priority. This network would be similar to the existing Internet in some respects but with government and industry contributing those elements which it is best able to provide.

From a more international perspective The Group of Seven (G7) ministers met in Brussels in February 1995 to discuss the emerging Global Information Infrastructure (GII). The conference was attended by science, technology and economic ministers of Canada, the United Kingdom, France, Japan, Germany, Italy and the United States, and focused on technological, cultural and economic issues regarding the development of an international infrastructure.

A free electronic magazine called G7 Live was used to deliver daily wrap-ups, commentary and news about the conference to Internet users worldwide. Specific issues covered by G7 Live include intellectual property rights, infrastructure building, cultural and regulatory considerations and descriptions of the more than 100 technology exhibits that were present at the conference.

Both the NII and the GII described above are important initiatives which should ultimately lead to the "Information Super Highway" that is presently the subject of so much discussion in the media.

Table of Contents IBM and the Internet