TCP/IP
With so many different network
methods, something was needed to unify them. Robert E. Kahn
of DARPA
and ARPANET
recruited Vinton Cerf of Stanford University to work with him on the problem. By 1973, they had worked
out a fundamental reformulation, where the differences between network
protocols were hidden by using a common internetwork protocol, and instead of the network being responsible for
reliability, as in the ARPANET, the hosts became responsible. Cerf credits Hubert Zimmerman, Gerard LeLann and Louis Pouzin
(designer of the CYCLADES network) with important work on this design.[25]
The specification of the resulting
protocol, RFC 675
– Specification of Internet Transmission Control Program, by Vinton Cerf, Yogen Dalal and Carl Sunshine, Network
Working Group, December 1974, contains the first attested use of the term internet,
as a shorthand for internetworking; later RFCs repeat this use, so the
word started out as an adjective rather than the noun it is today.
A Stanford Research Institute packet radio van, site of the first three-way internetworked
transmission.
With the role of the network reduced
to the bare minimum, it became possible to join almost any networks together,
no matter what their characteristics were, thereby solving Kahn's initial
problem. DARPA agreed to fund development of prototype software, and after
several years of work, the first demonstration of a gateway between the Packet Radio
network in the SF Bay area and the ARPANET was conducted by the Stanford Research Institute. On November 22, 1977 a three network demonstration was
conducted including the ARPANET, the Packet Radio Network and the Atlantic
Packet Satellite network.[26][27]
Stemming from the first
specifications of TCP in 1974, TCP/IP
emerged in mid-late 1978 in nearly final form. By 1981, the associated
standards were published as RFCs 791, 792 and 793 and adopted for use. DARPA sponsored or
encouraged the development of TCP/IP implementations for many operating systems
and then scheduled a migration of all hosts on all of its packet networks to
TCP/IP. On January 1, 1983, known as flag day, TCP/IP protocols became the only approved protocol on the
ARPANET, replacing the earlier NCP protocol.[28]
ARPANET
to the federal wide area networks: MILNET, NSI, ESNet, CSNET, and NSFNET
After the ARPANET had been up and
running for several years, ARPA looked for another agency to hand off the
network to; ARPA's primary mission was funding cutting edge research and
development, not running a communications utility. Eventually, in July 1975,
the network had been turned over to the Defense
Communications Agency, also part of the Department of Defense.
In 1983, the U.S. military portion of the ARPANET was broken off as a separate
network, the MILNET.
MILNET subsequently became the unclassified but military-only NIPRNET,
in parallel with the SECRET-level SIPRNET
and JWICS
for TOP SECRET and above. NIPRNET does have controlled security gateways to the
public Internet.
The networks based on the ARPANET
were government funded and therefore restricted to noncommercial uses such as
research; unrelated commercial use was strictly forbidden. This initially
restricted connections to military sites and universities. During the 1980s,
the connections expanded to more educational institutions, and even to a
growing number of companies such as Digital
Equipment Corporation and Hewlett-Packard,
which were participating in research projects or providing services to those
who were.
Several other branches of the U.S. government,
the National Aeronautics and Space Administration (NASA), the National
Science Foundation (NSF), and the Department of Energy
(DOE) became heavily involved in Internet research and started development of a
successor to ARPANET. In the mid 1980s, all three of these branches developed
the first Wide Area Networks based on TCP/IP. NASA developed the NASA Science Network,
NSF developed CSNET
and DOE evolved the Energy Sciences Network or ESNet.
T3 NSFNET Backbone, c. 1992
NASA developed the TCP/IP based NASA
Science Network (NSN) in the mid 1980s, connecting space scientists to data and
information stored anywhere in the world. In 1989, the DECnet-based Space Physics Analysis
Network (SPAN) and the TCP/IP-based NASA Science Network (NSN) were brought
together at NASA Ames Research Center creating the first multiprotocol wide
area network called the NASA Science Internet, or NSI. NSI was established to
provide a totally integrated communications infrastructure to the NASA
scientific community for the advancement of earth, space and life sciences. As
a high-speed, multiprotocol, international network, NSI provided connectivity
to over 20,000 scientists across all seven continents.
In 1981 NSF supported the
development of the Computer Science Network (CSNET). CSNET connected with ARPANET using TCP/IP, and ran
TCP/IP over X.25,
but it also supported departments without sophisticated network connections,
using automated dial-up mail exchange.
Its experience with CSNET lead NSF
to use TCP/IP when it created NSFNET, a 56 kbit/s backbone
established in 1986, to supported the NSF sponsored supercomputing
centers. The NSFNET Project also provided support for the creation of regional
research and education networks in the United States and for the connection of
university and college campus networks to the regional networks.[29]
The use of NSFNET and the regional networks was not limited to supercomputer
users and the 56 kbit/s network quickly became overloaded. NSFNET was
upgraded to 1.5 Mbit/s in 1988 under a cooperative agreement with the Merit Network
in partnership with IBM, MCI, and the State of Michigan. The existence of NSFNET and the creation of Federal
Internet Exchanges (FIXes) allowed the ARPANET to be
decommissioned in 1990. NSFNET was expanded and upgraded to 45 Mbit/s in
1991, and was decommissioned in 1995 when it was replaced by backbones operated
by several commercial Internet
Service Providers.
Transition
towards the Internet
The term "internet" was
adopted in the first RFC published on the TCP protocol (RFC 675:[30]
Internet Transmission Control Program, December 1974) as an abbreviation of the
term internetworking and the two terms were used interchangeably. In
general, an internet was any network using TCP/IP. It was around the
time when ARPANET was interlinked with NSFNET in the late 1980s, that the term
was used as the name of the network, Internet,[31]
being the large and global TCP/IP network.
As interest in widespread networking
grew and new applications for it were developed, the Internet's technologies
spread throughout the rest of the world. The network-agnostic approach in
TCP/IP meant that it was easy to use any existing network infrastructure, such
as the IPSS X.25 network, to carry Internet
traffic. In 1984, University College London replaced its transatlantic
satellite links with TCP/IP over IPSS.[32]
Many sites unable to link directly
to the Internet started to create simple gateways to allow transfer of email,
at that time the most important application. Sites which only had intermittent
connections used UUCP
or FidoNet
and relied on the gateways between these networks and the Internet. Some
gateway services went beyond simple email peering, such as allowing access to FTP sites via UUCP or email.[33]
Finally, the Internet's remaining
centralized routing aspects were removed. The EGP routing protocol was replaced by a new protocol, the Border Gateway Protocol (BGP). This turned the Internet into a meshed topology and
moved away from the centric architecture which ARPANET had emphasized. In 1994,
Classless Inter-Domain Routing
was introduced to support better conservation of address space which allowed
use of route aggregation to decrease the size of routing tables.[34]
