Networks exist in principle since the 60s. And on the communication level, the American Miltar took a top spot with its equipment on computers worldwide, but it also knew the Achilles heel of previous network technologies just the network nodes. If only one such node failed, that meant the stoppage of the entire network. And destroyed communication is seen in military circles as the strategic defeat par excellence. It was clear that for the future, technology was needed that maintained functionality even if several elements failed. The research department Advanced Research Projects Agency ARPA (1957), founded by the US Department of Defense, then supported research into resistant network structures and protocols.
As early as 1962, RAND Corporation published the first proposal for a packet-switched network, which guaranteed data transmission in the event of the failure of part of the switching computers. The main findings of this theory were the abandonment of central decision-making and the breakdown of messages into parts of constant packet size. These features also specify the operation of today's "Internet".
At that time, incompatibilities between the computers provided in the network already existed due to different hardware. Ergo one needed a uniform data format. Further research resulted in the introduction of an Interface Message Processor (IMP) that could communicate with any IMP attached to each host. The format conversions had to be done in the protocol between the computer and the IMP. The interface message processor could thus be described as the predecessor of the network card.
Once again, ARPA was in charge and initiated a bid to develop interface message processors and a network that could connect up to 16 sites. As a pilot project in late 1969, a network was built between the Universities of Los Angeles, Santa Barbara, Stanford and Utah. This reference network is considered the beginning of the ARPAnet.
Surprisingly, the fame of the first operational network on a packet-switching basis is due to the British, who already put such into operation in Great Britain in 1968.
Probably due to the different hardware platforms and the associated effort in the implementation of the protocols for each IMP computer combination, the desired target with 23 computers, which were connected to 15 nodes, reached only in 1971. During the same period, the first proposals for the protocols TELNET and FTP fell.
Two years later, in 1973, the first two overseas connections were added to the now 33 nodes within the United States (United Kingdom, Norway). For the first time, the problem of incompatible protocols was faced, as only in the ARPAnet with the Network Control Protocol (NCP), a uniform policy existed, the now added networks but their own protocols used.
In 1979, two students linked two Unix machines together over a telephone connection. For data transmission they used Unix to Unix Copy (UUCP). The network worked by transferring a message from a computer to a server which stored it and at the next opportunity (telephone connection) passed it on to the next server until all computers had a copy of the message. Another innovation was the hierarchical arrangement of the messages according to thematic aspects. This network has survived to this day as USENET.
The same year saw the laying of the foundation stone for the TCP / IP-based computer Scienes Network (CSNET), to which especially the non-ARPAnet integrated universities of the USA joined together. However, it took two years until the practical implementation, while the cheaper USENET within a year already counted 15 computers.
After the adoption of the standards for TCP / IP in 1982, the conversion of the entire ARPAnets. The NCP had served it out. In Europe, however, the EUnet was installed, which worked mainly on the basis of the USENET.
Also in 1982, the standard for an email protocol was passed.
In 1983, the ARPAnet was renamed the ARPA Internet and all US military facilities were integrated into a new network (MILNET).
Including the TCP / IP code in the Berkeley UNIX implementation (version 4.2) has greatly enhanced the protocol stack. Also MSDOS computers got with the FidoNet access to the developing Internet.
In 1984, the limit of 1000 networked computers was finally exceeded. The problem of addressing new computers led to the introduction of Domain Name Service (DNS). Until then, each integrated computer had to locally save a file that contained all the addresses of the accessible computers in the network. The size of the "database" has long been difficult to handle. As the network spreads, the question of the timeliness of such a file was added.
The New Network (NSFNET) of the National Science Foundation (NSF) in 1986 already joined together five US supercomputer sites. The transmission rate of the data was at least 56 kBit / s. NSF enabled outside institutions to connect their networks to NFSNET. Since the offer was very interesting, the performance of the connections was increased to 1.5 Megabit / s in order to meet the increased data volume.
At about the same time, the Network News Transfer Protocol (NNTP), which could transport the USENET data over TCP / IP connections, also introduced Mail Exchanger to allow machines to exchange messages that have no (permanent) access to the Internet had. The first European countries were integrated into the NFSNET in 1988.
Eventually, about 60,000 computers participated in the worldwide network in 1988, when the first "Internet worm" blocked numerous computers and for the first time clarified the problem of network security.
And in 1989 and others Germany also the connection to the NFSNET.