Other articles in the series:
- History of the relay
- History of electronic computers
- History of the transistor
- Internet history
- Era of Fragmentation
In the 1968 work "The Computer as a Communication Device", written during the development of the ARPANET, и declared that the association of computers would not be limited to the creation of separate networks. They predicted that such networks would coalesce into a "non-permanent network of networks" that would bring together "various information processing and storage equipment" into an interconnected whole. In less than a decade, such initially theoretical considerations attracted immediate practical interest. By the mid-1970s, computer networks began to spread rapidly.
Distribution of networks
They penetrated various media, institutions and places. ALOHAnet was one of several new academic networks that received ARPA funding in the early 1970s. Others included PRNET, which combined trucks with packet radio, and satellite SATNET. Other countries have developed their own research networks along similar lines, notably Britain and France. Local networks, thanks to their smaller scale and lower cost, multiplied even faster. In addition to Ethernet from Xerox PARC, one could find Octopus at the Lawrence Radiation Laboratory in Berkeley, California; Ring at the University of Cambridge; Mark II at the British National Physical Laboratory.
Around the same time, commercial enterprises began offering paid access to private packet networks. This opened up a new national market for online computing services. In the 1960s, various companies launched businesses that offered access to specialized databases (legal and financial), or time-sharing computers, to anyone with their own terminal. However, access to them across the country via the regular telephone network was disproportionately expensive, making it difficult for these networks to expand beyond local markets. Several larger firms (Tymshare, for example) built their own intranets, but commercial packet networks have reduced the cost of running them to reasonable levels.
The first such network appeared due to the departure of ARPANET experts. In 1972, several employees left the Bolt, Beranek, and Newman (BBN) company, which was responsible for the creation and operation of the ARPANET, to form Packet Communications, Inc. Although the company ultimately failed, the sudden shock served as a catalyst for BBN to create its own private network, Telenet. With Larry Roberts, the ARPANET architect at the helm, Telenet ran successfully for five years before it was bought by GTE.
Given the emergence of such diverse networks, how could Licklider and Taylor have foreseen the emergence of a single unified system? Even if it were organizationally possible to simply integrate all these systems with the ARPANET - which could not be done - the incompatibility of their protocols made this impossible. And yet, in the end, all these heterogeneous networks (and their descendants) really connected with each other into a universal communication system, which we know as the Internet. It didn’t start with any grant or global plan, but with a forgotten research project that was being worked on by a middle manager from ARPA .
Bob Kahn problem
Kahn completed his Ph.D. in electronic signal processing at Princeton in 1964 while playing golf on the courses near the school. After a brief stint as a professor at MIT, he took a job at BBN, initially with the desire to take a vacation to immerse himself in the industry to see how practical people decided which problems were worthy of research. Coincidentally, his work at BBN was related to the study of the possible behavior of computer networks - shortly after that, BBN received an order for ARPANET. Kana pulled this project, and he gave the bulk of the developments regarding the architecture of the network.
Photo of Kahn from a 1974 newspaper
His "short break" turned into a six-year job where Kahn was a networking expert at BBN while bringing the ARPANET to full working order. By 1972, he was tired of the topic, and more importantly, tired of fighting the constant politicking and fighting with the heads of BBN divisions. So he accepted an offer from Larry Roberts (before Roberts himself left to create Telenet) and became a program manager at ARPA to lead the development of automated manufacturing technology, with the potential to manage millions of dollars of investment. He abandoned work on the ARPANET and decided to start from scratch in a new area.
But within months of his arrival in Washington, DC, Congress hacked the automatic manufacturing project. Kahn wanted to immediately pack up and return to Cambridge, but Roberts persuaded him to stay and help develop new networking projects for ARPA. Kahn, unable to break free of his own knowledge, found himself the manager of PRNET, a packet radio network that was supposed to provide military operations with the benefits of packet-switched networks.
The PRNET project, launched under the auspices of the Stanford Research Institute (SRI), was to extend ALOHANET's basic packet-transmitting technical core to support repeaters and multi-station operation, including moving vans. However, it was immediately clear to Kahn that there would be no benefit from such a network, since it was a computer network with virtually no computers. When it went live in 1975, it had one SRI computer and four repeaters located along the San Francisco Bay. Mobile field stations could not reasonably handle the size and power consumption of 1970s mainframes. All significant computing resources were within the ARPANET, which used a completely different set of protocols, and was unable to interpret the message received from the PRNET. He wondered how it would be possible to connect this nascent network with its much more mature cousin?
Kahn turned to an old acquaintance from the early ARPANET days to help him with an answer. became interested in computers while a mathematics student at Stanford and decided to return to graduate school in computer science at the University of California, Los Angeles (UCLA) after working for several years in the IBM office. He arrived in 1967 and, along with his high school friend Steve Crocker, joined the Len Kleinrock Network Measurement Center, which was the ARPANET division at UCLA. There, he and Crocker became experts in protocol development, and were key members of the networking working group, which developed both the basic network control program (NCP) for sending messages over the ARPANET, and high-level file transfer and remote login protocols.
Cerf's photo from a 1974 newspaper
Cerf met Kahn in the early 1970s when the latter arrived at UCLA from BBN to test the network under load. He created congestion in the network using software created by Cerf, which generated artificial traffic. As Kahn expected, the network failed and he recommended changes to improve congestion management. In the years that followed, Cerf continued what appeared to be a promising academic career. Around the same time Kahn left BBN for Washington, Cerf traveled across the coast to take an adjunct professorship at Stanford.
Kahn knew a lot about computer networks, but had no experience in developing protocols - he was in signal processing, not computer science. He knew Cerf would be the perfect complement to his skillset, and it would be critical in any attempt to link ARPANET to PRNET. Kahn contacted him about internetworking, and they met several times in 1973 before burrowing into a hotel in Palo Alto to publish their seminal work, "A Protocol for Packet Interconnection", published in May 1974 in IEEE Transactions on Communications. . There, the draft "Transmission Control Program" (TCP) (soon to become "protocol") was presented, the cornerstone for the software of the modern Internet.
External influence
There is no couple or moment more connected to the invention of the internet than Cerf and Kahn and their 1974 work. Yet the creation of the Internet was not an event that happened at a certain point in time - it was a process that unfolded over many years of development. The original protocol described by Cerf and Kahn in a 1974 paper has been revised and tweaked numerous times over the years. The first communication between networks was tested only in 1977; the protocol was divided into two layers - the ubiquitous today TCP and IP - only in 1978; ARPANET only began using it for its own purposes in 1982 (this timeline for the emergence of the Internet can be extended to 1995, when the US government removed the firewall between publicly funded academic Internet and commercial Internet). The list of participants in this process of invention has expanded far beyond these two names. In the early years, an organization called the International Packet Network Working Group (INWG) served as the main body for collaborative work.
ARPANET entered the wider technical world in October 1972 at the first international computer communications conference, held at the Washington Hilton with its modernist curves. In addition to Americans like Cerf and Kahn, there were several prominent network experts from Europe, in particular from France and Donald Davies from Britain. At the instigation of Larry Roberts, they decided to form an international working group to discuss packet switching systems and protocols, much like the networking working group that established protocols for the ARPANET. Cerf, a recent professor at Stanford, agreed to be chairman. One of their first topics was the problem of interconnection.
Among the important early contributors to this discussion was Robert Metcalfe, whom we have already met as the Ethernet architect at Xerox PARC. Although Metcalfe could not tell his colleagues this, by the time Cerf and Kahn's work was published, he had long been developing his own Internet protocol, the PARC Universal Packet, or PUP.
Internet demand at Xerox increased as Alto's Ethernet network became successful. PARC had another local area network of Data General Nova minicomputers, and of course there was also the ARPANET. The PARC leaders looked into the future and guessed that each Xerox base had to have its own Ethernet, and that they would somehow have to be connected to each other (perhaps through Xerox's own internal ARPANET equivalent). In order to be able to pretend to be a normal message, the PUP packet was stored inside other packets on whatever network it traveled on - say, PARC Ethernet. When a packet reached a gateway computer between Ethernet and another network (such as the ARPANET), that computer unwrapped the PUP packet, read its address, and rewrapped it into an ARPANET packet with the appropriate headers, sending it to the address.
While Metcalfe couldn't speak directly about what Xerox did, his hands-on experience inevitably seeped into INWG discussions. Evidence of his influence is seen in the fact that in the 1974 work, Cerf and Kahn acknowledge his contribution, and Metcalfe later takes a little offense at not insisting on co-authorship. PUP most likely influenced the design of the modern internet again in the 1970s when pushed through the decision to separate the protocol into TCP and IP so as not to process the complex TCP protocol on gateways between networks. IP (Internet Protocol) was a simplified version of the address protocol, without any of the complex TCP logic to ensure every bit was delivered. The Xerox network protocol - then known as Xerox Network Systems (XNS) - had already come to a similar division.
Another source of influence on the early Internet protocols came from Europe, specifically in the network developed in the early 1970s as a result of Plan Calcul, a program launched by to grow their own French computing industry. De Gaulle has long been concerned about the rise of US political, commercial, financial and cultural dominance in Western Europe. He decided to once again make France an independent world leader, and not a pawn in the Cold War between the US and the USSR. In relation to the computer industry, two particularly strong threats to this independence emerged in the 1960s. First, the United States refused to issue licenses for the export of its most powerful computers, which France wanted to use in the development of its own atomic bombs. Secondly, the American company General Electric became the main owner of the only French computer manufacturer Compagnie des Machines Bull - and shortly thereafter closed several of Bull's main product lines (the company was founded in 1919 by a Norwegian named Bull, for the production of machines that work with punched cards - directly like IBM, she moved to France in the 1930s after the death of the founder). Thus was born Plan Calcul, designed to ensure France's ability to provide its own computing power.
To oversee the implementation of Plan Calcul, de Gaulle created a délégation à l'informatique (something like a "delegation for computer science") that reported directly to his prime minister. In early 1971, this delegation placed engineer Louis Pouzin in charge of building the French version of the ARPANET. The Delegation considered that packet networks would play a critical role in the computing of the coming years, and therefore technical knowledge in that area itself would be necessary if Plan Calcul was to be successful.
Pouzin at a conference in 1976
Pouzin, a graduate of the École Polytechnique de Paris, France's premier engineering school, worked as a young man for a French telephone equipment manufacturer before moving to Bull. There, he convinced employers that they needed to know more about cutting-edge US developments. So while at Bull, he spent two and a half years, from 1963 to 1965, helping build the Compatible Time-Sharing System (CTSS) at MIT. This experience made him the foremost expert on interactive time-sharing computing in all of France—and probably all of Europe.
Cyclades Network Architecture
Pouzin named the network he was asked to create Cyclades, after the Cyclades group of Greek islands in the Aegean. As the name suggests, each computer on this network was, by and large, its own island. The main contribution of Cyclades to network technologies was the concept - the simplest version of packet communication. The idea consisted of two complementary parts:
- Datagrams are independent: unlike the data in a phone call or ARPANET message, each datagram can be processed independently. It does not rely on previous messages, nor on their order, nor on the protocol for establishing a connection (such as dialing a telephone number).
- Datagrams are transmitted from host to host - the entire responsibility for reliably sending a message to an address lies with the sender and recipient, and not with the network, which in this case is just a "pipe".
The concept of a datagram seemed like heresy to Pouzin's colleagues at the French Postal, Telephone and Telegraph (PTT) organization, which in the 1970s built its own network around phone-like connections and terminal-to-computer (rather than computer-to-computer) connections. This took place under the supervision of another graduate of the Polytechnic School, Remy Despres. The idea of abandoning the reliability of transmissions within the network was repulsive to PTT, since decades of experience led it to make the telephone and telegraph as reliable as possible. At the same time, from an economic and political point of view, the transfer of control over all applications and services to host computers located at the edge of the network threatened to turn PTT into something far from unique and replaceable. However, nothing strengthens an opinion than a firm opposition to it, so the concept from PTT only helped Pouzin to verify the validity of his datagram - an approach to creating protocols that work to communicate one host to another.
Pouzin and his colleagues at the Cyclades project actively participated in the INWG and various conferences where the ideas behind TCP were discussed and were not shy about expressing their opinions on how the network or networks should work. Like Melkaf, Pouzin and his colleague Hubert Zimmermann deserve credit in the TCP work of 1974, and at least one other colleague, engineer Gerard le Lan, also helped Cerf polish the protocols. Cerf later recalled that " The sliding window method for TCP was taken directly from a discussion of this issue with Pouzin and his people ... I remember Bob Metcalfe, le Lan and I lying on a huge piece of paper on the floor of my living room in Palo Alto, trying to sketch state diagrams for these protocols " .
"Sliding window" refers to the way TCP manages the flow of data between a sender and a receiver. The current window consists of all packets in the outgoing data stream that the sender can actively send. The right edge of the window shifts to the right when the receiver reports that buffer space has been freed, and the left edge shifts to the right when the receiver reports receiving previous packets.
The concept of the diagram perfectly matched the behavior of broadcast networks such as Ethernet and ALOHANET, willy-nilly sending their messages into a noisy and indifferent air (in contrast to the more phone-like ARPANET, which required serial delivery of messages between IMPs over a reliable line from AT&T for normal operation). It made sense to tailor intranet protocols to the least reliable networks rather than their more complex cousins, and that's exactly what Kahn and Cerf's TCP did.
I could go on and on about the British role in developing the early stages of internetworking, but don't go into too much detail lest you miss the point - the two names most closely associated with the invention of the internet weren't the only ones that mattered.
TCP conquers all
What happened to these early ideas of intercontinental cooperation? Why are Cerf and Kan glorified everywhere as the fathers of the Internet, but nothing is heard about Pouzin and Zimmerman? To understand this, it is first necessary to delve into the procedural details of the INWG's early years.
Following the spirit of the ARPA Network Working Group and its "Requests for Comment" (RFC), the INWG created its own "general notes" system. As part of this practice, after about a year of working together, Kahn and Cerf submitted a draft TCP to the INWG as Note #39 in September 1973. This was essentially the same paper they published in IEEE Transactions the following spring. In April 1974, the Cyclades team, led by Hubert Zimmermann and Michel Ely, published a counter proposal, INWG 61. The difference was different views on various engineering trade-offs, mainly how packets were divided and reassembled across networks where the packet size was smaller. .
The split was minimal, but the need to somehow negotiate took on unexpected urgency due to plans to review network standards announced by the Comité Consultatif International Téléphonique et Télégraphique () [International Telephony and Telegraphy Advisory Committee]. CCITT, division , dealing with standardization, worked on a four-year cycle of plenary meetings. Proposals to be considered at the 1976 meeting had to be submitted before the fall of 1975, and no changes could be made between that date and 1980. Feverish meetings within the INWG led to a final vote in which a new protocol won, described by representatives of the most important organizations in the world for computer networks - Cerf of the ARPANET, Zimmerman of Cyclades, Roger Scantlebury of the British National Physical Laboratory and Alex McKenzie of BBN. The new proposal, INWG 96, fell somewhere between 39 and 61, and seemed to set the direction for interconnection for the foreseeable future.
But in fact, the compromise served as the last gasp of international cooperation in the field of interconnection, and this fact was preceded by the ominous absence of Bob Kahn from the INWG vote on the new proposal. It turned out that the result of the vote did not meet the deadlines set by the CCITT, and in addition, Cerf made the situation even worse by sending a letter to the CCITT, where he described that this proposal lacked full consensus in the INWG. But any suggestion from the INWG would still certainly not be accepted, because the telecom executives who dominated CCITT were not interested in the datagram-enabled networks invented by computer researchers. They wanted complete control over the traffic on the network, not delegating that power to local computers that they had no control over. They generally ignored the issue of interconnection, and agreed to adopt a virtual connection protocol for a single network, called .
The irony is that the X.25 protocol was backed by Kahn's former boss, Larry Roberts. Once a leader in cutting-edge networking research, his new interests as a business leader brought him to CCITT to authorize the protocols his company, Telenet, was already using.
The Europeans, mostly led by Zimmermann, made another attempt by turning to another standards organization where the dominance of the telecom leadership was not so strong - the International Organization for Standardization, . The resulting open systems communication standard () had some advantages over TCP/IP. For example, it did not have the same limited hierarchical addressing system as IP, the limitations of which required a few cheap hacks to deal with the explosive growth of the Internet in the 1990s (in the 2010s, networks finally begin to transition to IP protocol, which fixes problems with address space limitation). However, this process, for many reasons, dragged on and on indefinitely, without leading to the creation of working software. In particular, ISO procedures, well suited to the approval of established technical practices, were not well suited to emerging technologies. And when the Internet based on TCP / IP began to develop in the 1990s, OSI fell out of favor.
Let's move on from the battle for standards to the mundane, practical stuff of networking in the field. The Europeans in good faith undertook the implementation of INWG 96 to unite Cyclades and the National Physical Laboratory as part of the creation of a European information network. But Kahn and other leaders of the ARPA internet project weren't about to derail the TCP train for international cooperation. Kahn had already committed money to implement TCP on the ARPANET and PRNET, and didn't want to start all over again. Cerf tried to promote US support for the compromise he had worked out for the INWG, but finally gave in. He also decided to let go of the stresses of being an adjunct professor and, following Kahn's example, became a program manager at ARPA, stepping down from active participation in the INWG.
Why, then, did the desire of the Europeans to establish a united front and an official international standard succeed so little? Basically, it's all about the different positions of the heads of American and European telecoms. The Europeans had to contend with constant pressure on the datagram model from their Post and Telecom (PTT) executives, who operated as administrative departments of their respective national governments. Because of this, they were more motivated to find consensus in formal standards-setting processes. The rapid fall of Cyclades, which lost political interest in 1975 and all funding in 1978, provides material for studying the power of the PTT. Pouzin blamed the administration for her death . d'Estaing came to power in 1974 and recruited a government from representatives of the National School of Administration () despised by Pouzin: if Polytechnic can be compared to MIT, ENA can be likened to Harvard business school. The d'Estaing administration was building information technology policy around the idea of "national champions," and such a computer network needed the support of PTT. The Cyclades project would never have received such support; instead, Pouzin's rival Despres led the creation of an X.25-based virtual connection network called Transpac.
In the US, things were different. AT&T did not have the same political influence as its counterparts abroad, it was not part of the US administration. On the contrary, the government just at that time severely limited and weakened the company, it was forbidden to interfere in the development of computer networks and services, and soon it was completely dismantled. ARPA was free to develop its Internet program under the protective umbrella of the powerful Department of Defense, without any political pressure whatsoever. It funded the implementation of TCP on various computers, and used its influence to force all hosts on the ARPANET to switch to the new protocol in 1983. Therefore, the most powerful computer network in the world, many of whose nodes were the most powerful computing organizations in the world, became the site of the development of TCP. /ip.
Thus, TCP/IP became the cornerstone of the Internet, and not just the Internet, due to the relative political and financial freedom of ARPA compared to any other computer networking organization. OSI notwithstanding, ARPA has become the tail-wagging dog of the network research community. From a 1974 perspective, one could see that there were many lines of influence leading to Cerf and Kahn's work on TCP, and many potential international collaborations that could emerge from them. However, from the point of view of 1995, all roads lead to a single key moment, a single American organization and two famous names.
What else to read
- Janet Abbate, Inventing the Internet (1999)
- John Day, “The Clamor Outside as INWG Debated,” IEEE Annals of the History of Computing (2016)
- Andrew L. Russell, Open Standards and the Digital Age (2014)
- Andrew L. Russell and Valérie Schafer, “In the Shadow of ARPANET and Internet: Louis Pouzin and the Cyclades Network in the 1970s,” Technology and Culture (2014)
Source: habr.com