The CYCLADES packet switching network (French pronunciation: [siklad]) was a French research network created in the early 1970s. It was developed to explore alternatives to the ARPANET design and to support network research generally. The experiences with the network had influences on the design of Internet protocols.
The CYCLADES network was the first to make the hosts responsible for the reliable delivery of data, rather than the network itself, using unreliable datagrams and associated end-to-end protocol mechanisms. These concepts were later used in the Internet protocol; CYCLADES was one of the predecessor systems with the greatest technical influence on the Internet.
The network was sponsored by the French government, through the Institut de Recherche en lnformatique et en Automatique (IRIA), the national research laboratory for computer science in France (now known as INRIA), which served as the co-ordinating agency. Several French computer manufacturers, research institutes and universities contributed to the effort. CYCLADES was designed and directed by Louis Pouzin.
Conception and deployment
Design and staffing started in 1972, and November 1973 saw the first demonstration, using three hosts and one packet switch. Deployment continued in 1974, with three packet switches installed by February, although at that point the network was only operational for three hours each day. By June the network was up to seven switches, and was available throughout the day for experimental use.
A terminal concentrator was also developed that year, since time-sharing was still a prevalent mode of computer use. In 1975, the network shrank slightly due to budgetary constraints, but the setback was only temporary. At that point, the network provided remote login, remote batch and file transfer user application services.
CYCLADES used a layered architecture, as did the Internet. The basic packet transmission function, named CIGALE, was novel; however, it provided an unreliable datagram service (the word was coined by Louis Pouzin by combining data and telegram). Since the packet switches no longer had to ensure correct delivery of data, this greatly simplified their design.
“The inspiration for datagrams had two sources. One was Donald Davies' studies. He had done some simulation of datagram networks, although he had not built any, and it looked technically viable. The second inspiration was I like things simple. I didn't see any real technical motivation to overlay two levels of end-to-end protocols. I thought one was enough.”
The CIGALE network featured a distance vector routing protocol, and allowed experimentation with various metrics. it also included a time synchronization protocol in all the packet switches. CIGALE included early attempts at performing congestion control by dropping excess packets.
The name CIGALE—(French pronunciation: [siɡal]) which is French for cicada—originates from the fact that the developers installed a speaker at each computer, so that "it went 'chirp chirp chirp' like cicadas" when a packet passed a computer (Gillies and Cailliau 2000:38).
An end-to-end protocol built on top of that provided a reliable transport service, on top of which applications were built. It provided a reliable sequence of user-visible data units called letters, rather than the reliable byte stream of TCP. The transport protocol was able to deal with out-of-order and unreliable delivery of datagrams, using the now-standard mechanisms of end-end acknowledgments and timeouts; it also featured sliding windows and end-to-end flow control.
By 1976, the French PTT was developing Transpac, a packet network based on the emerging X.25 standard. The academic debates between datagram and virtual circuit networks continued for some time, but were eventually cut short by bureaucratic decisions.
Data transmission was a state monopoly in France at the time, and IRIA needed a special dispensation to run the CYCLADES network. The PTT did not agree to funding by the government of a competitor to their Transpac network, and insisted that the permission and funding be rescinded. By 1981, Cyclades was forced to shut down.
The most important legacy of CYCLADES was in showing that moving the responsibility for reliability into the hosts was workable, and produced a well-functioning service network. It also showed that it greatly reduced the complexity of the packet switches. The concept became a cornerstone in the design of the Internet.
The network was also a fertile ground for experimentation, and allowed a generation of French computer scientists to experiment with networking concepts. Louis Pouzin and the CYCLADES alumni initiated a number of follow-on projects at IRIA to experiment with local area networks, satellite networks, the Unix operating system, and the message passing operating system Chorus.
CYCLADES alumni and researchers at IRIA/INRIA were also influential in spreading adoption of the Internet in France, eventually witnessing the success of the datagram-based Internet, and the demise of the X.25 and ATM virtual circuit networks.
- Pouzin, Louis (1973). "Presentation and Major Design Aspects of the Cyclades Computer Network". "Proceedings of the NATO Advanced Study Institute on Computer Communication Networks". Sussex, United Kingdom: Noordhoff International Publishing. pp. 415–434.
- Pouzin, Louis (1974). "CIGALE, the packet switching machine of the CYCLADES computer network". "Proc of IFIP". Stockholm, Sweden. pp. 155–159.
- Pouzin, Louis (1975). "The CYCLADES network-present state and development trends". "Symposium on Computer Networks". Gaithersburg, Maryland: IEEE Computer Society. pp. 8–13.
- Gillies, James and Robert Cailliau (2000). How the Web Was Born: The Story of the World Wide Web. Oxford University Press. ISBN 978-0-19-286207-5.
- Louis Pouzin (editor), The Cyclades Computer Network: Toward Layered Network Architectures (North-Holland, Amsterdam, 1982)