Digital library (interview) RAI Educational

Vinton Cerf

Torgiano (PG), 13 September 1997

"The History of the Internet"

SUMMARY:

  • Cerf is known as the father of the Internet, but his partner Robert Kahn and many others were alsoinvolved in its early development (1).
  • The history of the Internet began in 1969 when the Defence Advance Research Project Agency, a part of the US military Defence Department, started the Arpanet. Kahn was at Arpa working on three different packing switching networks and realised that networks had to be connected and made to work as one. Kahn and Cerf solved this problem in 1973. It had nothing to do with nuclear war, as is sometimes thought (2).
  • Leonard Kleinrock, Paul Baran and Donald Davies were also involved in the Internet's early development. These three men had never met before the project that used packet switching, the Arpanet, was created (3).
  • The most popular aspect of the Internet today is the World Wide Web. About 65 percent of all traffic on the Internet today is Web-based. But the part that people use the most is electronic mail (4).
  • The Arpanet was used as a research vehicle until 1990 when it was retired. In around 1983 it was separated into two parts: one part stayed with the research community and was also ultimately retired, the other part went to the US military and was called Milnet. Now it has been marginally retired and replaced by an implementation of the Internet for the US military (5).
  • Today there are about 200,000 networks around the world connected to each other using Internet technology. They include local area networks, radio networks, satellite networks and big wide area networks that are used for telephone circuits to connect the pieces together. This will continue to evolve and satellite and cable television systems will probably be connected into the Internet in order to support multi-casting, the Internet's method of delivering sound and video to large numbers of people (6).
  • Internet packets can sometimes get lost. TCP protocols (transmission control protocol) is basically simple and is designed to make e-mail communication more reliable (7).
  • The TCP and Internet protocols were devised long ago and created a standard that everyone could rely on. These protocols have not stayed static; their design has changed and expanded in order to accommodate new functionalities. As new requirements come along, the protocols have to be reinvented to make them support new problems (8).
  • Now there is a new Internet protocol called Version 6 which has very large address space so that more computers can be put together on the Internet. Another development is the use of the Internet for real-time communications: voice, video and radio. These real-time communications require new functionality in the network, so priorities for certain packets in the Internet have been invented. Cerf and his colleagues are also experimenting with multi-casting techniques to distribute one packet to many different receivers, as well as video on the Internet (9).
  • The electronic mail protocols in the Internet are beginning to expand. The protocol that is used for delivering mail today is SMTP or simple mail transport protocol. It has been extended to ESMTP, extended simple mail transport protocol; the extension includes acknowledgements that your recipient has actually received his message (10).
  • Some applications use a lot more resource than others. Electronic mail uses very small amounts of the Internet, but the WWW, especially if it's transmitting audio or video, uses up a great deal more. MCI has decided that it is fair to charge differently depending on how much of the network resources are used to support the particular application (11).
  • It's hard to predict the future of the Internet. We know it is going to get bigger; by the year 2000 the Internet will be close to the size of the telephone system. Devices such as refrigerators, dishwashers and washing machines may become part of the Internet. They will have Internet chips in them and be connected to a network that goes around the house. Cars will be connected to the Internet and to a global positioning satellite system, so that they can get information about alternate routes, if there is traffic congestion (12).
  • Other forms of communication, like satellite communication and cable communication for television, can become a part of the Internet system. There are experiments with the transmission of Internet packets through television satellite and cable television signals. Some day the video itself will be carried as part of the Internet packets all the way to the destination (13).
  • There is no way any one company, or country or person could control the content that goes on the net. But there are reasons to be concerned about what is found on the Internet, particularly on what children can reach. We have to offer parents and teachers the ability to filter the information that comes out of the Internet, protecting children while allowing adults to continue to have a very powerful tool for research, education and just for fun (14).

homepage

lezioni


digital library

authorities
subjects
biblioteca digitale

autori

cerca

aiuto

INTERVIEW:

Question 1
You are commonly known as the father of the Internet. Why?

Answer
Lots of people have put this label on me. It is not quite fair. I had a partner whose name was Bob Kahn and together we designed the Internet and its basic protocols way back in 1973, some 24 years ago. But I think our public relations people got very carried away, and this label really belongs to at least two of us and probably to a thousand other people who have contributed to the development of the Internet.

Back

Question 2
Arpanet was the first information network What was its purpose and how did it develop to become the current Internet?

Answer
Arpanet started out as a research project by the Defence Advance Research Project Agency, a part of the US military Defence Department. They were very interested in finding a different way of communicating while they were out on the battlefield using computers, and they started with the Arpanet, which used telephone lines to connect computers. This was back in about 1969. This was a very successful project and they started to experiment with radio communications and satellite communications, all using this new technique called packet switching, very different from the way the telephone system worked. Bob Kahn, my partner, was at Arpa and was working on all three of these different packing switching networks and he realised that somehow we were going to have to connect the networks together and make them all work as if they were one. That was the Internet problem that he and I solved in 1973. So it started out as one network, then three networks, then we started to connect them together into the Internet. There is a misunderstanding about the history of the Internet. When the Arpanet, was designed there was an interest in resource sharing. They wanted to connect computers at about 30 different universities around the country that studied computer science and Arpa was funding them. So it had nothing to do with atomic bombs and nuclear war or anything like that. But what was interesting about the technology that they used - packet switching - is that it tends to be very robust because it was very decentralised. So the first network, the Arpanet, was simply used to connect a bunch of university computers together. After we realised that packaging was a very powerful technique we started to ask questions such as: is it possible to connect different kinds of packet nets together that could be used for military management? So for the first time we were starting to ask questions of a military type using this new packet switching technology. When the Internet was designed there was a question of how to make sure this it was robust, even if there were a war. If there are bombs breaking parts of the network, can it connect itself together automatically? So that was part of the design point for the Internet. The reason there is confusion about this is that in 1962 a man named Paul Baran experimented with an idea very much like packet switching to build a voice communication network that was intended to survive a nuclear exchange, but his design was never actually implemented.

Back

Question 3
What was the first established computer network ? How did it develop to become the present Internet and when?

Answer
In the early 1960s, several people had the same idea almost at the same time about a new way of communicating. Instead of working the way the telephone system works, which is called circuit switching, where the two telephones have a connection between them, which is set up and running all the time until they hang up, this new idea was called packet switching. The first to have the idea was a man named Leonard Kleinrock who was a kind of researcher at MIT and wrote his dissertation on the subject. The second to have this idea was Paul Baran. He was at Rand Corporation in Santa Monica, California, and he was studying a way of communicating by voice, but which took the sound and turned it into digital packets of data and could then be distributed through a network across the United States. His concern was to build a system that would resist a nuclear attack, and so the idea of packet switching, at least as Paul Baran conceived it, was intended to resist a nuclear attack. The third man to look at this idea, independently, was a man named Donald Davies, at the National Physical laboratory in London, and his interest was connecting devices, terminals and computers together with a switching system that would run very efficiently. Those three men never met each other before the project that used packet switching, the Arpanet, was created. The Arpanet project started in the late 1960s. It was funded by the Defence Advanced Project Research Agency, as I said, which is part of the US Department of Defence, but their mission had nothing to do with nuclear war. Their mission was to connect computers at universities where they were doing some research in computers science that Arpa was funding. So the first wide area packet switching network was designed to support research. After that successful project, which was demonstrated to the public in 1972, one of my colleagues, Bob Kahn, who worked on the Arpanet, went to Arpa and started to study how to use this packet switching idea in radio communication, mobile radio, or satellite communication. And at this point he called me, and said: "Vinton, I have a problem. We have different kinds of packet switching networks. But we want to connect them together because we need the military to use these communication tools in the field in mobile vehicles, on the ocean, where you can only use satellites to communicate from ship to ship and ship to shore, and we want to connect them up into the continental network" (which the Arpanet represented). So we had these three different network technologies to connect together. This problem was, of course, the Internet because it was connecting nets together. So Bob and I worked on that design from 1973.

Back

Question 4
Why is the Internet so popular today?

Answer
I think the most popular aspect of the Internet today has got to be the World Wide Web. We think about 65 percent of all the traffic on the Internet today is Web-based traffic. But although that is the most exciting part of the Internet, the part that people use the most is electronic mail. So behind the level of excitement of the World Wide Web is basic human communication - the ability to send electronic messages back and forth to each other - as an alternative to the telephone.

Back

Question 5
What happened to Arpanet? What is the military system network now and is it similar to the former Arpanet?

Answer
The Arpanet was originally conceived and built in 1969. It was used as a research vehicle until 1990 when it was retired. But before it was retired, around 1983, it was separated into two parts: one part stayed with the research community and that part was also ultimately retired, the other part went to the US military. Arpanet continued under a new name, Milnet. That network supported military operations for some time. By this time it has been marginally retired and replaced by an implementation of the Internet for the US military.

Back

Question 6
The structure of the Internet is made of single points. How will it develop in the future?

Answer
We all appreciate that the Internet was originally designed to connect different nets together. Today we think there are about 200,000 networks all around the world that are connected to each other using the Internet technology. Some of them are local area networks, some are radio networks, some are satellite networks and some are big wide area networks that are used for telephone circuits to connect the pieces together. We think this is going to continue to evolve, and in particular, we expect to see satellite and cable television systems connected into the Internet in order to support what we now call multi-casting, which is the Internet's way of delivering sound and video to large numbers of people.

Back

Question 7
Can you tell us in very simple words what a protocol is and why do we need to use it.

Answer
The Internet at its heart is very simple. For example, we all know what a postcard is: Internet packets are just like postcards, except they run about 100 million times faster. A postcard has a "from" address and a "to" address, and there is a limited amount of room to write things on. An Internet packet has a "from" address and a "to" address and it has a limited amount of space to carry information. Now, we know some other things about postcards. When you put a postcard into the post-box, it's not always sure that it will arrive on the other side. Sometimes postcards get lost. This is true of Internet packets also, they can also sometimes get lost. If you put three postcards into the postal system, they don't necessarily come out in the same order, they might not even come out on the same day. In the Internet we have the same situation: the packets can get out of order, they can get lost, and in a very strange circumstance, sometimes on the Internet an Internet packet can get copied, and so you may get two copies instead of one. So we have to make this rather unreliable communication network - the basic Internet packet or postcard - reliable for computers to talk to each other. The way we do that is to build another layer of protocol on top of the basic postcard and we call this layer TCP, or transmission control protocol. There is a very easy way to understand how that works too. Suppose that you wanted to send a book to a friend and the only way you could send it was with postcards. You would cut the book's pages out and glue them on the postcard. Then you'd remember that not every postcard would have the page number on it because you have the pages, so your friend may get these postcards out of order, so it would be a good idea if you numbered each postcard so your friend could put them back together in order. Then you remember that some of the postcards are going to be lost, so you keep a copy for yourself and you can send copies to your friend if he loses any. How do you know which postcard your friend has got? Good solution. You friend can send you a postcard every once in a while and say, I got every postcard, except postcard number 102, except then you remember that the postcard that tells you how many your friend got might get lost too. So now you have to decide, if I haven't heard anything from my friend about how many postcards he got, then I should start sending him copies of postcards that he has not acknowledged yet. And when he finally sends me an acknowledgement, you can throw any the postcards copies that you kept because he got all that he needed. That's all that goes on in the TCP protocol. It's very simple and the postcard protocol, the Internet protocol, is also very simple.

Back

Question 8
The TCP is pretty old but it is still widely used. Why?

Answer
The TCP protocol and the Internet protocol, although they were devised long ago, created a standard that everyone could rely on, so that any computer could implement those protocols and could talk to any other computer by the Internet and that has been its value. Standards make it possible for computers to communicate, even if they have never met each other. So the protocols have not stayed static; as the years have gone by their design has changed and expanded in order to accommodate new functionalities. As new requirements come along, you have to reinvent the protocol to make them support these new problems.

Back

Question 9
Can you tell us about new protocols that will carry video and audio?

Answer
Several forces are causing the Internet to change. For one thing the Internet is getting bigger, and we need to have more addresses in order to allow the network to get as big as it wants to grow. So we have a new Internet protocol called Version 6 which has a very large address space, just like a large telephone number space, so that more computers can be put together on the Internet. We are also starting to use the Internet for real-time communications, voice communications, video communication, and for radio. These real-time communications require new functionality in the network; some packets have to be delivered more quickly than others to keep the video pictures from breaking up or to keep the sound from breaking up. So we have invented priorities for certain packets in the Internet; some of them are treated better than others are in order to deliver them quickly. We are experimenting with multi-casting techniques to distribute one packet to many different receivers, just as you do when you listen to television or radio. We are seeing this application growing now on the network. There are about 2000 radio stations that transmit their sound through the Internet. We are also beginning to experiment with video on the Internet. Some day perhaps this program will be distributed through the Internet and received on the computer, not on a television set.

Back

Question 10
Sometimes e-mail is lost. You don't even know if someone received your mail. How can we fix this problem so that mail is a reliable system?

Answer
One of the applications of the Internet is electronic mail. And some people have noticed that just like the Internet or the electronic postcard, electronic mail can get lost too. There are several ways of dealing with this problem. First of all, if the e-mail cannot be delivered, and the system knows that it cannot be delivered, because the mailbox does not exist or the party's computer is asleep, then you get messages back saying: "this message could not be delivered". But there still are cases where you send a message and it doesn't get delivered and nobody knows this and of course you don't know if the recipient has got it or not. The electronic mail protocols in the Internet are beginning to expand. The protocol that is used for delivering mail today is called SMTP or simple mail transport protocol. It has been extended to ESMTP, extended simple mail transport protocol. The extension includes acknowledgements, so you can request when a message is delivered that an acknowledgement comes back to you. This will help people recover messages which have got lost.

Back

Question 11
Do you think that the Internet should have different costs for the final user if used for business or leisure? Should we have to pay in different ways?

Answer
The question of prices on the Internet is of great interest to everyone who uses it. For a long time the price of the Internet service was flat. Everyone paid the same thing no matter how much they used it. This was OK as long as nobody used too much. But as the Internet has grown, and as people's appetite to use the network for different applications has grown, some applications use a lot more resource than others do. Electronic mail, for example, uses very small amounts of the Internet, but the WWW, especially if it is transmitting audio or video, uses up a great deal more of the network resources. At some point you have to decide that people should pay for the amount of resource that they use, otherwise the person who uses it very little is subsidising the person who uses it a lot in a flat rate environment. Also, we know that for real-time applications, more of the network resource will be consumed than for non-real-time. MCI has concluded that the fair thing to do for our subscribers is to charge differently depending on how much of the network resources are used to support the particular application. So you will see differential charging in the industry as time goes on to account for these different facts.

Back

Question 12
What about the development of Internet in the near future.

Answer
It is hard to look into the future to see the new things that will happen on the Internet. We know several things are likely. First of all it is going to get bigger. We estimate by the end of the year 2000 that there will be somewhere between 200 and 300 million computers on the Internet. Today there are about 23 million. And just for comparison, on the telephone system there are about 700 million telephone lines, terminations on the global telephone networks, so by the year 200, the Internet will be getting close to the size of the telephone system. The second thing that we can predict is that devices are going to be on the Internet that you don't normally think of as being part of the Internet world. Today, personal computers, laptops, and desktop machines, and big web servers are part of the Internet environment. But we are starting to see little pieces, computer chips, that will do the Internet protocol being put into things like appliances. For example, your refrigerator or your hot water heater, the dishwasher and the clothes washer may become part of the Internet. They will have little Internet chips in them and will be connected to a network that goes around the house. You may ask why anyone would want to hook those appliances onto the Net. And there are some good answers to that question. The power companies would like to control some of the appliances so when there is peak in demand for electricity, they can turn off the hot water heaters and in that way moderate the demand for power during that period. It doesn't hurt the water heater and it keeps the power company from having to build much more capacity, which is very expensive. If they can save money by controlling the appliances, the consumers won't have to pay so much for electric power. That's just one example. The more interesting reasons are to connect up the automobile to the Internet, so that diagnostics can be performed on the car. Or you want to hook up the automobile to the Internet and to the global positioning satellite system, so that the car knows where it is and it can get information for you about alternate routes in case there is traffic congestion or an accident or the road is closed. So there are lots of reasons for communicating no matter where you are, and for devices that today are not on the Net that will be in the future.

Back

Question 13
And so satellite technology will join the Internet. Is there a big future for satellite technology as part of the system?

Answer
We already know that the other forms of communication - satellite communication, cable communication for television - can become a part of the Internet system. Today, we are experimenting with the transmission of Internet packets through television satellite signals. We are also doing this through cable television signals, sending Internet traffic along with the video. Some day the video itself will be carried as part of the Internet packets all the way to the destination. We are confident that all forms of transmission, whether optic fibre, radio, satellite, coaxial cable or ordinary twisted pair telephone, will all carry Internet traffic of one type or another.

Back

Question 14
Some say that we have to adopt censorship against the Internet and I would like your opinion. Who will control the Internet?

Answer
There are people who are concerned about the content of the Internet. We have to appreciate that today there are 23 million computers and 200,000 networks in something like 214 countries that are part of the Internet. There is no way any one company, any one country or any one person could control the content that goes on the Net. This would be like trying to control what people say to each other over the telephone. But there are reasons to be concerned about what is found on the Internet. Some people think that there should be some limitations on what children can reach. Since we cannot actually control the content of the network, what we have to do is offer parents and teachers the ability to filter the information that comes out of the Internet, which may go a young person who might need to have only limited access to the system. So we don't think that censorship is either possible or advisable for the basic Internet, but we do agree that it could be helpful to parents and teachers if there is a way of filtering information as it leaves the network for a youngster. That way the Internet is available for adult communication without limitation, and we still have the opportunity to allow children to enter the Internet and to use it as a very powerful tool for research and for education and just for fun.

Back

back to the top