Interview with Vinton Cerf One of the 'Fathers of the Internet' and Senior VP for Architecture and Technology, WorldCom
Share:
Vinton Cerf, One of the Fathers of the Internet
Senior VP for Architecture and Technology, WorldCom,
Chairman of the Board ICANN (Internet Corporation for Assigned Names and Numbers),
Board of Trustees, Gallaudet University.
AO/Beck: Dr. Cerf, it really is an honor to speak with you. I know that you are of the pivotal players in the development of the Internet, and I appreciate your time this morning.
Cerf: Hi Dr. Beck, the pleasure is mine. I'm happy to speak with you and the readers of Audiology Online and Healthy Hearing.
AO/Beck: I should explain to the readers that indeed, you and your area of expertise are a little out of the mainstream with respect to our normal interviewees and topics, but then again, I met you through the fact that your wife actually has a cochlear implant!
Cerf: Yes, exactly. In fact she is an Advanced Bionics cochlear implant patient, and she does very well with her cochlear implant. Additionally, I've worn hearing aids for some 46 years, so I am happy to work with you, Audiology Online and Healthy Hearing too. I think the recent progress with hearing aids has been fantastic, and thankfully, as my hearing loss has progressed, so too has hearing aid technology!
AO/Beck: I agree. We'll be interviewing your wife in a few weeks too, to discuss her thoughts and reflections on her cochlear implant. But in the meantime, I'd like to discuss the Internet and related issues with you, and we'll talk about hearing aids at a later date if that's acceptable? Let's start with your formal education and a little about the degrees you hold?
Cerf: Of course. My undergraduate work was at Stanford in 1961 to 1965. I graduated with a bachelor's degree in mathematics. Upon completion of that, I went to work for IBM in Los Angeles, and that's where I met my wife. After a a couple of years, I decided I needed to go back to school to get a little more advanced training in Computer Science, and so I went to UCLA and got my MS in Computer Science in 1970, and then earned my Ph.D. also in Computer Science in 1972.
AO/Beck: Pardon me for asking, but was that the first computer science degree from UCLA?
Cerf: No, unfortunately, it was the second! One of my colleagues, Kim Gostelow, beat me to it and I think I got the second UCLA Computer Science doctoral degree.
AO/Beck: Dr Cerf, despite the claim by Al Gore a few years ago, it seems to me that you are the true Father of the Internet. Can you comment on that please?
Cerf: Well, I think many people actually have a claim on that title, and perhaps I had a small claim on that too. In 1962, a man named Leonard Kleinrock was a graduate student at MIT and he wrote his PhD dissertation on a topic related to statistical analysis of communication networks and message flow and delay in that environment. His dissertation was published as a book by McGraw-Hill in 1964 entitled: Communication Networks: Stochastic Message Flow and Delay. Interestingly, that work explained the value of what we now call packet switching that serves as the underlying technology of most of today's data communication networks. Kleinrock went to UCLA, and wound up in the Electrical Engineering Department. Around 1962, Paul Baran was at Rand Corporation, and he was trying to design a voice control system for the US Military. The idea was to network a series of transmitters and receivers over the airwaves, and part of the plan was to take speech, digitize it, break it up into packets (Baran called them message blocks ), and then flood them across the network. In that way, if there were a nuclear attack, there would be multiple paths for information transfer, based on highly redundant paths across the network. Now what's interesting, is that the system was never actually built. Nonetheless, Paul wrote an 11 volume report called On Distributed Communication. Paul and I had the same thesis advisor at UCLA, Gerald Estrin. So, a lot of this was being in the right place at the right time and some of it was dumb luck. Anyway, there was interest at the US Defense Advanced Research Projects Agency (ARPA) to start hooking up computers used in the Computer Science Departments supported by ARPA across America. Larry Roberts was brought to ARPA from MIT's Lincoln Laboratory to run the ARPANET project. By 1968 I had been at UCLA for a year or so and Kleinrock was awarded a research grant from ARPA to model, measure and analyze the ARPANET. ARPA awarded the design and implementation of the packet network to a company in Cambridge, Massachusetts, Bolt Beranek and Newman (BBN). UCLA's Kleinrock ran the Network Measurement Center and I served as principal programmer in that group. So there was an amazing convergence of people working on the packet switched net and the project was very successful. The conventional telecommunications companies thought it was silliness, and they didn't believe in it, but of course we went ahead anyway. The technical people really came together at various universities, with the military and the private sector, and the result has been far more amazing than any of us could have predicted. After the first four nodes of the ARPANET were installed in late 1969, Robert Kahn, then at BBN, came to UCLA with one of his colleagues, David Walden, to carry out a variety of tests on the network. I worked closely with Kahn to carry out a variety of performance tests. In 1972, Kahn went to ARPA and I went to Stanford. In the spring of 1973, Kahn visited me at Stanford and described to me a problem he was working on: the interlinking of several kinds of packet networks (the ARPANET, the packet radio net and the packet satellite network). Our solution to this internetting problem became the technical basis for the Internet. I would say that our efforts to interconnect all of the different packet technologies in 1973 was probably among the watershed events in the development of the Internet. In May of 1974, we published the very first paper on how the Internet could work. Included in the paper was the description of what we now call the TCP/IP protocols and the basic architecture of the Internet. It took another five years to standardize these protocols and then five more to implement them on a sufficient number of operating systems that ARPA could require that all computers on its several packet networks switch over to run the new TCP/IP protocols, converting a system of distinct networks into the combined Internet. The cutover date was January 1, 1983. It had taken ten years from the concept of Internet to its deployment in ARPA's research community. I joined ARPA as a program manager in 1976 to run the Internetting and packet technology research programs and stayed there until just before the 1983 cutover to the Internet protocols. It would be another twenty years before the general public became aware of this new technology, thanks in part to the advent of the World Wide Web application running on the Internet system.
AO/Beck: I'm sure it must be very satisfying to know that just in the USA there are at least 110 million people using the Internet, as the result of the work you and your colleagues initiated?
Cerf: Yes, it is. I have heard numbers that are actually higher, perhaps closer to 168 million, but who really knows?
AO/Beck: Dr. Cerf, where is this all going? Where do you think we'll be three to five years from now? What are the three major issues coming down the information superhighway?
Cerf: Great question. Several things are going on now, which will likely impact where we are then. The first is the recognition that we may actually run out of 32 bit Internet Address numbers that are the lower level numeric addresses of the Internet Protocol (IP). They serve the same role on the Internet at telephone numbers play on the telephone network. I've been personally pushing for the implementation of IP version 6, which is the next version of what is being used today. Actually at this time we're running version 4, and that begs the question what happened to version 5? Well, that was an experiment that didn't pan out, and it expired in the 1980s. Nonetheless, the issue is how long it will take until we run out of numbers, and whether or not we really will run out. I think we will. The maximum number of addresses, based on the current 32 bit system is approximately 4.3 billion, or 2 to the 32nd power.
AO/Beck: And so what is the actual correction for this problem?
Cerf: Well, there have been a number of hacks, or patches called network address translation devices done over the last several years. But realistically, these are pretty limiting. The bottom line is that we need to use a new system that offers 128 bit address space, which is 10 to the 38th power. This will be more and more important as we move to even more cell phones and digital devices working through the Internet, and then when we add more Internet enabled PDAs (Personal Digital Assistants), more Internet enabled appliances and on and on, and we know it will work best if each system and device has its own address.
AO/Beck: So based on your best analysis, when do you suppose we'll actually start running out of addresses?
Cerf: Well, that is the key question, and the answer is truly that no one knows. My best guess is that 2006 is the right target date, by which time we need the 128 bit system up and running. Because Internet began as an experiment in the US, a number of US institutions and businesses have large allocations of Internet address space. Later entrants in other countries have smaller allocations so it is no surprise that there is a lot of interest in the larger address space in countries outside the US. Japan, China and Europe are enthusiastically working towards the implementation of IPv6 in part for that reason.
Additionally, another major issue right now is mobile access to the Internet. More and more people use radio based devices to connect to the Internet, and these are getting better very quickly. They are going to increase in popularity because they are inexpensive, relative to a PC, and because they have no wires, they are truly portable portals to the Internet. So, mobile access is the second major issue.
The third major issue at this time is convergence, meaning running voice and text over the Internet with radio and video simultaneously. So higher capacity access to the Internet is increasingly important, and there is no end in sight. So we're already seeing a shift from cable modems and digital subscriber loops, which have just really started to get their market penetration, to the next big thing, which will be Gigabit Ethernet. Gigabit Ethernet provides a one billion bit per second interface, and its impact will be significant.
AO/Beck: Can you tell me anything about the Internet and communication in space?
Cerf: For the last 4 years, I've been working with the Jet Propulsion Laboratory (JPL) and we're in the third iteration of our proposed system for deep space communication. The basic application is to extend Internet access across the solar system for space exploration. At this time, the only way to communicate into deep space is by radio though perhaps eventually we will be using light. Each space communication protocol previously developed was mission specific - and so prior protocols have not really been an asset. There is a standard link layer protocol, called CCSDS Link Layer, and that has been a useful standard. About half a dozen of us have been working at JPL to try to standardize higher layer communications protocols, so that we can communicate more effectively and efficiently with each subsequent space craft, using the communication assets of previously launched missions. The current standard Internet protocols, TCP/IP, are not well suited for deep space communication. It could literally take minutes to hours for a two-way communication to occur. Between Earth and Mars, even at the speed of light, the round trip transmission time is up to 40 minutes. Of course, even if we used lasers, the speed of light is the speed of light, and we cannot transmit faster than that at this point!
AO/Beck: Dr. Cerf, based on all of your knowledge and history, what has been the biggest surprise to you, regarding computer technology?
Cerf: For me, the biggest surprise has been the World Wide Web (WWW) itself. The man credited with developing the WWW is Tim Berners-Lee, and he worked at CERN in Geneva, Switzerland in 1989 or so. Of course Tim's work has its roots in the work done by Douglas Engelbart at SRI International in the mid-late 1960s. What astonished me was the outpouring of content that people took the time and trouble to create and place on the Internet using HTML (the early language of the WWW). The fact that people were willing to share so much information freely has been surprising to me.
AO/Beck: Before I let you go, can you give me your thoughts on the music industry and the person-to-person sharing and related MP3 download issues?
Cerf: The music industry has reacted negatively to the technology, but I really think there is no effective way for them to stop people from sharing music on the Internet. The entire music industry is built on the sale of albums, and that is just not efficient. People really seem to not want to pay 20 dollars for 12 songs they don't really want, packaged along with 2 songs they do like. So it's a real dilemma now, and the music industry will have to see what sort of new working business models can be devised.
AO/Beck: Dr Cerf, I really hate to let you go, but then again, I am way past our time limit. Thanks so much for sharing your time and thoughts with us today.
Cerf: You're welcome Doug. Thanks for your time too.
For more information on Advanced Bionics click here.
Click here to visit the Advanced Bionics website.
Senior VP for Architecture and Technology, WorldCom,
Chairman of the Board ICANN (Internet Corporation for Assigned Names and Numbers),
Board of Trustees, Gallaudet University.
AO/Beck: Dr. Cerf, it really is an honor to speak with you. I know that you are of the pivotal players in the development of the Internet, and I appreciate your time this morning.
Cerf: Hi Dr. Beck, the pleasure is mine. I'm happy to speak with you and the readers of Audiology Online and Healthy Hearing.
AO/Beck: I should explain to the readers that indeed, you and your area of expertise are a little out of the mainstream with respect to our normal interviewees and topics, but then again, I met you through the fact that your wife actually has a cochlear implant!
Cerf: Yes, exactly. In fact she is an Advanced Bionics cochlear implant patient, and she does very well with her cochlear implant. Additionally, I've worn hearing aids for some 46 years, so I am happy to work with you, Audiology Online and Healthy Hearing too. I think the recent progress with hearing aids has been fantastic, and thankfully, as my hearing loss has progressed, so too has hearing aid technology!
AO/Beck: I agree. We'll be interviewing your wife in a few weeks too, to discuss her thoughts and reflections on her cochlear implant. But in the meantime, I'd like to discuss the Internet and related issues with you, and we'll talk about hearing aids at a later date if that's acceptable? Let's start with your formal education and a little about the degrees you hold?
Cerf: Of course. My undergraduate work was at Stanford in 1961 to 1965. I graduated with a bachelor's degree in mathematics. Upon completion of that, I went to work for IBM in Los Angeles, and that's where I met my wife. After a a couple of years, I decided I needed to go back to school to get a little more advanced training in Computer Science, and so I went to UCLA and got my MS in Computer Science in 1970, and then earned my Ph.D. also in Computer Science in 1972.
AO/Beck: Pardon me for asking, but was that the first computer science degree from UCLA?
Cerf: No, unfortunately, it was the second! One of my colleagues, Kim Gostelow, beat me to it and I think I got the second UCLA Computer Science doctoral degree.
AO/Beck: Dr Cerf, despite the claim by Al Gore a few years ago, it seems to me that you are the true Father of the Internet. Can you comment on that please?
Cerf: Well, I think many people actually have a claim on that title, and perhaps I had a small claim on that too. In 1962, a man named Leonard Kleinrock was a graduate student at MIT and he wrote his PhD dissertation on a topic related to statistical analysis of communication networks and message flow and delay in that environment. His dissertation was published as a book by McGraw-Hill in 1964 entitled: Communication Networks: Stochastic Message Flow and Delay. Interestingly, that work explained the value of what we now call packet switching that serves as the underlying technology of most of today's data communication networks. Kleinrock went to UCLA, and wound up in the Electrical Engineering Department. Around 1962, Paul Baran was at Rand Corporation, and he was trying to design a voice control system for the US Military. The idea was to network a series of transmitters and receivers over the airwaves, and part of the plan was to take speech, digitize it, break it up into packets (Baran called them message blocks ), and then flood them across the network. In that way, if there were a nuclear attack, there would be multiple paths for information transfer, based on highly redundant paths across the network. Now what's interesting, is that the system was never actually built. Nonetheless, Paul wrote an 11 volume report called On Distributed Communication. Paul and I had the same thesis advisor at UCLA, Gerald Estrin. So, a lot of this was being in the right place at the right time and some of it was dumb luck. Anyway, there was interest at the US Defense Advanced Research Projects Agency (ARPA) to start hooking up computers used in the Computer Science Departments supported by ARPA across America. Larry Roberts was brought to ARPA from MIT's Lincoln Laboratory to run the ARPANET project. By 1968 I had been at UCLA for a year or so and Kleinrock was awarded a research grant from ARPA to model, measure and analyze the ARPANET. ARPA awarded the design and implementation of the packet network to a company in Cambridge, Massachusetts, Bolt Beranek and Newman (BBN). UCLA's Kleinrock ran the Network Measurement Center and I served as principal programmer in that group. So there was an amazing convergence of people working on the packet switched net and the project was very successful. The conventional telecommunications companies thought it was silliness, and they didn't believe in it, but of course we went ahead anyway. The technical people really came together at various universities, with the military and the private sector, and the result has been far more amazing than any of us could have predicted. After the first four nodes of the ARPANET were installed in late 1969, Robert Kahn, then at BBN, came to UCLA with one of his colleagues, David Walden, to carry out a variety of tests on the network. I worked closely with Kahn to carry out a variety of performance tests. In 1972, Kahn went to ARPA and I went to Stanford. In the spring of 1973, Kahn visited me at Stanford and described to me a problem he was working on: the interlinking of several kinds of packet networks (the ARPANET, the packet radio net and the packet satellite network). Our solution to this internetting problem became the technical basis for the Internet. I would say that our efforts to interconnect all of the different packet technologies in 1973 was probably among the watershed events in the development of the Internet. In May of 1974, we published the very first paper on how the Internet could work. Included in the paper was the description of what we now call the TCP/IP protocols and the basic architecture of the Internet. It took another five years to standardize these protocols and then five more to implement them on a sufficient number of operating systems that ARPA could require that all computers on its several packet networks switch over to run the new TCP/IP protocols, converting a system of distinct networks into the combined Internet. The cutover date was January 1, 1983. It had taken ten years from the concept of Internet to its deployment in ARPA's research community. I joined ARPA as a program manager in 1976 to run the Internetting and packet technology research programs and stayed there until just before the 1983 cutover to the Internet protocols. It would be another twenty years before the general public became aware of this new technology, thanks in part to the advent of the World Wide Web application running on the Internet system.
AO/Beck: I'm sure it must be very satisfying to know that just in the USA there are at least 110 million people using the Internet, as the result of the work you and your colleagues initiated?
Cerf: Yes, it is. I have heard numbers that are actually higher, perhaps closer to 168 million, but who really knows?
AO/Beck: Dr. Cerf, where is this all going? Where do you think we'll be three to five years from now? What are the three major issues coming down the information superhighway?
Cerf: Great question. Several things are going on now, which will likely impact where we are then. The first is the recognition that we may actually run out of 32 bit Internet Address numbers that are the lower level numeric addresses of the Internet Protocol (IP). They serve the same role on the Internet at telephone numbers play on the telephone network. I've been personally pushing for the implementation of IP version 6, which is the next version of what is being used today. Actually at this time we're running version 4, and that begs the question what happened to version 5? Well, that was an experiment that didn't pan out, and it expired in the 1980s. Nonetheless, the issue is how long it will take until we run out of numbers, and whether or not we really will run out. I think we will. The maximum number of addresses, based on the current 32 bit system is approximately 4.3 billion, or 2 to the 32nd power.
AO/Beck: And so what is the actual correction for this problem?
Cerf: Well, there have been a number of hacks, or patches called network address translation devices done over the last several years. But realistically, these are pretty limiting. The bottom line is that we need to use a new system that offers 128 bit address space, which is 10 to the 38th power. This will be more and more important as we move to even more cell phones and digital devices working through the Internet, and then when we add more Internet enabled PDAs (Personal Digital Assistants), more Internet enabled appliances and on and on, and we know it will work best if each system and device has its own address.
AO/Beck: So based on your best analysis, when do you suppose we'll actually start running out of addresses?
Cerf: Well, that is the key question, and the answer is truly that no one knows. My best guess is that 2006 is the right target date, by which time we need the 128 bit system up and running. Because Internet began as an experiment in the US, a number of US institutions and businesses have large allocations of Internet address space. Later entrants in other countries have smaller allocations so it is no surprise that there is a lot of interest in the larger address space in countries outside the US. Japan, China and Europe are enthusiastically working towards the implementation of IPv6 in part for that reason.
Additionally, another major issue right now is mobile access to the Internet. More and more people use radio based devices to connect to the Internet, and these are getting better very quickly. They are going to increase in popularity because they are inexpensive, relative to a PC, and because they have no wires, they are truly portable portals to the Internet. So, mobile access is the second major issue.
The third major issue at this time is convergence, meaning running voice and text over the Internet with radio and video simultaneously. So higher capacity access to the Internet is increasingly important, and there is no end in sight. So we're already seeing a shift from cable modems and digital subscriber loops, which have just really started to get their market penetration, to the next big thing, which will be Gigabit Ethernet. Gigabit Ethernet provides a one billion bit per second interface, and its impact will be significant.
AO/Beck: Can you tell me anything about the Internet and communication in space?
Cerf: For the last 4 years, I've been working with the Jet Propulsion Laboratory (JPL) and we're in the third iteration of our proposed system for deep space communication. The basic application is to extend Internet access across the solar system for space exploration. At this time, the only way to communicate into deep space is by radio though perhaps eventually we will be using light. Each space communication protocol previously developed was mission specific - and so prior protocols have not really been an asset. There is a standard link layer protocol, called CCSDS Link Layer, and that has been a useful standard. About half a dozen of us have been working at JPL to try to standardize higher layer communications protocols, so that we can communicate more effectively and efficiently with each subsequent space craft, using the communication assets of previously launched missions. The current standard Internet protocols, TCP/IP, are not well suited for deep space communication. It could literally take minutes to hours for a two-way communication to occur. Between Earth and Mars, even at the speed of light, the round trip transmission time is up to 40 minutes. Of course, even if we used lasers, the speed of light is the speed of light, and we cannot transmit faster than that at this point!
AO/Beck: Dr. Cerf, based on all of your knowledge and history, what has been the biggest surprise to you, regarding computer technology?
Cerf: For me, the biggest surprise has been the World Wide Web (WWW) itself. The man credited with developing the WWW is Tim Berners-Lee, and he worked at CERN in Geneva, Switzerland in 1989 or so. Of course Tim's work has its roots in the work done by Douglas Engelbart at SRI International in the mid-late 1960s. What astonished me was the outpouring of content that people took the time and trouble to create and place on the Internet using HTML (the early language of the WWW). The fact that people were willing to share so much information freely has been surprising to me.
AO/Beck: Before I let you go, can you give me your thoughts on the music industry and the person-to-person sharing and related MP3 download issues?
Cerf: The music industry has reacted negatively to the technology, but I really think there is no effective way for them to stop people from sharing music on the Internet. The entire music industry is built on the sale of albums, and that is just not efficient. People really seem to not want to pay 20 dollars for 12 songs they don't really want, packaged along with 2 songs they do like. So it's a real dilemma now, and the music industry will have to see what sort of new working business models can be devised.
AO/Beck: Dr Cerf, I really hate to let you go, but then again, I am way past our time limit. Thanks so much for sharing your time and thoughts with us today.
Cerf: You're welcome Doug. Thanks for your time too.
For more information on Advanced Bionics click here.
Click here to visit the Advanced Bionics website.
