Kahn has a communications perspective and Cerf has a computing perspective, and together they work to connect different computing networks – up to 256.
译自 50 Years Later, Vint Cerf and Bob Kahn Remember the Birth of TCP/IP，作者 David Cassel。
Vint Cerf, along with 85-year-old Bob Kahn, gave a special talk in the room where they first wrote the TCP/IP protocol more than 50 years ago.
"It's thrilling to be here to celebrate the 50th anniversary of this work that Bob Kahn and I have started and the contributions of so many people," Cerf said during the week-long celebration speech.
He also received some glowing applause for the philosophical narration. "To be able to celebrate the 50th anniversary, it's important that you're still here to celebrate it!"
Cerf is now the vice president of Google and their chief Internet evangelist. Kahn is now the Chairman and CEO of the non-profit National Research Initiative, Inc., which promotes research and development of national information infrastructure. But Kahn used the internet connection, which he helped write the underlying protocol, to appear virtually with Cerf – waiting where the event took place – so that together they could tell the story of the birth of the internet.
and some of the people who helped them create this world-changing moment......
Network effects
At another event that weekend, Cerf's memories go back to his days as a UCLA graduate student, working on a bulky Sigma7. In September 1969, Cerf joined the first node of ARPANET, and the next connected node was added in SRI in October. Cerf is part of the "Network Working Group" created by Steve Crocker, who helped develop ARPANET's protocol.
Cerf remembers that Bob Kahn was also "very involved in the design of ARPANET" (ARPANET, Cerf explains, was "a precursor system designed to prove that packet-switched communications were going to be a big thing for computers.") But they're not the only ones thinking about networking. In 1971, the University of Hawaii also established ALOHAnet, the world's first wireless packet network, based on radio frequency.
Both ARPANET and ALOHAnet "really lead the way...... What we saw in early 1973," Cerf said later.
Or, as Kahn puts it, at the beginning of the 70s of the 20th century, "many different networks" already existed. By 1972, Bob J. Brown, who was working at DARPA at the time, was in the driver's seat. Kahn even realized that similar technology could be used for military command and control. (Cerf later joked that transmitting packets on telephone lines wouldn't work for the military because "tanks would run over wires and ships would get entangled.") So Kahn had to switch to radio. Specifically, a packet radio system using a moving vehicle and a land radio repeater......
Now they need a way to connect these networks – to allow the radio-based network to actually reach the computers on the ARPAnet (which, as Kahn puts it, "in order to make this mobile radio network useful in every sense of the word." ”）
There are different interfaces, data rates, and error control – but in another talk, Cerf succinctly completes the story. "So Bob· Kahn had a lot of things in his head in mind, and when he showed up at Stanford, we started trying to figure out how to connect these different types of networks to make them more unified from the perspective of the host computer at the edge.
"It took us about six months to get the job done."
Tips from PARC
There is also a participant. Xerox PARC joined ARPANET back in 1973 – Cerf noted its proximity to Stanford University, which undoubtedly led to some cross-pollination. Then, Cerf surprised the audience by having another important figure from 50 years ago – John Shoch – appear in front of them. At the time, Shook was a student at Stanford University, working at PARC. (Shook remembers, "It was a great time in this particular field of computer and communications research.") It's hard to recreate the passion in a whole new field. ”）
But Cerf specifically remembers Shoch's "involvement" when he attended a Cerf workshop at Stanford — Shoch noted that at the time, "I was technically a full-time employee and a full-time student, which was a bit odd." "Shoch is working on a very relevant and relevant project – but also a proprietary one. (PARC Generic Packet, which later led to the Xerox Network System.) ）
Cerf recalls with a laugh, "They couldn't really tell us exactly what they were doing in the workshop. But they're excellent collaborators – so they're trying to hint at something that could go wrong that we should be aware of. ”
"So I want to thank John for his efforts in helping us steer this complex path to our final destination!"
In a later presentation, Cerf noted that the early ARPANET protocol "assumed absolute success of the network." So it's very helpful to see examples of three different alternatives – those that have built-in considerations for end-to-end packet loss recovery (Ethernet for packet radio, packet satellite, and PARC).
Cerf also admits that "we had the help of many participants, because around the end of 1972, an organization called the International Network Working Group was formed. ”
An enduring paper
Kahn has a communications perspective and Cerf has a computing perspective, and together they solve this problem. At an upcoming seminar in the UK, they sat down and tried to document their work.
Cerf laughs when Kahn recalls that Cerf didn't write anything five to ten minutes later. "This is understandable, as this is a complex issue with multiple factors...... So I started writing. Because when I'm documenting things, I tend to keep writing, keep writing, and then look at what I've written, and then figure out how to structure it to get to the end I want.
"So I've written eight pages. Vint came back, looked at it, and said, 'Okay. I know what to do. I think he wrote the next 20 to 30 pages or so, and then we iterated on what was on paper...... It was a great experience.
"Vint seems to remember that it lasted a day and a half or two, and I seem to think it was only one day. It could be any one, it really doesn't matter because we got the job done. ”
Cerf also adds, "There is a plaque next to the elevator describing the work we did back in 1973......"
Cerf notes that the final paper was published in May 1974. "So it's been 50 years – the internet is thriving in almost every possible dimension."
Over the weekend, Cerf remembers where he was on the day of the Internet's 25th anniversary: for a photo shoot for Newsweek with Steve Crocker and Jon Postel, who together implemented the ARPANET protocol.
Judith Estrin, 50 years old
The story continues with the appearance of Judith Estrin, who is now the CEO of JLAbs but was a master's student at the time, joining the project after the specification (and the first implementation) was in place. After creating the TCP/IP protocol in 1974, "I started testing – late at night, early in the morning – with UCL [London] and BBN. Estin said she continues to pursue a career "helping to build the networking industry, including LANs and the Internet." (Among other things, according to Forbes, Estrin served as Cisco's chief technology officer from 1998 to 2000.) ）
But 50 years later, Erin shares three principles he learned from Vint and Bob. "One of them is to plan for uncertainty, and the agreement doesn't assume something from below or above." A related principle is "interconnecting diversity for collective power, not homogeneous expansion" – which implies an open systems environment and a degree of collaboration and interdependence.
Estrin went on to point out that the world today seems to be moving away from this idea of open systems to the integration of digital services, "which are now getting more and more powerful...... We make the rules from the community, and you have the rights and responsibilities...... Now the rules are set by the leader......
"I think we should step back and think about these philosophical questions as digital services have come to dominate our lives, and their impact outside of technology is increasing."
感谢 The Jetsons
Shoch recalls Cerf's point that engineering is all about turning science fiction into reality. He then thanked the pioneers who created "Detective Di Renjie" – "So, broadly speaking...... invented the Apple Watch" — and Hannah Barbera, the creator of The Jetsons, a 1962 animated series set in the future where people can make video calls......
It took half a century to define the architecture for the first time—and then to implement it over the next few decades. But looking at it all, the architecture proves to be both "very important" and "well-conceived...... "Your paper and the work of others have come up with ideas that have stood the test of time." ”
Shoch admits some architectural errors. They initially only used an 8-digit number to represent the network, "because we thought there were only 256 ...... 'How could there be more than 256 networks?!' ’”
But what impressed him most was that this half-century of evolution proved that the architecture framework was truly flexible – "able to adapt to orders of magnitude changes in the number of users, bandwidth, and geographic complexity." No one could have predicted all this. (Cerf also adds "technical diversity" and notes that fiber optic connectivity emerged in the 80s of the 20th century...... ）
"Looking ahead, I see a lot of uncertainty and things that I can't predict. I predict the future is not as good as Hannah Barbera. But having an open architecture – where we can have long discussions about what that means – gives me confidence that there's enough flexibility to accommodate new applications, new communications – and we want that to continue. ”