Professor Nick Mckeown Freng

PROFILE EVOLVING THE INTERNET Professor Nick McKeown FREng He may have given the world the technology that speeded up the internet, but in his next move, Professor Nick McKeown FREng plans to replace those networks he helped create. It would be hard to think of anyone more MIND MADE UP appropriate to interview over an internet Although McKeown did what he was told link between Sussex and California than by his new employers, this had not always Professor Nick McKeown FREng. After all, his been the case. Had he heeded his careers PhD research findings delivered a tenfold master at school, he would have dismissed increase in the speed of routers, which engineering altogether. When McKeown let enabled the internet to handle the traffic on that he was considering an engineering created by Skype and other services. education, inspired in part by his father, Back in the 1980s, McKeown was assigned Professor Pat McKeown FREng, an engineer his first research tasks at the Hewlett Packard and entrepreneur, he recalls that the careers (HP) Labs near Bristol. At the time, he hoped master’s response was: “You are too smart to work on artificial intelligence or computer Nick McKeown gives a TED talk at a conference in to be an engineer. You should go into Monterey in 2006 © Robert Leslie architecture, the popular topics of the day. something creative.” McKeown laughs and However, HP decided to focus McKeown’s says: “From that moment my mind was time and research resources on internet made up. I decided: ‘Right, I am going to be router architecture, which quickly became an engineer!’ one of the hottest areas of technology. This McKeown, now Professor of Electrical was before the advent of the World Wide Engineering and Computer Science at Web and before most people had heard of Stanford University, delights in relating the internet, when network connections this anecdote, and how it left a lasting were mostly between a scattering of big impression. He ensures that he creates links computers in large organisations. Since that with schools, including his old one near fateful HP decision, McKeown has been at the Bedford, Wootton Upper School, helping cutting edge of research into the architecture to ensure that future generations do not of networks rather than computers. encounter similar ignorance about the INGENIA ISSUE 64 SEPTEMBER 2015 47 Professor Nick McKeown FREng PROFILE McKeown studied for a degree in electrical However, it wasn’t speed that mattered and electronic engineering. to Cisco at the time, so much as adding On graduating in 1986, McKeown joined new features to its router boxes. This was HP Labs in Bristol. At that time, a PhD was three years before the first web browser, not essential to get a job in research in the Mosaic, and only shortly after the industry UK. Everyone at HP’s main R&D centre, in had settled on the best protocol, TCP/IP, to Palo Alto, California had to have a PhD, but coordinate the data traffic between systems. there were not enough people in the UK with The early work on routers was all about engineering PhDs. “So they took a load of the algorithms that control the packets of us fresh out of our undergraduate degrees. data that the hardware has to process and I came to realise later how little I knew. They ensure that they get to the right place in gave us this very expensive, well-funded the right order. After Bay Bridge came Tiny- education, paid us a salary, allowed us to Tera,which McKeown says “was our attempt make lots of mistakes, build stuff. It was an to see how you could build a one-terabit incredible apprenticeship.” per second switch out of semiconductor This thought sets McKeown reflecting technology. That seemed like a very high on the state of the UK at that time. HP’s recruitment problems were, he says, speed at the time. It needed specific algorithms for scheduling packets.” Those “a very sad reflection of the awful state of Professor McKeown directs a programme in Istanbul that teaches Turkish high school students the engineering and engineering education.” At scheduling algorithms were at the heart of fundamentals of computer science. The primary purpose of this program is to give rising high school juniors, especially girls, their first experiences of coding © Asena Gencel that time, he says, the government’s view his PhD. was that there was no future in technology The networking industry eventually and no one cared about manufacturing. cottoned on. After completing his PhD in CHEAPER CHIPS difference between it and the SRAM that it would replace.” This explained why, as McKeown recalls it, 1995, McKeown joined Cisco Systems for One of those developments came as a when he graduated, the most popular career a few months before moving to become surprise to the makers of expensive static As with Abrizio, McKeown set up another for electrical engineering graduates was a professor at Stanford University. At Cisco RAM (SRAM) chips. Dynamic RAM (DRAM) business to help to transfer the technology Professor Nick McKeown FREng accounting, followed by chartered surveying. he “helped architect”, as he describes it, the chips of the sort used in PCs are inexpensive, from the university into commercial use. Engineering came third. gigabit switched router (GSR), that became high-volume devices, but they have There was more to the idea than just importance of engineers, and the creativity mechanical engineer as a mechanic!” Cisco’s flagship router. limitations. The physical dimensions of DRAM algorithms: it needed chips, designed to work involved in engineering. McKeown’s own journey to California, WEST COAST CREATIVITY McKeown’s portfolio in innovations chips were great for making cheap chips, but those algorithms. In 2003, along with one of One way in which he does this is by and into computer networking, was not For an aspiring engineering researcher, went on to change the networking they limited the speeds switches and routers his PhD students, he set up Nemo Systems, hosting a course for school students based straightforward. He initially went to the US certainly looked more attractive, business. Tiny-Tera gave birth to a spin- could handle as data packets came and went. another chip development business, this on the teaching methods developed Durham University to do a degree in especially for someone whose work exposed out company, Abrizio, which made chips “As a network gets faster, you need faster time making DRAM chips for networking at Stanford. Four students from his old mechanical engineering. However, he him to the excitement of California and its for switching hardware. It was set up with and faster memories in order to buffer these systems. Once again, within two years, the school in Bedfordshire joined a class of found Durham very old fashioned and “a burgeoning high technology sector. venture funding in 1997, while McKeown packets,” says McKeown. It got to a stage fledgling business changed hands, when 100 or so in Istanbul. Here, McKeown and bit out of date from an engineering point It was McKeown’s masters research at took time off from Stanford and then, where packets were arriving faster than Cisco acquired it. colleagues teach high-school students of view”. So he left after a year, to go to the University of California at Berkeley and two years later, PMC-Sierra bought the the chips could handle them. So router McKeown’s early days are littered with from Turkey and the UK how to program Leeds “to swim with Adrian Moorhouse”, then his PhD work that shook up thinking company. makers went for faster SRAM technology, patent applications that went on to earn in Java, as an introduction to the field who went on to win a gold medal in the on computer networks. The official title of Those were heady days in Silicon Valley. even though SRAM chips were 75 times Stanford and Berkeley millions of dollars. of ‘computer science’. McKeown prefers Seoul Olympics in 1988. McKeown himself his PhD project was Scheduling Cells in an McKeown was, as he puts it, in the right more expensive per bit than DRAMs. “The However, he brought that flow of patents to this term to ‘ICT’ or ‘IT’ which he feels are narrowly missed being in the UK’s Olympic Input-Queued Cell Switch, but what it really place at the right time. A key event was networking industry was using about $500 an end in 1999. “I don’t patent anything now. associated with the maintenance and swimming team in 1984. “There were just meant, McKeown explains, was how to make when he saw the Mosaic web browser. million per year of SRAMs. And they were too I put everything into the public domain, administration of computers and networks too many good breaststroke swimmers in switches and routers go faster. It began with “I couldn’t have predicted what would small, too expensive, too power-hungry.” all of the software, all the ideas, all the rather than software development, creation the UK at the time.” the masters course when, with three fellow happen after that, through the rest of the Cue McKeown, with his algorithms: algorithms.” and programming. He says “These are Leeds wasn’t just about swimming masters students, he built “the world’s fastest 90s, but I realised that this was huge. I spent “We figured out a way that, with a clever This reflects McKeown’s views on the necessary roles but they differ greatly from though. Armed with a scholarship from router”, 10 times faster than anything that the next 10 years on algorithms, techniques, algorithm and a one-page proof, you could role of research in universities.

Load more