Category
| Beauty , Fashion & Style |
| Jobs & Career |
| Travels |
| RelationShip |
| work & Life |
| Health & Wellness |
| Hot Stuff |
| Leisure |
| opinion |
| yoga |
| Vastu |
| Wellness & Beliefs |
Entries archive
Our poll
Main » 2010 September 30 » Network visionary
11:59 AM Network visionary |
| Cisco is in networking what Intel is in PC processors and Microsoft is in PC operating systems. So dominant that you don't often hear about its rivals. One company though that has successfully challenged Cisco in core areas of networking, such as routers and switches, is Juniper Networks. It was founded in 1996 by Pradeep Sindhu, a graduate from IIT, Kanpur, and a PhD from Carnegie Mellon University, US. Wikipedia describes him as "a technical visionary in the field of high performance and networked computing". Sindhu, now 57, was born in Bombay. His dad was in the army, and they moved around all over India, so, he says, he had no real hometown. "I guess the first home town was Chandigarh, where we built a house. And then we moved to Delhi, where I did my high school." He went to the US to do his Masters in 1975, and America has been his base since then. Juniper today is a $3.6 billion company, with 30 to 40% market share in several major categories of routers (those electronic devices within the network that ensure, for instance, that the email you send reaches the intended person, or that the website you ask for is what you get). The company, in 1999, had one of the most successful initial public offerings in American stock market history. By the end of the first day as a publicly traded company, Juniper's stock rose by 190%, taking its market capitalization to just below $4.9 billion. Sindhu served as chairman and CEO of Juniper for eight months between February 1996 and September 1996, and since then has been its vice chairman and chief technology officer. He was recently in Bangalore, where Juniper has one of its biggest R&D centres, and spoke exclusively to us on a range of issues, from what got him interested in networking to the future of networking as he sees it and the contribution the India centre is making. How did Juniper happen? I was working at Xerox PARC (Palo Alto Research Centre), the centre where the personal computer, ethernet and a lot of the things in high technology were invented. In 1987, a group of us from PARC had gone to Sun Microsystems. We designed, architected and launched the first set of Sun servers, the SS1000 and SS2000. After that the team came back to Xerox and made a pitch that we should use this technology to build image processing equipment for Xerox itself. That went on for three years. It wasn't very satisfactory. PARC understood software and systems, but Xerox Corp understood only toner and paper. And the conclusion I came to was I wanted to start a venture of my own, independent of a company's internal desires, and in a framework that was venture funded. I took a break of 3 or 4 months, and that was the time the idea of Juniper came about. Why did networking appeal to you? At that point it was becoming very clear that networking, especially wide area networking, was ripe for reinvention, because the networking industry was 10-15 years behind the computing industry. Given the exponential rate at which performance requirements were increasing, a new approach was needed, and that was the central insight of Juniper. What was the difference you made? We reinvented the way routers were built. At that time there was one technology for each network application. TV broadcast used one technology, telephone lines used another technology, communicating between different companies used a third. If there is one insight I can give you about networks, it is this: networks provide the maximum value when you have a single network and a single networking technology. Bob Metcalfe has said that the value of a network is proportional to the number of people and things connected. Nobody thought that IP (internet protocol) could be that unifying technology. Everyone was taking different bets. My insight was that IP can be that unifying technology, because scale is very important. Nobody really knew how to make IP go very fast. People like Metcalfe were predicting that the internet was going to implode because nobody knew how to scale it. The reliability of routers being made by Cisco, 3Com, Bay was awful, because they were being built with general purpose processors that were just not fast enough. The technical challenge Juniper solved was to build IP routers that scale at the speed that was required. The difference in our price-performance compared to the fastest routers available at that time was 25 times. Technologically what was the difference? We divided the work that routers did in two parts. The part that handles data packets as they come in and go out, that was done completely in hardware, so we redesigned silicon engines with the specific purpose of moving packets. This was something that a lot of people had said was impossible to do, you are crazy etc. The other part was the software, the control plane that determines how a packet should be forwarded through which interface, which we did in a general purpose processor. This separation of control plane and data plane was fundamental. And today, everybody builds routers this way. Cisco remains much bigger, why? Cisco was the incumbent, started over 25 years ago. We started out in the service provider market and we entered the enterprise market only four years ago. We have an over 30% share of the service provider market. We are the only company that has been successful against Cisco, and we have taken market share consistently. Cisco is a great company, very interesting model with a fantastic marketing organization and a system for buying up companies and feeding them into the distribution channel. Juniper's strategy is very different. We focus on high performance networking. We don't have the largest catalog of equipment. Why? Because networking is strategic both in the enterprise and service provider segments. More and more of the networking market will take the characteristics of high performance networking. How do you define high performance networking? It is networking at large scale, it is networks that have very high reliability, very high security, and are fully general purpose. Fully general purpose, can you explain that? Even with IP, people want networks that are specialized. We want to build equipment that is as general purpose as possible and put all of the value in the software on top. The computing world moved to that model some time ago, with Intel and Microsoft, and the same idea is there in cloud computing today. In networking, people are still building box based solutions, different kinds of routers for different things. But we believe it will go the computing way. Apple created the iPhone and allowed lots of people to build applications for it. So it caught fire. And it went through three revisions quickly, each time becoming more powerful. In networking, the development ecosystem outside -- partners, customers -- don't get any value other than the direct products that you give. Much of the complexity in building and running networks is in the operational part, the applications part. In taking a horizontal platform-based approach, we would be able to marshal resources outside Juniper. We are only about 3,600 engineers. But there are thousands of engineers outside whose talents are currently squandered in thousands of silos. They will now turbo charge the rate of innovation in the networking industry. Because of the traditional box-based approach, the degree of automation in networks is very primitive. This will change. Aren't others going the same way? They can't. Cisco has 6-7 different operating systems. Every time they do a new product line, acquire a company, it's a new operating system. Same thing with an Alcatel-Lucent or Huawei. We have always had the vision of having a single operating system, called Junos. This strategy of an open software ecosystem is about three years old. It's very complicated to execute if there are multiple operating systems. How important is the India R&D centre to you? Of our 3,600 engineers, 1,400-1,500 are in Bangalore. For us, the centre here is a competence centre, not a cost centre. The work we do in India spans our entire gamut of products. They work in the software space, in our security and routing products, in building of systems, in service and support. Some of the work on the most high end routers we build, such as the T1600, happens here. |
|
|
| Total comments: 0 | |
