Avatar

Ever since I can remember, I’ve loved solving problems. But my dream, or at least my Jewish mother’s dream, was to become a lawyer. It was an accidental meeting with an old friend and my wife saying that I’m too naïve to be a lawyer which made me change direction and become an engineer. Two years later, I began working in a small and promising startup named ‘Galileo’. It was the start of a challenging and enjoyable career, and it made me admit that engineering is more for me. (Luckily, I listened to my wife and didn’t become a lawyer!)

In the last 23 years, I worked in some of the iconic silicon startups in Israel: Galileo – acquired by Marvell, Dune – acquired by Broadcom, and Leaba – acquired by Cisco. I was privileged to work with legendary engineers and entrepreneurs – Avigdor Willenz, Eyal Waldman, Nafea Bshara, Ofer Iny, and Eyal Dagan. It was and still is a fascinating journey that raises the inevitable question:

“What does it take for a successful technological initiative?”

Putting luck – which is always a blessing to have – aside, I’ve identified and distilled five essential points. All these points can be found in the successful startups I mentioned above, and we followed them while creating Cisco Silicon One TM.

  1. Have a good understanding of the market you play in. Ask yourself who are the players, what are the market drivers, and what are its limitations. Without understanding it, you may find that your brilliant technical solution cannot be adopted because of a limitation that may not even directly relate to the problem you solved or the solution you came up with.
  2. Find a hard problem. Once you’ve identified the problems, select a difficult one. Working on an easy one is the worst thing you can do – you waste your time on something that others can also solve, and instead of being a leader you position yourself as “yet another solution.” We always want to be second to none.
  3. Ensure a thorough understanding of the essence of the problem. Finding a problem is only the first step. You need to get to the root of it and understand every aspect. Most of the time these problems are multi-dimensional, and without a deep understanding of all dimensions, you won’t be able to come up with a good enough solution.
  4. Develop a creative solution. For this, you need to have a group of out-of-the-box thinkers that challenge all that we know and bring forward a surprising answer.
  5. Deep technological knowledge is essential. Let’s say you understand the market, you’ve found a difficult problem and came up with a brilliant solution. Can you execute it? Do you understand the technological barriers that you’ll need to cross or bypass? What is the best technology for implementing your solution? Only deep technological understanding can bridge the gap between a brilliant logical solution to an implementable, production-worthy product.

Cisco Silicon One is an excellent example that demonstrates how following these principles results in an innovative solution.

When we started thinking of Cisco Silicon One, as a team we already had more than 100 years of experience and a good understating of the market we play in. We questioned all known assumptions and tried to focus on the most challenging questions that no one managed to solve until then. Here are a few examples:

  • Why do people build different products for switching and routing?
  • Must programmability come hand in hand with lower performance?
  • Can we overlay all roles of a line-card device, a fabric-element, and a standalone device on one piece of silicon?
  • How do we untie the Gordian knot of high performance and high power consumption?

We tore into these difficult challenges and looked at them from all angles, trying to get to the root of each problem. Cisco Silicon One architecture is the result of this thorough thought process and tremendous intellectual efforts by all team members.

Dealing with unsolved challenges also means pushing the technology to the edge. We used our wide and deep technological knowledge throughout the silicon development process to come up with innovative implementations and bypass all technological barriers.

Q100 and Q200 devices are the realizations of all these efforts. These products took Cisco Silicon One architecture from the drawing table to the real world, demonstrating unmatched performance and processing capabilities.

Want to get a glimpse of how we solved one of the challenges? Read my white paper on hybrid memory architecture that enabled us to unify the product lines of high bandwidth switches and the sophisticated, heavy processing routers to a single product line.



Authors

Aviran Kadosh

Principal Engineer

Hardware Engineering