SM: What is the nature of the application delivery kernel? What kind of optimization can you gain?
LC: We designed all of our own hardware, and we used mostly merchandised chipsets. Because of that, our hardware is very efficient. We are designed for very fast application layer processing and deep packet scanning. When most other vendors go from layer 4 to layer 7, their performance drops 50%. In our case it is a 10-20% drop in most cases. We do not have ASICS but we do have our own FPGA, which does a lot of the routine work for us.
SM: Your secret sauce is in your handling of the application traffic. The layer 4 to layer 7 translation is something you do faster than anyone else on the market. The power savings are gained by using a parallel architecture, as opposed to a sequential architecture. Is that a fair summary?
LC: It is. The other thing is that we have a comprehensive design that is more efficient.
SM: This took you two and a half years to build, and you financed the whole thing yourself?
LC: It has been funded by me and a really friendly investor. We have a few institutions now, but they are not traditional venture companies; rather, they are friendly overseas venture companies. It is passive money.
SM: Who else was involved with the creation of this product?
LC: I have two of the best engineers from Foundry, both of whom had retired. They are working more on a volunteer basis. They are not getting paid either. They are really doing it for fun, although they have a little bit of equity. We now have 115 employees and over 100 customers.
SM: How much money have you guys pulled into this?
LC: We have raised up to a C round today. Together it totals $39 million.
SM: You had a site in Beijing and one in India, which gave you a global presence. What are some of your other milestones?
LC: In June 2005 we finalized our architecture design. Winning “Best of Show” at Interop Las Vegas and Tokyo in 2007 was significant, and that was in conjunction with the formal announcement of our product in the US in May and Japan in June 2007. We have the AX series product, which is application delivery focused; the ID series, which is focused on identity management; and the EX series, which focuses on bandwidth management. Our first major customer came at the end of 2007, when we won Toyota.
Most players in our marketplace build a layer 4-7 product. We say that layers 4-7 are the integration of layers 2-3. We integrated layers 2-3 into layers 4-7. We have the intelligence to detect fail-over quickly. They way other players detect a fail-over is by monitoring an application, and those tend to take a longer time to fail. We monitor at all layers, so we know immediately when it is failing and we can switch over. Our competition will do fail-over anywhere from two to 45 seconds. We typically do it in half a second, and on the long end will do it at two seconds. That is the reason we won Toyota, our first major win, and that was a milestone for us as well.
SM: Were your customers involved at all in the process of building the product?
LC: I knew the customers very well because of my ties to Foundry. I talked to the customers before I even started the company. I figured once we had the platform right, we could add additional features very quickly.
SM: You knew that customers had a defined problem that you could solve?
LC: Yes. Toyota was one of those interested. After Toyota, we have had all of the key wins. We have one of the largest cable companies on the East Coast as a customer. For that customer we added a notification platform into the product, so we can provide notice at 50% bandwidth, 80% bandwidth, etc. We offer the load balancing and do it well enough to compete with anyone, but we have extra features that others do not have. We can customize our product for key customers.
This segment is part 6 in the series : Still Innovating in Networking: A10 CEO Lee Chen
1 2 3 4 5 6 7