Counting the New York Stock Exchange and Chicago Board Options Exchange amongst its customers, Solarflare is currently lighting up the trading business. High Frequency Traders sat down with CEO, Russell Stern, to find out what the future holds for high-speed networking.
HFT: Can you give us a little background on Solarflare? In what ways are your solutions being used in the high frequency trading space?
Russell Stern: Our initial background was as a semiconductor company. Our chief investor Bandel Carano of Oak Investment Partners was also involved with a Cambridge-based company called Level 5 Networks. It had been founded by people from the University of Cambridge Computer Science Laboratory under Dr Andy Hopper and the AT&T research team. These guys were basically working on an R&D project to create the world’s fastest network. When AT&T shut down the research project, Andy founded Level 5 and we acquired them in 2002. Even at that point Solarflare was focusing being a semiconductor company, but about a year and half ago we decided to enter the end user channel directly with our own branded products. That’s what started this rocket ride that we’ve been on.
We decided to focus on the HFT because of some particular characteristics of our product. Namely that we are able to deliver extremely low-latency and extremely low latency jitter at extremely high message rates, all without requiring the customer to modify their application. We spotted another trend as the industry started to adopt 10gb. This was that the end users were becoming fundamentally as sophisticated as us in developing the technology, that they were hiring people with similar skills and capabilities and that they were making decisions about what type of networking adapter and connector that they were putting into their applications. I think it was largely overlooked by the server vendor community that the first thing an application touches as it runs in the server is the network adapter in their system. We are their connection to the outside world. How we interface with that application is very complicated and very intricate and is the source of much of the performance problems that they were trying to solve.
We believed that we had an answer for these end users and they seemed to be both capable and willing to make these types of decisions on their own rather than asking the OEMs to make them for them. They also had the budgets to buy this type of technology. That encouraged us to get into that market, so we started working with a big customer in the UK that gave us the impetus to enter the market. We also did some work with STAC Research, an independent benchmarking organisation in Chicago that fundamentally tries to compare different solutions. Consequently, we published some benchmarks with partners like IBM, Cisco and Arista.
That gave us a little bit of visibility. We started building up our manufacturing capability and we now have over 250 customers, made up of the biggest banks and organisations like the New York Stock Exchange, the Chicago Board Options Exchange and some others. It’s a market where companies not only understand where they can get the benefits in their applications, but that also provides direction for us in developing new products.