Thought Leaders in Big Data: Franz Aman, Chief Marketing Officer, Silicon Graphics (Part 2)

Sramana Mitra: I would like to talk in more detail about some of the customers and problems you discussed. You can pick whichever your favorites are. Let’s do three or four use cases. I would like to dive into the depths of what problems you are solving and how you are solving them. What role does SGI play in terms of what you need in order to solve those problems? What does a big data solution look like in various scenarios?

Franz Aaman: One of our principles is to deliver complete solutions and focus on the outcome for the customer, not just the technology ingredients. To that extent we bring together our own capabilities, but we also integrate the part of the software it takes to come up with a complete solution.

Here are a couple of use cases which I find very illustrative of what we do. You are going to see a lot of common trends. One of those trends is that we know how to deal with massive amounts of data. I just looked at a slide of SGI from 1993, and one of the labels on that slide was “Big Data.” Everyone is talking about big data today, but here are some examples: One of the customers we are working with is the Square Kilometer Array. In particular I am talking about the part C center in Australia. At present, the knowledge of mankind today equates to about 1.5 exabytes in total. This is the size of our digital universe. Those radio telescopes that are being built in Australia will eventually receive radio signals that are going to equate to about 1.5 exabytes of data a day. One of the big problems is going to be determining which data we are going to throw away. We have to do so quickly, because we can’t possibly process and store all that information. The particular challenge for that project is taking all that data and making sure we store the relevant items and make them available not just to researchers, but to a broader public as well. We also help with analysis and visualization. To me that is fascinating, because it may finally answer the question, are we alone in this universe?

Another example, which has a greater impact on a day-by-day basis, is fraud detection. It is very much a real financial transaction use case. It is not a scientific application; it is something very real. We help a large global carrier figure out fraud based on duplicate electronic postage and electronic currency. Some people who have used electronic postage have figured out that they can use not only one, but 10,000 [of the same electronic postage stamp]. You have to figure out how to recognize those duplicates right away and then destroy whatever mail is associated with them. This is just one of our systems. The system has an in-memory database which decides in real time whether something is a duplicate. Whichever envelope is zipping by at an amazing speed, we take a quick read and check it against an in-memory database. Have we seen it before? If not, we store it. If we have, the envelope goes right down the shoot and disappears. The ROI for that system was not measured in months or years, but in weeks. This is something very real at a human scale, and I think it is amazing. In that case I looked closely at an external integrator to put together the solution we have used in the memory database technology from Oracle, and we put all that together.

In the case of the Quick Kilometer Array, we are using a lot of our own computing capability as well as storage from a third party. Then we use software from SGI that makes it possible to track all the data, virtualize different storage tiers, and give everyone easy access to find the data whenever they need to, regardless of where it is stored. Many of today’s manufacturing and supply chain companies already started to compute and design – they are now starting to model and simulate. Even smaller companies are taking this step because they want to increase the quantity, and they want to improve the capabilities they deliver to the supply chain further down to some of the bigger manufacturing companies, whether it concerns airplanes, cars, etc.

There are so many things you can do that you couldn’t do with traditional scale product types and tests. Today’s problems are becoming much more multi-physical and don’t just have to do with aerodynamics, where you can save a wind tunnel, for example. Now it is also about problems with interactions, thermal properties, or interactions between structural capabilities. You can use new manufacturing capabilities like additive manufacturing. In 3-D printing you can manufacture nice curves and bends for the first time, and you don’t have the limitations you had with traditional drilling and C&C type technology. There are companies that have started printing houses in 3-D. It is a fascinating area. Some of our customers in Formula 1 [auto racing] don’t have time between races to do prototype testing in wind tunnels. They have to get new car parts ready in time for a race in three or four days. They only have the time to design and model it. The rest of the time goes into building it. Those are some of the most interesting examples.

This segment is part 2 in the series : Thought Leaders in Big Data: Franz Aman, Chief Marketing Officer, Silicon Graphics
1 2 3 4 5