Take that over to the security world and ask yourself which parts of that standard definition are applicable. Having quantities of data that are so large that you can’t deal with them through traditional means is really not going to apply to traditional on-premise systems because in a single-customer facility you’re really not going to generate data that’s that large. Cloud companies, though, since they span hundreds, thousands, tens of thousands of customers and aggregate all that data together, are going to have large enough data aggregations that these techniques that are developed for Big Data would actually be relevant to doing analyses across customers — norms, averages, trends, things like that. That’s the first thing. The second thing is there is perhaps a fine line between the term Big Data and an older, more well-traveled term called business intelligence
. Business intelligence is in the business usually of using traditional databases, often going up against large but not super large datasets, and providing information that’s been extracted from that dataset.
As I say, there is a fine line between business intelligence and Big Data, and a lot of people that used to call themselves business intelligence companies are now calling themselves Big Data companies because it’s more popular. But in the security world, in terms of specific applications, if you look for example at the flood of data that comes out of cameras or in our case the flood of data that’s coming out of access-control systems and all flowing back to one place, that makes for an interesting opportunity to do work against very large datasets and perhaps give people information that they couldn’t get, just based on the data generated inside of their own organization.
It’s been said ‘cash may be king, but data is golden.’ Is that a fair statement?
If the integrators can — and I agree with that statement — deliver systems, they’re not going to create the systems that do this because that’s not what they do. Generally speaking, they’re not software programmers or data analysts, but they can purchase systems from people who make them and put them at the customers’ disposal.
The thing they can do is nobody has enough security people to understand all of the data that’s coming back from their systems. These business intelligence or Big Data systems can help to summarize information into a more useful form. When you start looking at data being golden, people are usually talking about using this information for other operational purposes. For example, in retail the information that comes in off of cameras is used to measure customer behaviors and to change characteristics of the store or of the selling proposition based on that information, and now you’re actually creating more business value.
The hope is that through Big Data techniques or business intelligence techniques people can learn something from their security data that not only is relevant to the security mission, but also more broadly relevant and profitable, creating a return on the investment for the business as a whole.
Do you see Big Data as a significant revenue generator for integrators?
As I was trying to politely say earlier, they might be able to help install something like this but they’re not going to create it, and they’re not going to decide what the measures or metrics need to be. We’re exploring this area with customers right now and the metrics for one customer are different than the metrics for another customer. For example, a retail customer that has 500 stores is going to be interested in extracting very different kinds of analytics and comparisons and data histories and things like that than somebody who is in the medical industry and has a totally different set of interests. That’s something that our integrators aren’t really equipped to have a dialog about or to help shape. They can sell our solution and we can do that, but they’re really kind of along for the ride, for lack of a better term.
Will Big Data applications include situational awareness in real-time?
It could. One of the examples I have in that category, we have a service that aggregates all of the camera motion-detection events that are happening at our customers’ facilities, so that’s tens of thousands per minute just streaming in. A lot of that stuff is only interesting when there’s an exception because you might say I can see there was some motion at this time, in this place; I’m going to look at that piece of video.
That still leaves you with a lot to look at, and it’s really only for the purpose of looking at a piece of video. If you instead say well, video cameras are for recording video, but you can also look at them as a sensor, and say the only thing I’m interested is whether there’s motion here or not. For example, if a certain part of my business is normally trafficked on average with let’s say 50 motion events per minute during a certain time of day, and I’ve got a year’s worth of history showing me that, and suddenly that part of business is trafficked by three times, four times, five times that much motion, there’s probably something going on there.
These kinds of analytics allow for that kind of historical perspective and averaging and accounting for days of week and times of year to filter out the noise from the information and present you with events or patterns or trends that truly are an exception, and do represent a departure from the norm. That’s an example of how you can apply some of these techniques to regular security activities.
Page 2 of 3 pages <