 Live from Washington D.C., it's theCUBE, covering AWS Public Sector Summit 2017. Brought to you by Amazon Web Services and its partner ecosystem. We're welcome inside the convention center here in Washington D.C. You're looking at many of the attendees of the AWS Public Sector Summit 2017. We're coming to you live from our nation's capital. Several thousand people on hand here for this two-day event, three-day event. We're here for two days. John Wall's along with John Furrier. John, good to see you again, sir. Thank you. We're joined by Jay Littlepage, who is the VP of Infrastructure and Operations at Digital Globe. Jay, thank you for being with us here on theCUBE. Good to have you. First off, I mean, you're a company, high resolution earth imagery satellite stuff out of this world business, right? Tell our viewers a little bit just about what you do and the magnitude of what you and obviously the environmental implications of that or the defense, safety, security, all those realms. Okay, well, stop me when I've said too much because I get pretty excited about this. We work for a very cool company. We've been taking earth imagery since 1999 when our first satellite went up in the sky. And as we've increased our capabilities with our constellation, our latest satellite went up last November. We're flying basically a giant camera that we can fly like a drone. So, and when I say giant camera, it's about the size of a school bus and the lens is about the size of the front of a school bus and we can take imagery from 700 miles up in space and resolve a pixel about the size of a laptop. So that gives us an incredible amount of capability and the flying like a drone, besides just being really cool and geeky, we can swing the lens from basically Kansas City to here in Washington in 15 seconds and take a shot. And so when world events happen, when an earthquake happens, they're generally not scheduled events. We don't have to have the satellite right above the point where there's something going on on the ground. We can take a shot from an angle of 1,000 miles away and with compute power and good algorithms, we can basically resolve the curvature of the earth and it looks like we're right overhead and we're getting imagery out immediately to first responders, to governmental agencies so they can respond very quickly to a disaster happening, save lives. So obviously the ramifications are endless, almost. Yes. Where you can, all that data, I mean we can't even imagine, I mean when you talk about storage, so that's certainly a complexity and then they're making it useful to all these different sectors. Without getting too simple, I mean, how do you manage that? Well, you know, it's a big trade off because ideally if storage was free, all of our imagery and its highest consumable form would be available all the time to everybody. Each image, high resolution image might be 35 gig by itself. So you think of that long of flying a constellation, we've got 100 petabytes of imagery. That's too much, it's too expensive to have online all of the time and so we have to balance what's going to be relevant and useful to people versus cost. You know, a lot of the imagery kind of goes through a cycle where it's interesting until it's not and it starts to age off. The thing about the planet though is we never know what's going to happen and when something that aged off is going to be relevant again and so the balance for my team is really making sure we're kind of hitting the sweet spot on there. The imagery that is relevant is readily accessible and the imagery that's not is in its cheapest form factor possible which for us is it's compressed and it's in some sort of archival storage which for us now that we've used the snowmobile is glacier. Jay, I want to ask you a few thoughts. I want you to talk about Digital Globe before that some context. This weekend I was hanging out with my friends and Santa Cruz, the kids were surfing. He's a big drone guy. He used to work for GoPro and she used to fly the drones and I go, hey, how's it going with the drones? Kind of boring, here's a great photo I created but after a while it just became like Google Earth and it got boring. Kind of pointing out that he wanted more and we've got virtual reality, augmented reality, experiences coming to users. That puts imagery, place imagery, the globe, pictures, places and things is what you guys do. So that's not going away anytime soon. So talk about your business, what you guys do, some of the things that you do, your business model and how that's changing and how Amazon here in the public sector is changing that. Well, that's a fantastic question and our business is changing pretty rapidly. We have all that imagery and it's beautiful imagery but increasingly there's so much of it and so many of the use cases aren't about human eyeballs staring at pixels, they're about algorithms extracting information from the pixels and increasingly from either sort of the breadth of pixels instead of just looking at a small area you can look around it and see what's happening around it and use that as signaling information or you can go deep into an archive and see the same location on the planet over and over over years and see the changes that have happened in terms of time frame. So increasingly our market is about extracting information, extracting insights from the imagery more so than it is the imagery itself and so that's driving an analytics business for us and it's also driving a services business for us which is particularly important in the public sector to actually use that for different purposes. You can imagine the creativity involved in developers out there watching or even thinking about using satellite imagery in context of other data. Remember during the web 2.0 craze in the earlier in the last decade you saw mashups of API with Google Maps. Oh yeah, put a little pin, then a mobile came but now you're seeing mashups go on with other data. And I've heard stats that Uber for instance remaps New York City every five days with all that GPS data of the cars which are basically sensors. So you can almost imagine the alchemy, the convergence of data. This is exciting for you guys, I can imagine. You wouldn't share with us anecdotally or statistically what you're seeing, how this is playing out. Well you know some of our biggest commercial customers of our products now are location-based services. So Uber's using our imagery because the size of the aperture of our lens means we have great resolution and so they've been consuming that and consuming our machine learning algorithms to basically understand where traffic is and where people are so that they can refine on an ongoing basis where the best pickup and drop-off locations are. That really drives their business. Facebook's using the imagery to basically help build out the internet. They want to move into places on the planet where internet doesn't exist. Well in order to really understand that they need to understand where to build, how to build, how many people are there and you can actually extract all that from imagery by going in in detail and mapping roof shapes and roof sizes and from there extracting pretty accurate estimates of how many people live in a particular area and that's driving their project which is ultimately going to drive access for places on the planet. So intelligence and software to look at imagery. I mean we hear at Amazon recognitions their big product for facial recognition among other pictures. But that's what they're getting at, this notion of actually extracting that data. Well you think about it, once the data is available, once our imagery is available, then the sky's the limit. We have a certain set of algorithms that we apply to help different industries, to look at rooftops, to look at water extraction so after a hurricane we can actually see how the coverage has changed. But you look at a Facebook and they're applying their own algorithms. We don't force our algorithms to be used, we provide the information, we provide the data, companies can bring their own algorithms and then it's all about what can you learn and then what can you do about it and it's amazing. So here's the question in the whole Polygot conversation of multiple languages that people speak as translated to the tech industry. And interdisciplinary forces are in play. Data science, coding, cognitive, now machine learning. So the question is for you is that, okay, as this stuff comes together, do you speak DevOps? It's kind of a word we hear people say, is that like Russian or is that like English? DevOps is a cloud language mindset. And so that brings up the question of, are you guys friendly to developers? And because people want to have microservices. If I'm a developer, I'm like, hey, I want those maps. How do I get them? Can I buy them as a service? Can I, are they loaded on Amazon? How do I engage with Digital Globe whether I'm a developer or a company? Well, you think about what you just said and the customers I just talked about, they're not geospatial customers. They're not staffed with people that are PhDs in extracting information. They're developers that are working for high tech companies that have a problem they want to solve. They're running mobile apps or doing some cool database work or anything. So we're using, providing the raw imagery in the algorithms to very tried and true systems where people can plug into work benches and build artificial intelligence without necessarily being experts in that. And as a case in point, my team is an IT team. We've got a part of the organization that is all staffed with PhDs. They're the ones that are driving our global. PhD is a service. Well, kind of. I mean, you think about it, they're off, they're driving the leading edge for these solutions for our customers, but I've got an IT team and I've got this problem with all this data that we talked about earlier. Well, how am I actually going to manage that? I'm going to be pulling in all sorts of different sources of data and I'm going to be applying machine learning using IT guys that are PhDs to actually do that. And I'm not going to send them to graduate school. They're going to be using standard APIs and they're going to be applying fairly generic algorithms. And, you know, the- So is that your model just API it out? Is there other- Well, I think the real key is, you know, the API makes it accessible, but a machine learning algorithm is only as good as it's training. So the more it's used, the more it refines itself, the better our algorithm gets. And so that is going to be the type of thing that the IT developer, the infrastructure engineer of the future becomes. And I've already basically in the last couple of years as we started this journey at AWS, 28% of my staff now, same size staff, but they're software developers now. So, I mean, we'll take this to the government side. We've talked a lot about commercial use, but to the government side, I'm thinking about FEMA disaster response, maybe core of engineers, you know, bridge construction, road construction, postline management, are all those kind of applications that we see on the dot gov side? There are all things that you see that can be done on the dot gov side, but we're doing them all in the commercial environment, the US East region for AWS. And I think that's actually a really important distinction. And it's something that I think more and more of the government agencies are starting to see. We do a lot of work for one particular government agency and half for years, but 99. something percent of our imagery is commercial unclassified. And it's available for the purposes that our customer use it for, but they're also available for all those other customers I talked about. And more and more of what we do, we are doing on the completely open but secure commercial environment because it's ubiquitous for our customers. Not all of our customers do that type of work. They don't need to comply with those rigid standards. It's generally where all AWS products that are released are released to with the other environments lagging. They probably don't want me saying that on TV, but I just did. And it's cheaper. We're a commercial company that does public sector work. We have to make a profit. And the best way to do that is to put your environment in a place where if you're going to repeat an operation, like pull an image out of glacier and build it into something that is consumable by either a human or an algorithm and put it back. If you're going to do something like that a million times, you want to do it really inexpensively. And so that's where- So Amazon's ethos is lower prices, make things faster. Jeff Bezos' ethos shipping products on these books in the old days. Now they're shipping code and making lower latency systems. So you guys are a big customer. What are the big implementation features that you have with AWS? And then the second part of the question is, are you worried about locking? At some point you're so big, the hours are going to be so massive and you're paying so much cash. Should you build your own? That's the big debate. Do you go private cloud? Do you stay in the public? Thoughts on those two questions? Well, we have both. Right now we're running a 15-year-old system. Which is where we create the imagery that comes off the satellites and it goes into a tape archive. Last year at re-invent- Hey, tape's supposed to be dead. Tape? Tape, we'll die someday. It's going to die in digital globe really soon, but at the re-invent conference last year, AWS rolled out a semi-truck. Well, the real semi-truck was in our parking lot getting loaded with all those tapes. And it's at- Did you actually use the semi-truck from the service? We were the first customer ever, I believe, of the snowmobile. And so it takes a lot of time and effort to move to basically ingest 12,000 LTO5 tapes, loaded onto a semi and send it off. You know, that represents every image ever taken by DG in the history of our company. And it's now in AWS. So to your second part of your question, we're pretty committed now. We feel good about- And you're okay with that? We're okay with that for a couple of reasons. One is, I'm not constraining the business. AWS is cheaper. It will be even cheaper for us as we learn how to pull all the levers and turn all the dials in this environment. But you know, you think about, we ran a particular job last year for a customer that consumed 750,000 compute hours in 22 days. We couldn't have done that in our data center. We would have said no. And so I would have been constrained to my business. We can't do it. Or we could do it, come back, the answer will be here in, oh, six months. So and time is of the essence in situations like that. So we're comfortable with it for our business. We're also comfortable with it because increasingly, that's where our customers already are. We were creating something in our current environment and shipping it to Amazon anyway. Yeah, so our movie about you, would Jim Carey? Yes, man? You can say yes to everything now with Amazon. Okay, but this is a good point. I'm just joking aside. This is interesting because we have this debate all the time. When is the cloud prohibitive? In this case, your business model based on the fact that the variables spend that you turn up your compute is based upon cadence of the business. That's exactly right. You know, the thing that's really changed for the business with this model is, historically, IT has been a cost center. And moving into Amazon, I manage our storage and I pay for our storage because it's a shared asset. It's something that is for the common good. The business units, the different product managers in our business now have the dial for what they spend on the compute and everything else. And so if they want to go to market really rapidly, they can. If they want to spin it up rapidly, they can. If they want to turn it down, they can. And it's not a fixed investment. So it allows business philosophy that we've never had before. Jay, I know we're getting tight on time but I do want to ask you one question. And I did not know that you were the first Snowmobile customer. So that's good trivia that have on the cube and great to have you. So while we got you here, with the Snowmobile, being the first customer of AWS Snowmobile when they rolled out at Amazon re-invent, we covered on SiliconANGLE. What's been your experience? Why did you jump on that? And how has your experience been? Share some color onto that whole process. Okay. It's been an iterative learning process for both us and for Amazon. We were sitting on all this imagery. We knew we wanted to get it to AWS. We started using the Snowballs almost a year and a half ago but moving 100 petabytes, 80 terabytes at a time, it's like using a spoon to move a haystack. So when they, Amazon approached us about knowing the challenge we had about moving it all at once, I initially thought they were kidding. And then I realized it was Amazon. They don't kid about things like this. And so we jumped on pretty early and worked with them on this. So you got a blown away, like what? What's the catch? Really? A truck? Really? Yeah, but really. So it's secure to possibly be, we're taking out the internet and all the different variables in that, including a lot of cost and bandwidth constraints and basically parking it next to our data. And it's basically a big NFS file system and we loaded data onto it. The constraint for us being basically that tape library with 10,000 miles of movement on the tape heads, we had to balance between loading the snowmobile and basically responding to our regular customers. We pull four million images a year off that tape library. And so loading every single image we've ever created onto the snowmobile at the same time was a technical challenge on our side more so than Amazon's side. So we had to find like that sweet spot and then just let it run. And now it's operational. The snowmobile is gone. AWS has got it. They're adjusting it right now into the West region. And we're looking forward to being able to just go wild with that data. What's the theorem? Snowmobiles, we got SIMIs, we have satellites, we have it all, right? We have it all, yeah. It's massive, obviously, but impressive what you're doing with it. So congratulations on that front and thank you again for being with us here on theCUBE. Thanks for having me. You bet, we continue our coverage here from Washington DC, live on theCUBE. SiliconANGLE TV continues right after this.