 theCubeSiliconAngle.com, wikibon.org, exclusive coverage of HP Moonshot in New York City, special event as they announce their next generation major inflection point seminal moment in the server data center market, introducing some killer technology, has some amazing stats, we're here to break it down. I'm John Furrier, the founder of Silicon Angle, joining my co-host. I'm Dave Vellante from wikibon.org. Laurent Clerk is in the house on theCUBE. He's with CGG, he's the vice president of IT at CGG. Welcome to theCUBE. Well, thanks for having me. Oh, you're welcome. Tell us about CGG. It's not a company that's a household name, although it's quite an interesting background. Yes, indeed, CGG is a geophysical contractor, so what we do basically is we acquire, process, and deliver images of the underground. Our clients, typically, are the oil and gas companies, and what we do with that, obviously, is they then look at it and figure out what they want to do in terms of drilling or finding oil and gas. Yeah, so talk about how you do that today from an IT standpoint, what your infrastructure looks like, your applications, and we'll get into sort of where Moonshot fits in, all this. Yes, definitely. What we get from the field, and we do acquire data ourselves, that's one of the activities that we have, what we get from the field is a fairly large amount of data that if you look at it without doing anything to it, it is next to impossible to interpret. So there is a step where you take that data and then you process it and you refine it to the point where you get an image of the underground that you can actually understand and use, and that requires software, obviously, and we do design our own software, but that also requires a fair amount of processing capacity, both IO-wise, we have a lot of data, a lot of throughput. Typical data set is in the range of 10 to 100 terabytes, and then a lot of processing that goes with it. Typically, our type of activity is very, very similar to what you find in the other high-performance computing environments. Right, so that makes it well-suited for something like Moonshot. So maybe you could describe your current infrastructure today and then, as I say, we could talk about where Moonshot fits. All right, well, historically, we went from a very large system such as the craze and even convexism, then we moved to mid-range servers, and then finally, at some point in early 2000, we moved to clusters, just like pretty much everybody else. And the clusters we have are clusters that are designed around processors that are capable of doing a lot of floating-point operations. And the network that feeds the data to those nodes is designed around, typically, a 10 gig today, 40 gig for the connection between the switches. So it's a pretty large set of these. We have thousands of servers, literally. We have a number of data sets, data sites. We have probably an excess of 40 sites today around the world doing processing. Not all the same size, but... And they are basically designed to do number crunching of large data sets. And are the servers fairly underutilized? Is that fair characterization or not necessarily? What do you mean? I say, you know, the processors, are they underutilized in terms of... Well, no, not really. What we try to do since we own the software as well is we try to develop the software in a way that it's going to work well for the hardware that we can acquire. And actually, that's where the moonshot is a significant change for us. Yeah, I also explained that. Well, typically up until now, you were bound by the design of the hardware systems as they are. There is some flexibility, obviously. You can change some components. You have a wide range of CPUs. But basically, it's a box with a CPU and then some cards connected to the box. And you had to design your software around those constraints. With a moonshot, you now have a box of components that you can mix and match and you can then look at your software in a different way and say, well, what would work well with what in terms of hardware and then acquire just the bits and pieces that you need and make them work together well? Yeah, okay. In the natural resource area, I was asked a question last week. I was on a panel leading a disruptive technology and the energy question came up and we talked about, say, big data, convergent infrastructure for lead infrastructure. And it's not an area that we cover deeply, but I wanted to ask you a question around that world that you live in in terms of what you need from compute and software at the scale. Kind of give us a picture of the kind of scale that you do and what does it do for your data center? Is it a lot of big data? Just take us through the use case real quick and then how moonshot fits in. All right, well, typically today, the infrastructure that we've developed and deployed in our sites is well into the tens of petabyte of storage and we're well into the, well, terabytes, teraflops of processing capacity. So these are pretty large sites. They are not as large as what you would find in the very large web server forms, for example, but they have the next step behind in terms of size and complexity. But they are not so much dedicated to transaction work like most of the very large sites. They are really dedicated to number crunching and we funnel through that architecture a massive amount of data that goes through a number of different processing steps that are all CPU intensive. So we have all the complexity of not just moving data through that but also processing it in terms of scientific processing. So really we have all the problems at the same time and that is why a flexible architecture is going to enable us to have just enough of what works well for each and every cell. The data pipelining and the compute, obviously, big, big thing. What about at the application layer latency and any kind of SL requirements around latency? Well, a lot of what we do is steal batch jobs because their nature is such that they are very long. So all of the batch activity what we want is make sure that it actually works and works well and is stable. There's not so much a latency issue in terms of how quickly you get the data. But it's raw machines that you need. So in this case, this is nice for you because you get 1800 servers on a chassis. It's perfect. And how about the power? What's your take on the power? You've got a chance to kick the tires on this product. Explain to the folks the order of magnitude difference between non-moon shot and a moonshot box. Well, it's still very early in that process. So I'm not going to be able to give you numbers like HP could. However, what we're seeing definitely in terms of power is that all our processing centers are power bound just like everybody else is. Be it because regulations force us to take a look at it or just the financials of the equation. It's expensive to buy all that electricity. So we have to be very careful about the way we use the electricity and definitely anything that is going to save electricity for us is not just going to enable us to get a better price performance point, but it's also going to enable us to have more equipment to produce a better picture for our client. Yeah, more Protestant. Yeah, because at the end of the day, all about our clients want to be able to see something. So what we need to do is produce the best quality data that we can so that they can then see what they want to see in there. How about just in terms of your supplier relationships? Obviously, you're an HP customer. Are you primarily an HP shop? Do you have a diversity of suppliers? We have a diversity of suppliers. We do work with the suppliers depending on the technology that they bring to the table. And HP with the moonshot is bringing a technology that's definitely very interesting. So it would be, and we couldn't afford to miss that opportunity. So we're working with them on that, definitely. Yeah, okay, so it's not a case of you're okay, we're an HP shop, anything HP comes out with, we buy, it's more opportunistic with regard to the technology that's coming. And the applications that you've cited, they're HPC-like, if I can say that, right? So they lend themselves well to this type of architecture. Do you see things like moonshot applying to more traditional IT? Well, the beauty of moonshot, as you've heard with the previous people that you've talked to, is that it is not so much, I'm getting the CPU, I'm getting this memory, I'm getting something where I can put pretty much anything I want. And if I don't find it, I suspect that you can go to HP and say, hey, there's a market for that, can I have that particular product in there? And so, yes, definitely, I see that box being pretty universal in terms of usage. Which for us makes it interesting because we're a niche market, so it's very unusual that companies will spend the money to develop something specifically for a niche market. So if we can piggyback on a much broader market to get the things we need, that's even better for us. You're obviously, at least in part, a data-driven organization. Are you having conversations internally about different ways of monetizing that data and how is that changing over time? Well, you know, that is not an area that I really feel comfortable discussing. There is more sales and marketing and I don't normally participate in those conversations. So that's interesting because to the extent that sales and marketing are gonna drive that and they require a technology component, would they go off and just sort of do that themselves or would they not consult with you? Or are these secrets that you can't share? That's possible too. Well, there is a bit of, we like to keep the edge that we have to ourselves in that discussion, but basically what we do as an IT team for an organization like CGG is we view ourselves as a service provider internally. So what we do is we listen to the problem that they have and then we come with solutions for them and we tell them here is what we found, here is what we think works best and then it's up to the business managers to decide what makes sense for them in terms of what they sell. From an application development standpoint, how do you see that changing the processes or the way in which they interact with the infrastructure? How do you see that changing over time as a result of things like Moonshot? Well, if you look at the Moonshot as a collection of different parts that you can provision depending on your needs, of course you're gonna have to look at the architecture of the software itself and figure out how to recompose it to take advantage of that. And that could be a difficulty because every time you move to more complex hardware, multi-cores or things that are not obvious, not just a bump in the CPU capacity, then you have to ask yourself what software architecture do I need for that? So there will be some work to be done around that, definitely. Final question, we're getting tight on time here, is what's your overall take of the announcement here at Moonshot and what would you share with the audience around what it means to the industry as folks are just, I mean, you guys are ahead of the curve but there are some a lot of corporate IT departments putting their toe in the waters. They see cloud as in roadmap, they know there's issues and all that good stuff, they're transforming their businesses. What would you share with those folks out there to explain kind of this impact of what Moonshot can do for them? Well, the fact that a well-established vendor like HP comes up with that sort of architecture should enable a lot of people to actually, as you say, put their toes in the water and figure out if that will work for them. And it's pretty much a risk-free operation here. You can work with HP and figure out what you can do with a box and if it works for you, then that's a solution you need. And you got the savings on the energy size, that's a good footprint issue. That's what was good. Okay, we're here inside the Cube New York City Special Edition, breaking it down, we have a lot more coming, stay with us all day. And then Dave and I will also do a wrap up towards the end of the day and give our editorial spin on what's happening and we'll take it from there. So have a great day and stay with us here at HP's Moonshot on SiliconANGLE.com.