 Live from Las Vegas, Nevada. Extracting the signal from the noise. It's theCUBE, covering IBM Edge 2015. Brought to you by IBM. Okay, welcome back everyone. We are here live in Las Vegas for IBM Edge 2015. This is Silicon Angles theCUBE, our flagship program. Where we go out to the events and extract the signal from the noise. I'm John Furrier. My co-host is this segment, David Floyer, chief analyst at Wikibon. Our next guest, Matthew Schumann. Enterprise Luzard at Melanox. Welcome to theCUBE. Thank you. So we're going to get down and dirty in this segment. David Floyer is in the house. So talk about open power. Give us the take on that. Open power. I mean, Melanox is really excited to be, invited to be part of the Open Power. Of course, consortium. I mean, it's a really exciting thing with Open Power Foundation. The idea of innovation, collaboration, bringing the idea of innovation, bringing the change in market differentiation to the market, giving people choice. Where from a solid enterprise grade compute platform with network and other partners like Google and NVIDIA with the compute, it's really just a really great, a great, a great, great industry to just go forward and move in a different direction. I mean, it's got some legs to it. I mean, when Doug Balog was on theCUBE two years ago when they announced it, or less than two years ago, you know, he was, you know, all pumped up, the marketing spiel. But they didn't really know how it would turn out. It really ended up being. There was five of us at the start. Melanox was one of the first five. 127 people now. Share with us the early days because it was a good bet. A lot of smart people making the good bets. But at that time, they were just getting out of their servers, their server commodity servers. It was a time of great change. Open compute was exploding. A lot of architectural shifts. What was it like in the, in that? In the early and interesting days, it was, I mean, there was a lot of energy from IBM. That was the thing which really, Melanox really saw. We got involved in projects like the TurboLamp project where they were bringing together innovation partners from across the industry to get together and build solutions. Not just building a horizontal solution but a vertical solution where all the partners were actually optimizing at their own horizontal layer but also optimizing at the vertical layer as well so that you had tight integration so you could scale and deliver solutions. So, excitement was really the key thing around Open Power Foundation at the beginning because with the scale out platforms they delivered, they delivered a really interesting platform, a lot of compute power, a lot of bandwidth that you could access from the memory that could allow you to do a lot of exciting things. Things that have been brought out today into the genomics work that Melanox and IBM are exploring today together. So, if I can ask a question on that, where are you finding the value of doing this? What sort of workloads? If I look at the marketplace at the moment, there's a lot going on with low latency flash for example and PCI switching and memory switching. There's a whole lot going on. How is Melanox adding value to the applications? What we do, Melanox's role is we really extend the efficiency that IBM's created with the Power 8 compute platform and extend it across the network. So, we deliver obviously low latency, it's our heritage, where we come from with the Invinivan space. It's work that's been leveraged with NVIDIA and IBM around the Coral project and so that's really the thing. So, we're seeing that we're working, Melanox as a general company working with the flash and the storage industry to bring new solutions to market where they are leveraging that low latency because it's no wait time, it's a waitless environment. Everything has to be timely. It's a now network. It's a now environment. Everybody expects something now. You don't want to pick up your phone and play with your apps and have things happening slowly. You want it now. So, when you've got the bandwidth and the process and the data management and data movement capabilities of Power, Power 8, you need the network to match that and be able to do that. You can invest in the best GPUs, the best servers, the best storage, but if you don't invest in the network and have the right network, you're diluting that investment that you've made and all those things. So, infrastructure matters. I mean, that's a tagline from IBM, but I mean for us... Everyone's been trying to crack this code for years, the network efficiency of compute on the network is a huge opportunity. Is this kind of where it's connecting? This is my personal play from being in the industry and being working for IBM and working in a number of different networking companies. This is where we've taken that. With Cappy, we have taken the CPU and with the FPGA and you've made that extra core on the system. Now we can actually take it and bring it right down on the network. And in Finneban, you've been talking about low latency. Finneban's always been talking about low latency. Cappy's reducing that by another magnitude. So where are we down to now? So can we get to switching below 100 microseconds? 100, so I mean, our new technologies are around 220 nanoseconds. 220 nanoseconds. Nanoseconds. Right. That's not end-to-end, but... That's port-to-port. Port-to-port. End-to-end latency is sort of around 300 to 500 nanosecond. And that's through the card up, both ends of the card, so it's a very, very low latency. And that's with our new 100 gig EDR technologies. So that's game-changing, if you can put that in a flash memory to memory. That's the thing. I mean, now a computer that is adjacent to that is now almost in the same box. That's moved so much closer together. If you look at what's happening with Coral, with the NVLink work that NVIDIA is doing with IBM around Power 9, the work we're doing with Power 9, we are bringing those as closely together as we possibly can. So using the Cappy interface so that we can get that ultimate thing of very, very low latency. We can build that scale-out environment for the national labs as they go forward. What's the challenge for others? Now, you have an inner circle of the formation. You've got 127. Where's the impact? Where's the meat and the bone going to come out in the ecosystem? Are you starting to see it now and can you share what's coming out of the oven if you will? So Open Power's interesting in the sense that there's different things. We've done collaborations with software vendors like Ubuntu, MariaDB, ZenPHP, we're a networking company. It's not usually that you have networking guys saying, hey, I'm talking about a PHP server here. I'm talking about a database here. That's the thing. We're working in the NFE space with Altaira, who's an FPGA company. So obviously FPGAs and high-speed networking, they're sort of similar. But then you're working with something like Xwind who has the Data Plane Development Kit. Data Plane Development Kit was developed on the Intel platform by Intel. They actually created it. We've moved that onto Power and we're delivering comparable performance for Power as we can on Intel and we've got more CPUs, better CPUs, stronger CPUs. So it's a really exciting thing. So that's sort of the innovation you're seeing and you're seeing people that getting together who you wouldn't expect to get together. But when you think about it in sort of a small sense, it doesn't make sense. When you think of it in the big sense, it makes hey, they have got similar interests. But the thing about OpenPower is it's innovation. It's exploration. It's integration. We look at the solution. Sorry. There's very interesting about the database, isn't it? Because MariaDB were one of the early ones with NVMe and Atomic Rights and all of that sort of thing. Is that the sort of area that you're looking at with them? Well, when we looked at with MariaDB and the Turbo Lambda stack, we just wanted to do a sort of a crawl-walk run. So they locked us in a room for two days for July 15th last year and then they said, oh, by the way, you've got to have something demonstrable by the 16th of October. So I had five companies that had IBM in the mix and then said, okay, we want to deliver something. So we had to do some stuff. So we did basic blocking and tackling TCP IP and those things. Now we've got that up and running. We're demonstrating it again on the floor today. Today on the demo floor on the solution floor. And then, so where does it go? I mean, do we add storage opportunities there with Elastic Storage Server? Do we add NVMe? Do we add Reader's Flash? No SQL to the solution? There's all sorts of different ways that we can go and that's sort of just, it's just a continued evolution and integration and exploration. A lot of people talk about software, threads involved, all the stuff's happening, these new innovations. What is the innovation on the software side you see coming here? I mean, I was interviewing a customer last week and said, they were not bullish on NFV. Now that's their personal opinion. Maybe they had a little, how they integrate. But open power is about integration. It's about having flexibility. Where's the software role in all this? And what's your vision as this evolves? Because there are going to be diverse opinions out there about things like NFV or other pieces of the puzzle. NFV's diverse opinions is that. I mean, that's very clear. But from a solution standpoint, we want to have the software will also be determined. Yeah, I mean, I'm a solution architect at the end of the day, right? And so I start at the network and then I work up the stack and then I'm thinking and where is the right, what is the problem? Rather than, is NFV the solution? Or is it the problem? And software. It's just a coin. It's like, does Tom Brady deserve to be suspended for four days? Half the people think he does, half they don't. Right, so when you're coming to a solution with open power foundation is what are the problems that we're trying to solve, right? We saw that identified that we needed to deliver a better lamp stack. We're seeing that we can deliver a better solution in the NFV space. We're seeing that we can do interesting things in the genomic space. We're seeing we can do interesting things in the, for coral, for the national labs. So where it's coming from the software point of view is what are the problems that you're trying to solve? We work in the big data space with IBM around the IBM data engine for analytics, the elastic storage server, the spectrum scale. What are the problems that you're saying? Because those are all infrastructure. But what am I trying to do? Am I trying to decide that I'm going to trawl all the government stock SEC filings for seeing who's doing the investment in buildings and trawl all the building permits and come down and say, if I'm a company that builds chemicals, am I going to build it in this city because this company here is filing here and doing that, which is solving a problem with all that infrastructure. The open power community has those people that can help you solve those problems. And really it's just kidding. That's the beautiful thing about the ecosystem. It's a solution architect's dream. Think about it. I mean, that's what you're talking about. I mean, I've got a broad range of platforms I can choose from. I've got chip manufacturers. I've got FPGA manufacturers. I've got networking. We're now getting application and management companies joining the solution. So it's an amazing thing. As an architect, you're right. It's an absolute dream. So say, what's your dream of about an application or application of this technology that you really think will make a difference to the world? Well, one of the things I sort of was talking about the other day with some people was you take the big data solution, the analytics, right? You run it at retail and you've got a retail environment, right? You now want to do store purchasing. You talk about the model, the examples they talk about where you're going in. You're clicking on the website. You're getting suggestions. You've been building a house. So you now start looking at paint and then, okay, ladders and rollers come up and all that stuff happens in real time, right? So it's not Nestrib 1 or it's not one solution. It's taking those components and being able to reshape them to do various different things. Now, that big analytic solution that you're using to extract that data for contextual delivery of solution e-commerce transactions can also be reshaped to deliver that building scenario where you say you're trolling the SEC records, you're trolling the building permits and finding out where those things intersect. So it's not just that what you've got and as you said graphically, it's a solution architect's dream. OpenPower brings that to the solution architect is that you can go and talk to TyAnne, you can go talk to Sushon, you can talk to the various different companies in the group to get what you're wanting to build. You've got partners like Google and Rackspace building OpenPower solutions. It's just... It's a lot of choice but at the same time the benefits... Dave and I talked about this all the time, Dave Vellante and Dave Fleurer. Competitive advantage is lock-in. I say lock-in metaphorically speaking. Everyone needs to have an ability to make money. A lock-in was the old way. Now you have choice, so the lock-in becomes the platform. The ecosystem is an opportunity. I should say lock-in. Competitive advantage. The competitive advantage or the differentiation is the fact of the innovation. The enthusiasm and the energy around innovation. Look at all the partners that are in OpenPower. That's what they bring to the marketplace. I mean the founding members, Google. I think innovation in Google are almost synonymous these days. NVIDIA, Melanox. Micro-Intein. Differentiation is the new lock-in. In an open choice way. It's a very different way of looking at the world. It's not like you can buy from this. This is a broad consortium of people. A group foundation that is really great. It's the ethos of open source. Matthew, thanks so much for sharing with us. Get down and dirty here in the queue talking about Melanox, all the greaties of OpenPower. We'll be right back after the short break.