 Live from the San Jose Mechanery Convention Center, it's theCUBE at Open Compute Project US Summit 2015. Okay, welcome back everyone. This is SiliconANGLE Media's theCUBE, our flagship program. We go out for the events, extract the synth from the noise. I'm John Furrier with my co-host, Jeff Frick who's a general manager of our CUBE business. Jeff, we've got Aaron Sullivan, Distinguished Engineer of Rack Space. Welcome to theCUBE. Thanks, good to be here. Open power, open compute, power's huge. But let's get back, let's get into the conversation. Tell us what's going on with you this year and your perspective, looking back from last year, what's going on? I mean, bigger, smaller, or robust? Yeah, lots of stuff, lots of advancement since last year. When we talked last year, we were just launching our Open Compute systems onto most of the apps that are running Open Compute at Rack Space today. We've launched eight applications at Rack Space supporting Open Compute. We run it in our back office for some of our business intelligence and analytics platforms. We run it in front office for a variety of our OpenStack cloud services. We built our on-metal cloud service, which we're very proud to launch a very distinctive offer last summer. We built that on Open Compute. It was an exclusive sort of thing. Like we couldn't have done it without Open Compute. Like in which way? Like what was there that wasn't, like gave you the ability to do that? So we needed to design a service that gave people bare-metal cloud at hypervisor cloud provisioning and security-type speeds and feature sets, right? So that's a tough thing to do. You gain a lot from that hypervisor in terms of security and speed and ease of provisioning. And when we took that out, we had to do things in the firmware of the server to replace some of those core functions and safety mechanisms that we needed. And the fact that we had Open Compute and the engagement with all the different engineers in the community to work with helped us get there. But we also kind of hit a wall at that point. We sort of realized the limits of where we could take open on the platform we were at at that point in time. And we think the cloud is going to continue to evolve and need to get vastly more efficient than it is today. In order to do that, we need to let developers like the guys that work on our on-metal platform and so forth, get deeper down into those systems, right? And they're all developers and hackers, right? Those sorts learn best when you just open the whole thing up and let them- Yeah, but you guys built a great system on Open Source as well. And here you have a community that's bringing Open Source. We were just talking to the GM at Microsoft. They brought large-scale management templates and reference architectures to the table. That's real IP. That's real IP. That's real IP. And we need more of that, right? That's what we need. We need IP exposed at that layer and deeper on all the systems that Open Compute delivers today. All right, explain to the folks out there what this means. Cause it's like, to me, I look at it like Christmas. You know, the presents are all there. You know, Microsoft comes with all this IP and the community around. As an engineer, what's it like? I mean, are your eyes popping out of your head? Are you a kid in the candy store? What's happened? What's the psychology of the in the trenches developer? No, I think when you first get in your eyes are popping out of your head and you're really excited. And then after a while, this is the new normal and you feel much more empowered than you did before and you realize how much more you could do. But now you got to go do it. And it's a new, it's a more competitive world that's enabled by all of this open technology. So it changes the playing field and it forces people to be innovative because I know everybody else can be that much more creative and innovative as well. Right, right. The Mark Shuttleworth demo today with the metal as a service. If you saw that John MAAS first time to ever see that and that would look like a pretty spectacular amount of automation. That's what on metal is at Rackspace, right? They have MAS in that service. There's a similar service in OpenStack that we built called Ironic that delivers a very similar service, right? We call it Ironic. You can use Ironic to stand up your OpenStack hypervisor cloud. You can also use it to run a bare metal cloud service. So same idea. I point, I click, and in moments I get a bare metal image on a server configured the way I want it. So on the OpenStack Rackspace situation here, where does all this intersect with say OpenStack? Because you have OpenStack growing significantly with a vibrant governance model, the community's growing. That's the past, that's infrastructure service. Now you talk on the bare metal as a service. Softlayer was bought by IBM. So you got all this kind of lines going on. How does this relate to this development ecosystem with OpenStack? It seems to be a nice hand in glove situation. Yeah, I think it does. I mean, there doesn't have to be overlap, but there are many places where it overlaps really nicely. Where having openness at each layer actually enables advancement from all the communities. All three groups can benefit from having open access to some other parts. If you think about what you see in bare metal provisioning prior to on metal, you usually had to wait a few hours to get that server. And the way it came up, you know, it may not have all of the pre-configuration, all of the, it may not be set up exactly like you want it. You know, a lot of clouds you can upload your image and you can say, just launch another one, launch another one. With on metal, it's the same sort of idea only you get it bare metal, right? OpenStack enabled that, right? If we hadn't had everything that was already in OpenStack in orchestration, you know, it would have been that whole thing plus a bare metal element we would have had to add. And instead we were able to just take that piece, that bare metal provisioning piece and build it in. So tell us what's going on with OpenPower because that's something that's very interesting. It's a consortium that's coming together. Can you give us an update what you guys are doing with that? What's it mean to the end user? Who's driving it? What's happening with that? Yeah, so Rackspace got involved with OpenPower before it launched. We were engaged in with the group of companies that were talking about the bylaws and the board and the formation of it. We decided not to go public at the time that it launched, but very involved. We'd been looking at power for about a year and a half prior to that. There were certain things about power that were far more efficient in some of its core features for some of our applications than what we had with any other platform. So we were excited about it but we kept coming back and saying, you know what, our data center operators grew up in the next 86 world with all of the stuff around it. You know, not just the processor but the management systems, the tools, all the rest of that. And if we want to leverage this cool new technology and power, we need to go enable it by changing all those other parts around it. And over time, that conversation switched from changing a feature, changing another feature, changing a feature to just, you know, we're changing things so fast. Why don't we just open everything up, right? And a lot of those things that we wanted to leverage a few years ago were still there and they're advancing. You go out, you can do a search for something OpenPower did with a company called Redis Labs. You know, Redis is a really hot application. You can use some functions that are available in OpenPower and only open, right, in OpenPower. So it's a much wider sort of product ecosystem that can plug into it. You can take and enable that for Redis and you can take a server where you needed 20, 25 of them on any other platform and on a power platform, you can do it with one, right? I mentioned, when's the last time we had an advancement in servers that was that order of magnitude, right? It's been a long time. So what's the interest in the ecosystem here right now this year for last year? So you're seeing, like, the scalities here. You're not necessarily making hard of it. It was a good storage solution. There's a software thing going on here. Software defined data center we were speculating earlier. Is it a land grab? I mean, I'm looking at this going, hey, if I'm an entrepreneur, I'm in these trenches. Orchestration, automation. These are cloud concepts. This is a software construct. You can come in and say, I want to own the software defined data center. That means you have the data center operating system is on the table. Can there be one more than one? Is there only one? What's your take on that? I know it's kind of outside your scope but as an industry observer, you got to see people making their move here. I think for our company, I think that we sort of exist outside of that fray to a certain extent. We've had to be operating with that mentality, the whole verticalized data center stack in order to work at the scale we are. I think that's becoming a reality for smaller organizations and operations today. In a large part because of what we and some of the other cloud oriented movements went, whether it's a service provider or software developer, I think that has turned that on. I think there's some sort of, I don't know if land grab is the word I would use but there's definitely a new stack forming and there's competition to say what's that top to bottom stack but I don't think it's just software defined stuff. That's there, right? Software defined network, software defined data center, like software defined storage, those things are all there. They're going to take a while to become a real products that people want to use. Some things are there, some things are not. Like we use a lot of software defined network at Rackspace today. But to see it hit like everywhere else in the industry, I think we're still, we're going to keep moving there. There's going to be another thing that gets turned on and another thing that's turned on over the course of the next few years. I think there are some really radicals changes down at the bottom of that stack in the chips and the servers where we're going to realize, like I was saying earlier, if we rethink that stack all the way down, not just the software defined stuff but all the way down to the bottom, we can get a great deal more efficiency and flexibility in what we have today. So what looks like software defined sort of stack wars today I think turns into a whole vertical, top down stack, not just servers that are abstracted for 50 different purposes, but servers that are extremely vertically integrated but also open. Open and here's the thing that we're learning and the queue for the folks watching they know the queue workloads matter. And now it's not a one trick pony for every single workload. You can't have general purposes over now. You don't have to have a general purpose computer. You're going to have a purpose built app that on purpose built provisioned open hardware with tailored software, virtualization. Talk about that trend, how real is that now? And where are we from the start line I guess? Or are we on the start line? Is it happening now? I think we're past the start line. I think the demand of our customers and our end users for ever more efficiency and performance, ever more scale and ease to start their app and their idea, it's pretty relentless. So we've been feeling that heat for years now and it forces you to evolve, right? It forces you to evolve to stay relevant and keep giving those customers what they want. So it's really workload centric tuning as opposed to just system tuning as well, right? That's right, because for many of them, for many of them the system itself, generation over generation is not delivering the returns that it was. You know, we went from high clock speeds to multiple cores and we would regularly get huge boost generation over generation and performance. Right now that's really slowing down and so you start looking for other ways to squeeze more value out of the system, right? If you listen to Jay Parikh in the keynote this morning, you know, last year he was saying, you know, we saved all this money with open compute. This year he's saying we're saving all this money by taking a holistic look at the stack, right? All the way through, we're, you know, we're wringing out extra performance and efficiency by doing that, right? But the trend of what you're arguing basically is the future, which is there will be unlimited opportunities to have differentiated stacks. So it's not an interest, it just brings up the open source thing, okay? Standardized hardware and everything, but I'm a developer. I can create a unique stack for a unique workload and have some differentiation. Yeah, competition's always going to be there, but there's no lock-in, but I can build my own stack. I can, I can. And there's ways, there's, you know, a lot of people think open source just sucks all the value out of this for anybody that's creating equipment and you're forced to live on really lean margins, but I kind of look at it differently. I think what it forces you to do is burn your boats. It forces you to say, this got me here and I got this much value out of it, but it's open and I can't just sit on that anymore. I have to go do the next big thing. I think one of the next big things in that space is going to be, you know, we'll come up with a name for it that's different because this one's been used, but it's going to be middleware for these very advanced purpose-built systems, right? Which is called that software. Yeah, it's going to be something that basically says that server is an alien compared to what we saw four or five years ago, but I have software that knows how to intelligently use it for what it was built for so that my services developers don't have to have all of this, you know, deep system stuff. Yeah, but this is interesting. Burning the boats is a great analogy to deal, hey, we burned the boats, now we're stuck on the island, we're going to make it work, but platforms are out there, so there's an island, if you will, a land to live off of, you have a platform, so what this brings to question is value-shifting. So commoditization raised to zero, so-called, you know, cloud or whatever, is one thing that's commoditization, but value-shifting. So we've heard from executives saying, hey, you know, entrepreneurs, value-shifting to where the apps are. So there's interesting debates in the tech community right now of that's a bad business model. You know, you're seeing, you know, things. We've dealt with that last year in a big way, right? They're all these price cuts in the cloud, and we were wondering for a little while, you know, there were people wondering, is this the end of us? Cause we're just, if it's a race to zero on cost of infrastructure, right? And the people who get all the value are the ones writing the apps, right? Then many of us are, you know, we're doomed, right? But it's really hard to run this stuff at scale. The clouds we've created today were sort of built to help people prototype and experiment at a really low cost, and just iterate constantly, right? So the entry cost to your idea is really low, but when that idea really takes off, where do these apps sit? They live in Google's cloud or Facebook's cloud, or it's like a Netflix with a huge engineering staff. Most of the customers that come to Rackspace don't have that level of expertise. So what they said was, look, we're moving into more advanced applications, so we need you to move with us, right? We're customers. Well, more advanced applications, but also they're adding more features, faster, more agile. That's right. On a software release cycle, so the software life cycle's fast. Yeah, and as they reach a certain size and scale, their customers are going to demand more, so they're going to keep feeling that pressure, and they're going to say, look, it was easy when I only had 10 servers to run this app, but now I'm at 1,000, and I don't have the expertise, you know, to optimize all of this to run on a cloud. I need your help, you know? I want the best kind of cloud and the best kind of integration service and support to help me run my app. You know, Jeff and I and Dave Vellante, who's not here, who's at back east, I'm our other co-host, we always talk about this. Competitive advantage in an open world. The old world was, we all know Microsoft, operating system monopoly with the, with the apps. We're probably one of the best competitive strategies in the history of the computer industry. Cisco's up there too, and there's Oracle. You name the big apples now in there, so they have competitive advantage. It's some lock-in, but now in open, it's harder to do lock-in. So the new barrier to entry or inimitability or competitive advantage is scale. So the fence to jump as a new entry into the market is scale. Do you agree? I do, but I think it's not just infrastructure scale. It's operational scale, scale of expertise, right? All of those things are, how do I do it bigger, smarter, with better service at scale? We always use the Chipotle example on our team. Chipotle's real simple concept, but they scale. It's hard to emulate that fast. So there's just economies of scale. So what you're saying is, like software, it's concepts, hey, I want to do a cloud. I'm going to be like Amazon. Well, good luck with that, right? I mean, try to copy Amazon over decade. So what do people do? So maybe this consolidation happening, so engineers are worried about platforms, right? They don't want to build on a platform that's not going to be closed down tomorrow. So they want it open. So what is the developer mindset? I guess the question is, what is the developer psyche right now? Because I want to work with people that I know that are, or have interest levels as me, and I want to get distribution from my product or whatever the code is. And I don't want it to be out of business. Yeah, yeah. I think there's a demographic to developers. At certain points, they care about certain things, right? For years, we face competition from from other companies who offered services like infrastructure services that were like ours at a lower cost. And the way our business was structured around support and everything, we carried overhead to offer that service that they didn't. And so we would have customers that would say, hey, I'm going to this lower cost place, or we had customers that would never start with us because this was cheaper. That's all they were looking at. And then they started selling. It's like the insurance, right? And the app took off and things got hard and they're calling that service provider and they're saying, hey, I'm broken. Get me online, right? At this point, the five buck difference you're going to spend on a virtual machine or whatever, it doesn't matter anymore. You have a business to sustain. You're not a big enterprise yet, but you're mission critical at this point, right? And when I think about that, I say there's differentiated forms of service in this business that different types, different demographics of customers will look for. You have some that say, I'm just going to do it all on my own, right? Every last bit of it from the concept of what the service is down to all the engineering work. A lot of those guys start in big clouds and sometimes they move to Colo or they just master those clouds. You have others that are sort of looking at it more like I already have a business. I have something, I realize I need to evolve it, right? For that sort, they want more of that service and more of that help getting in, right? It's easy to think like all the applications that exist today were just created in the last few years and those are the only ones that matter. To the vast majority of the spend in this market is still with an evolving IT group, right? With an evolving set of apps that people continue to value. And I think it's really easy to forget about them. We want to be able to support all of them, right? But you've got to meet them where they're at. And I think in this market, different providers differentiate on how they meet those customers and where. So Aaron, final question for you. For the folks watching out there, describe to them what's going on at this event. What's the main vibe? What is this all about? I've heard this open compute summit thing. What's happening here on the ground, around us, state of the industry? How would you explain it to them? Yeah, I think we're remaking the systems and the technology upon which most of this world operates. And we're doing it a piece of the world at a time, right? And it's continuing to expand from there. And we're doing it in a way that is empowering the operators, the end users, and every other participant to do it the way that they want to do it and determine which one has the best model. I mean, in microcosm, I think that's basically what we're doing. At a large scale, I think we're actually creating an approach that's going to drive other industries to think and act the same, even if it's not directly IT. You look at what's going on here and how fast this industry is changing as a result of this sort of thing. And then you start looking at some other industries that are more stagnant. And you start saying, man, how could I enable my industry to go through change that fast and that disruptive? And I think some want to go drive and do that. That's great. We appreciate your time on theCUBE. We've opened Power and Rack Space. You guys have done some great work. Congratulations. And again, it's super exciting here. This is where the next industry innovation will come out of. Our prediction is there's unicorns in this room, billion dollar valuations. I was kidding, certainly there are companies here with a billion dollar value. Intel, I mean startups. Big whales are here, but really entrepreneurial activity. It's really hot. This is theCUBE. Of course, we're on the trenches extracting that signal and sharing that with you. We'll be right back with our next guest after this short break. Thank you guys.