 Live from Vancouver, Canada, it's theCUBE at OpenStack Summit Vancouver 2015. Brought to you by headline sponsors EMC and jointly by Red Hat and Cisco with additional sponsorship by Brocade and HP. And now your hosts, John Furrier and Stu Miniman. Okay, welcome back everyone. We are here live day three at the OpenStack Summit in Vancouver, British Columbia. I'm John Furrier, the founder of Silicon Lam. I'm John Stu Miniman, chief analyst on the cloud and infrastructure at wikibond.com. And our next guest, Mark Interante, who's the SVP of engineering at HP Cloud Group. Welcome back to theCUBE, CUBE alumni. Great to see you. Thanks, great to be back. We love always having your talks on theCUBE here. Also, you're talking to everyone. There's a lot of engineering involved this year, more than ever. This is the theme this year, is engineers are here building, not just kicking the tires, sessions are packed, people are sitting on the floors with their laptops open. So it's a pretty amazing environment. So what's up with the code? Where's all the action? What have you guys shipped? Give us some of the data, share what's going on at HP Cloud. So the Keyler release is really, let me start with the OpenStack community itself. And the Keyler release I think is fascinating. Couple of really big things landed that mattered to customers. So one, federating Keystone. So what that means is you have multiple clouds, you connect them together. You saw the demo that we, BlueBox and a digital film treat it. So showing a complex multi-cloud, private and public cloud federation working seamlessly. That's great. That's really enabling multi-clouds happening. You know, another thing I'm excited about is in Swift, we finally cracked the nut and really got into production something called erasure coding. And you're like, what's that? Erasure coding allows you to more efficiently store at same reliability level, large scales of data. And so, classic Swift, three copies for the kind of redundancy you need. Erasure coding about 1.6. We just cut the price in half. And that's like, it's awesome to have software cutting the price of hardware. Mark, I mean, at Wikibon, we've talked a lot about, you know, killing raid. Yes, yes. And, you know, therefore object storage, you know, that's where erasure coding's there. Awesome, that's how the cloud should be. Wait, the future's here now. It's just unevenly distributed. Absolutely. Absolutely. It totally is. So I think those are just kind of two of the things that I'm most excited about when I talk to customers. A lot of people are really interested in scale-out storage. And then they're trying to figure out how do they connect their multiple clouds? People have multiple private clouds, whether it'll be a production cloud, a DevTek cloud, or a business line cloud. Now some want to connect them together because they want to be able to move some stuff back and forth. And this allows that. And, you know, we were talking with the Cisco guys earlier in this whole internet working phase that TCPIP enabled back in the computer industry. Cloud is inter-clouding, not internet working, but like the inter-cloud relationships are very interesting, is that you're seeing people wanting not one cloud. They want the ability to work. And you guys talk about bursting. Explain that dynamic from a technical perspective, that from a client perspective, you look at a pool of resources out there where the storage and whatnot and a set of services, service architecture, whatever you want to call it. And the cloud opportunity, it's not just one cloud. And you guys have that philosophy, explain that. Sure, well I think if you look at a cloud as a set of resources that are tied to the actual underlying infrastructure, people are going to be building differently-abled clouds with different capabilities and they're going to be at different time zones, different geographies. You can't get away from capacity throughput, latency, and geographic, the geographic nature of this. So people are going to want to have, this is why people are building certain clouds in Manhattan. Because they want speed of light to the stock exchange. They want different kinds of computes in those areas. They put them in Chicago in different areas. So we see, and we want to enable people to build a wide class of them to fit their needs and make the most efficient use of compute and storage resources. So customers don't talk about infrastructures as a service or platforms as a service or softwares as a service. That's not their language, that's the industry language. Those are the layers people like to put boxes in. They talk in, what is the language of the customer? From an engineering perspective, you have to map. I mean, we talked about services architecture earlier. But they don't say I want to pass. Or I want some cloud foundry. They do it for a reason, because they're architecting. What's your view of the current language or conversation that customers want to have? What's some of the linguistics, you know? A great question. So I was in Europe for a couple of weeks, just here in the last month, and here's what customers were talking to me about. And typically these are kind of VPs of infrastructure. These might be CIOs. These could be VPs of application development. And the conversations absolutely sorted around I want to build an environment for my application development teams to move faster with less friction so they can innovate more. And so how do we build that environment that allows application developers to have a good productive, happy life in developing? And that's what the CEOs are asking them to do. That's what the CIOs are taking that charter down. And the infrastructure guys are there to enable those application developers to help software take over the world. And if you don't have this a good environment, what happens? Turnover goes up, productivity goes down. And the conversations are all about automation, DevOps, CICD, and the talent transformation. How do you make use of this underlying technology? So it's very little about kind of actually, I as is the enabler for developer services. And that's what developers care about. So I got to ask you the philosophical question. Sure. Cloud flips things upside down. And we've heard a lot of those conversations on theCUBE where it's upside down, things that think people are thinking differently. What's upside down? That's an opportunity. What is the cloud done from a shifting of mindset? How does it change? What's flipped upside down? Where's the action? How can someone take advantage of and move quickly and be agile and go from old way to new way? Sure. I think the thing that has flipped is very straightforward. It's flipped us from a land of scarcity to a land of seemingly infinite abundance. So when you have seemingly infinite, you do things with those resources that you would have never done. Because it would have been silly to waste a hundred thousand transistors on to make a digital watch. Whereas today, there's like billions of transistors all around me. So the cloud enables seemingly infinite storage, seemingly infinitely configurable networks. And of course, seemingly infinite slice-upable compute. Which means that I as a developer can get things almost instantly and then manipulate them. So that opportunity is what's making application developers. And Stu and I were talking earlier about this notion of like, things are being invented for the first time. Because if you take that, if you believe what you just said, which we do, there are an infinite amount of, well, infinitely infinite storage. Seemingly infinite compute, storage, bandwidth. Internet of things are on the horizon. People are connected to the network as distributed computing, et cetera, et cetera. New things are being invented because you couldn't do that before. That's right, yes. What are some examples in your mind that you could share where you see this happening? Some proof points where you say, man, this was never possible years ago. This is what's happening. Here's how customers are thinking about the value chains of IT and whatnot. So maybe just maybe in the, if you pick the airline industry and the airline manufacturing industry, the amount of data and the amount of understanding about the dynamics of a single airplane. So a single airplane generates about one terabyte per flight. And that's kind of with current level sensors. I bet you in 10 years or five years, it'll be five times that amount. If not more. If not more, if not more. And so what that allows people to do is understand wear and tear, reliability, performance characteristics, another three or five percent improvement in the next year or two on operational efficiency and help reduce global warming and all of these things and enable a data-centric enterprise for manufacturing. And we just have not had that. What we had was, I think the gear on my plane needs to be about this big and the engineer drew the gear in a CAD system and they've got their gear and they put it in the plane and they never change it for 30 years. Today they take the gear, they take them out, they do image processing against it, they look at the wear characteristics, they go, if we can make this thing 50% lighter by doing a different alloy on this edge and all of a sudden they take 10 pounds out of the plane. You do that times a hundred or a couple of hundred and you end up with a radically different process. Yeah, fuel savings. Yeah, so Mark, if I look at things from a system design, I love what you're saying about kind of infinite this, infinite that, but there's no such thing as removing a bottleneck from the system. It's correct. It's only moving it somewhere else. So is it the people that is the bottleneck in the system now when I think your SVP of engineering? Yes. How do you get the right people? When I talk to the customers, when I talk to the startups, talk to everyone, it's not enough well-trained people and we spent years talking about, oh, there's going to be a shorter data scientist, the shortage of this and that. Is there a shortage of people or what's the big challenge? Well, I think there's absolutely a shortage of good engineers that want to work on your problem. Again, there's lots of problems. The good news is we're expanding rapidly. So this person wants to work on that problem. It's not yours. They want to work on somebody else's problem. So I think we've got a shortage there. I believe that we got to start early in the pipeline in high school and we need to show people how powerful STEM jobs are, how they're not that, they're not nerdly. They're actually quite fun and we got people here having a lot of fun. We got a lot of people designing things, problem solving. I really want to address the critical shortage of women in our community. It is a big deal. I think that that is an area where we have to work on the pipeline so we're getting people earlier who want to go into these fields. I think that's a huge challenge right now. Talk about cloud foundry. You guys are a big part of that. HP's differentiating with your unique view of your customer base. We heard Bobby Patrick on talking about bursting and that it's not about public cloud and clarified that you're in the public cloud. You're not going to get out of the public cloud capabilities but it's not necessarily a business in the sense of in that misquote that the New York Times quoted Bill Hylthon. It's all going to work together. So he explained it works together. So cloud foundry is a big part of that. What are some of the milestones that you guys are seeing from that effort and how are you guys differentiating against the foundation? So we've just had our third big release of our cloud foundry system since November and this was a big release because this has got native.net support. Again, 40% of enterprise developers are building in the.net environment and what they would like is they'd like to be able to use more cloud. Cloud has come to cloud, yeah. Exactly. And so it's native. We've got support for the entire Microsoft tool chain and IDEs embedded in there. The Foundry has actually accepted our code for incubation which we're very happy about. We've contributed all that back. I did a keynote about 10 days ago and shared all that information. And I think part of it is for us is really becoming a more major player in the cloud foundry core like we are here in OpenStack. You guys are happy with what's going on there. Yeah, we're absolutely happy. We've got great interactions with the foundation and the other leaders in that group. And it's getting a lot of good traction early which is great to see. They're getting some ubiquity. They're getting a lot of great edge options from the industry. It's not just one vendor. It's like a lot of great people. It is. But there's a lot of co-opetition now involved. Sure. Which is okay. HP's used to dealing with that. We do that all the time. What are you guys doing and differentiating on top of? What is the unique HP twist? Because what's interesting is that everyone's got their own unique twists, right? So our twist is we believe that infrastructure matters and we're tuning and optimizing Cloud Foundry to work really, really well with OpenStack. And we're unique in doing this. We've got a much tighter set of bindings and a much more broad set of capabilities that enable the Cloud Foundry user to be able to access and make good use of all the resources of OpenStack. But you're plugging in like .NET is interesting because now you're tapping into a whole other development community that has huge enterprise experience. That kind of needs a shot in the arm. I mean, I'm not going to, I'm not saying they're irrelevant, but it's a bit older. Web 1.0, 2.0, but now with Cloud, Agile seems to be a competitive advantage. And that's what developers are trying to figure out where the competitive advantage is for them and who could they play with nicely in the stack, if you will. Absolutely, yeah. How do you attract them? What's the conversation? Hey, guys, join us, and they can brush in and join you, or what was the dynamic? Because that's a tough nut to crack that developer community. You know, I don't spend a lot of time on our developer evangelism side. So I'm not in as much of those conversations as maybe Bobby or some of other people are. So I probably don't have quite as many rich experiences on that front. I'm trying to hire developers to help us build the cloud, and that's more of the conversations I have every week. Let's talk about helium, right? So HP Discover's coming up. So I know that you guys usually got a couple of things surprises. What can we expect at HP Discover coming up in a couple of weeks? I can't quite tell you that stuff that we got coming up. But I hope, are you guys going to be there? The cube will be there. Oh, excellent. We'll have a conversation then. Because we've got some good announcements coming out. We've been working hard. So you will have something? Oh, absolutely. Yeah, okay. I guess, Mark. So let's bring it back to OpenStack then. We've made a lot of progress. I think the integrator, going away from the integrator release, going to the Big Tent, talking about Power by OpenStack, something we've talked a lot about a lot. One of the big gaps I saw is migrating from one version of OpenStack to the next. We need to get there. What else? What's the big challenge that we need to work on I think we've got clean line of sight to maturity. Yes. And we're almost there. But what do you think we need to do and how's HP helping the community? So my big theme in my keynote earlier this week is around reducing the operator complexity and operator load for running these things at whatever scale you're at. So I want us to drive down the operator load by a factor of five in the next three releases. So we've been having conversations like three or four times a day in the last couple of days with other teams, with other people about what do you mean by that? How do we do it? Well, I've got all these tools where we've got different tools. How do we find where the time goes in running a pick a number? A 10,000 node cloud. And how do we take that time from this to this, you know, by Barcelona? And part of that it's just, it's starting to really understand where does, why is monitoring as complicated as this? How about playbook remediation for normal anomalies? We've done a lot of investments here in the last two releases on HA and the ability to lose part of a service and have it fail over. That's already gotten better. And HP has put a lot of its expertise in HA into those areas. Yeah, you're bringing up a great point. I tell you, especially with big enterprises have what I call hyperscale envy. At Facebook, at the Open Compute Summit, they said, we manage 20,000 servers per admin. Well, the thing is, is really, they don't really have server admins, they have people that build applications and then they've rolled out a data center scale not even a rack scale and they manage it. How does OpenStack in HP, Hylian, really move the needle to get us from, managing all the bits and pieces of the spoke infrastructure to, what's the new metric? How do we say that we've got success and we're doing a lot more? I think it's admins per 10,000 nodes and whether you get that from X to one fifth X is what I'm trying to drive to. And I'm also trying to get people to, to us to agree on a set of metrics, have teams go back and look, just share the stuff. And the community's pretty open. I mean, we've shared a lot of things over time. We've shared some of our size and latency of our cloud files instance and had great feedback from the Swift community about our level of transparency on some of the dynamics there. So I think part of it is let's share more actual data, figure where the time is going and show our development teams how they can make a big impact. You guys have been great contributors of OpenStack and you guys don't get enough props because the PR teams are all the mucky muck levels. But on the ground here, OpenStack, you guys dominate a lot of the code. You guys, very, participants, hundreds of people here from your team, HP Cloud, great community participating, been there from the founding beginning. So give a little props there and give a little shout out there. The team's doing great. You guys are doing great. You guys can share a lot of real code. So what's next for you guys? You're the engineer and that's where the action is. You're under the hood. You're the chief mechanic of the cloud. What's the hotspots? Where's the focus? You said you mentioned making the infrastructure matter. Okay, get that. But what specifically are you guys focused, lasered on in terms of engineering? Okay, well actually before I get to engineering, I think where I'm trying to put, encourage energy in the community is in learning operator knowledge and operator patterns and techniques and on the other side of the fence, encouraging people to think a bit more like a product, some product management thinking. So I did props out to the product SIG, which is a cross company group that's exploring, like how do we get clear on our personas? How do we get clear on the use cases? How do we help people plan out a large cross cutting set of features? Reference architectures, things of that nature? Reference architectures, yes. So if you think about cross cutting set of features, it's really hard to do that in open stack. It takes many, many versions. What does that mean? So how would I do IPv6 end to end? How would I do better NFV support end to end? So the way to do that is to get really clear on all the 12 pieces that have to change, kind of product management. Get really clear on, and make sure everybody understands that six subsystems have to change in the next release and get people starting to talk about them and have a higher chance of landing those in six months whereas today it takes us 18. It's like feeding the community data and discipline and practice that helps scale. This is a real opportunity for the community because this is a scaling challenge, growing pains, right? It is. You know, we hope to stack some to say it's adolescence phase, it's on a path to maturity, but this is the bones that need to be set in the skeleton of scale, right? You've got to have some product management disciplines, not just throw code over the transom. That's right. And I think part of it is, it's always been a very developer and technical-led community, and what we're trying to find out is how do we have product managers enter the conversation, offer data, feedback, insights, kudos and complaints from users in a way those can be consumed in problem solvers. The transparency is a closed loop. You've got feedback from the community, customers all feeding in, so it can be a challenge, I can imagine. All right, final words since we're cut on time. What's your goals next year, OpenStack? What are we going to be talking about? What do you see happening and what successes do you see? What's going to be the transformation of the next year? That's a great question. So I want to totally land all the dimensions of Federation from the UI to the billing to the authentication to all those dimensions. So I think we've landed the first piece in this one. I think the next two releases, when we're in Austin next time having some barbecue, I expect that we're going to be talking about a whole set of federated clouds that operate seamlessly that are able to, I'm able to burst up resources, get them billed all back to me. That's probably the biggest thing. And I want to see massive upgrades to our operator capabilities and tools. I want people to say, look, we're deeply on that road to getting the 5X and people are not complaining or nervous about deploying an OpenStack because they feel like they know how to do it. Just quick poke at that. Cloud bursting is something we all kind of laughed about for many years. I like a little bit more kind of the app catalog that we have. Why is federated a little bit different? I know we're short on time. Yes, I'm not a big believer in bursting because what happens is people just, there's specialized cases where it happens. What people want and what we built in the Valley for the last 15 or so more years is complex multi-DC always up apps. And so I was, I led Yahoo Finance for many years. So in doing that, Mithing runs in five DCs, you'll lose a DC, you know, a page blips, you're all back up. So more and more people want those kind of apps and they need those with data center resilience, room resilience and service resilience. And that's much more of the hybrid sophisticated app that we're trying to enable easier to do. It was a real pain 15 years ago to do this. A lot of PhDs just developing that code. It's really that distributed world that we've been working towards. Yes, exactly. Awesome. Mark Interanti, SVP of engineering at HP Cloud bringing his expertise and the leadership to the community and sharing that. That's what's great about this community. Again, it's on the path of maturity. Congratulations Royce. And again, props to HP for really a lot of great work you guys have done, being a big part of OpenStack, fostering innovation, keeping the culture and passion and organic growth from the ground up, while bringing some mojo from the big company, discipline in. So congratulations, Mr. Cube. We're bringing more data right after this short break. We'll be right back live. Day three coverage of the OpenStack summit here in Vancouver Bridge, Columbia. We'll be right back.