 Live from Boston, Massachusetts, it's theCUBE. Covering HPE Big Data Conference 2016. Now, here are your hosts, Dave Vellante and Paul Gillin. Welcome back to Boston, everybody. This is theCUBE, the worldwide leader in live tech coverage. We're here with Robert Young-Johns, who's the executive vice president of the software division for Hewlett Packard Enterprise, longtime CUBE alum. Good to see you again, Robert. Yeah, fantastic, thanks for having me. You're welcome. So, enjoyed your keynote. Building on the themes that you talked about at HPE Discover, you talked about software eating the world, and now analytics is sort of eating software embedded everywhere. Talk a little bit about that. Well, I think if you look at the evolution, I think the thing I talked about this morning is the way we used to devise applications as we go and study a business process, you know, walk around with that clipboard, who's doing what, we then turn that into a process with a nice diagram, we then try to automate that process, and then we decide what data we needed to support it. And I think what's happening today is the world's almost turned upside down. People are starting with the data, they're using analytics on the data to expose patterns and potential new business processes, and then they're building new business processes on top of that, and then they're transforming and disrupting industries. And I think that's a huge change and it's accelerating right now. I think that's pretty insightful. I want to make an observation that's related to that and get your feedback. So, in the 80s and 90s, the technology was mysterious. It was unknown. The processes, however, were known, and today the technology's sort of been demystified and the process is unknown. And your premise is that data is the starting point and then you're going to move down, maybe serially, maybe not, and ultimately you're going to design processes that fit the market and the data. Yeah, and can completely transform industries as a result. And that's what you're saying today. Because you're actually attacking known paradigms and most other people in the industries have worked on that 80s and 90s model and you're coming in and saying, hey, maybe that doesn't work anymore. Maybe the data is showing me there's a different process. The taxi driver versus Uber is a great, very stark example of that, of how that's happened. But it continues to be an exception, a rare example. And I wonder, do you see other industries, many other industries, primed for disruption the way the taxi industry was? A lot of things were sort of set to be disrupted in that business. Or is Uber more the exception? I don't think it's the exception at all. I mean, I think you look almost every industry segment, you'll see some major disruption going on. I had a fascinating conversation with some people in the airline industry last night about what's happening there. Not so much with the basic provisioning of transportation, that's much the same as it ever used to be. You fill planes with gas, you put passengers in them, you fly them from A to B. But the way in which you attract passengers, the way in which you expose fair structures to them, that's being disrupted very heavily by a number of players. And what was a strength, I think, for many aircraft companies, a few years ago, common reservation platform, all that sort of thing, is actually morphing into a weakness because they're getting abstracted. Unless they respond, there's a danger they do get abstracted to be the commodity provider. So it's one big example. I think you go through almost every industry, you'll find some example like that. It's an interesting example because in that case you have, there's mystery in the data. And the data is it was actually a competitive secret. And once that data was exposed, others were able to take it and co-opt it and use it against the airlines. Totally. I mean, you know, even it's very obscure places. I mean, my favorite sort of app of the moment, I'm not suggesting this is actually necessarily a big data app, is one that it may not even be anywhere, other than Silicon Valley, it's called Fill D. And basically I leave my car in the parking structure, I get to the app, I say, fill me with gas. I walk away, I come back later on the afternoon, my car is full of gas. Amazing. It's $5 or $3 if you don't mind when it is. I pay that every day, not to have to go to a gas station. I'm not sure what that's transforming, but if you read the traditional gas outlet. You talked about the fear of machines. So let's get into that topic a little bit. You know, machines have always replaced humans. That just seems like now they're doing it with cognitive functions. Why are you an optimist? Why are you not concerned about that? Well, I am concerned about it in the long term. But I think right now, as you think forward in the next 10, 20 years, actually it's quite the reverse. I think machines can unleash human intelligence in a way that's never been possible before. Which is why I talk a lot about this concept of augmented human intelligence. It's we as people don't have the capability to absorb the vast datasets being thrown at us, the amount of information that's available to us right now. But if you can have filters and preprocessing essentially, it can allow you to make better decisions. It can allow you to do what we do best, which is apply judgment. And look at situations and go deep into what root causes are, analysis, and so on. I was driving in this morning and I heard a story on NPR that Marvell Comics is putting out a comic book with Trudeau, the Prime Minister of Canada, is one of the, you know, he's a protagonist. And the idea is they're using machine learning and artificial intelligence to predict crimes. And they're trying to determine should we arrest people? You sort of- Yeah, that was a minority of course. A minority of course. Should we, right, okay, so of course, should we arrest people beforehand? And that's the, they're making a comic book about it, and they go to Trudeau for his advice as to whether or not he's the, you know, scion. Well, what ethical challenges are gonna be brought forth due to machine learning and privacy, ethical? How are we dealing with those? I think they're all tough. I mean, and I think, unfortunately, for an industry point of view, they're very hard for the industry to actually come to conclusions on, because the technology tends to leap ahead. Just when you think you've got it solved, you probably solved for a five-year-old problem. You haven't solved for today's problem. So I'm honestly not sure. It's not, it's really not, it's your point that it's not the technology industry's role to sort of determine that, it's public policy and let the politicians deal with that. Exactly, exactly. I mean, the only problem is that they tend to deal with what happened five years ago by the time they fixed it, life's moved on. It's outdated, okay. So I also want to ask you, you made a statement. You basically threw down the gauntlet and said, you know, IBM Watson made a lot of noise, great marketing, I'm sure, you know, great product. You said, I think we have a better solution. Unpack that a little bit, why? Tell us about your solution, why do you think it's better? I think it's partly a point of view. And it's not necessarily right or wrong, it's just very different. My perception of the IBM point of view is that they want to be content vertical specialists. They want to be specialists in the weather industry or they want to be specialists in healthcare. And they've acquired companies and so on to do that. They want to own all the data sets associated with those. And that's fine, that's a legitimate business. But to me, that's a business that's led by specialization, it's led by consulting, and then there's tooling that sits behind it. I think our view is very different. It's almost completely, not the opposite, it's sort of orthogonal, in the sense we want to be the toolset that anyone who wants to expose the value in data can use. So I don't want to be the weather provider. I don't want to be the healthcare service provider or the transportation industry provider. I'd rather people who are specialists in those domains and those sectors were the face to the customer. What I want to be is the arms provider, the tool provider if you like, that provides the analytic capabilities to those industries. That's a great point and it's a very clear distinction between yours and IBM's strategy clearly. You had the video this morning that you showed of Haven on Demand was really compelling. It made it look like all these 70 services that you're exposing, how are you planning to position them? IBM had the free Watson analytics, come and test drive it yourself. How are you planning to position the Haven on Demand so that really it spreads virally that a lot of people are able to tap in and use it? That's exactly been our approach. I mean, what we've taken the view of is that we need to expose it as a web service to make it freely available to any developer. We've done that since its origin. We have over 18,000 developers signed up. And I've said repeatedly, I'm willing to go head to head API to API with Watson because I think we can beat them in that area. But you're absolutely right. It's got to be easy to consume, easy to use and orientated towards the developer. Now, as people move through and move into commercial exploitation, then we'll start to think about what the right business model is, how we get paid for it. And we're beginning to see some breakthroughs in that area too. But it's got to start with getting developer traction. So you said you had the strongest platform, you said the strongest platform in the industry. And that is the point of view. You talked about that, arms dealer versus vertical. There's the API comment that you just made. Can we peel the onion on that? Is it what, easier to use? More of them, more robust? What makes them more appealing? Well, I think firstly there's more of them, but that's not necessarily a differentiator. What I do think is- There's more things you can do. What I do think is that they're all pretty deep because what we did is take 15, 20 years of idle experience with our product and we exposed functionality. We were originally built in idle through these APIs. So these weren't things we created from fresh. These are things we were already working on. So our image analysis APIs were based on image analysis we built into idle, which was used by security services worldwide to detect bad guys. So we have significant experience in that. Voice detection, similarly, we had some call center software that we were using to do real time voice analysis. We've taken that capability we've exposed as an API. And there's some new ones as well. We've been working with labs on some really interesting graph search capabilities to connect entities within datasets. That's a new capability. But there's deep science behind this. I constantly get asked, why do I have dev labs here in Cambridge Mass and Cambridge UK? And I just go, that's where the really smart people are. And there's a sort of consequence of that, which is I sort of have to pay a little bit more for them. But I think it's the right thing to do. So if software is eating the world, the developers are feeding the software beast. You talk about 18,000 developers and you and I and John Furrier have talked a lot about the importance of developers. Give us an update. Starting to gain traction. What are you doing for the developer world? Where do you want to see that go? What's the strategy? Well, I take two places because part of what we're doing as well is trying to use the analytics platform to turbocharge if you like. Some of our more traditional applications in application, life cycle management, quality management, performance testing, and so on. As well as trying to appeal to the raw analytics developer. So I think I talked about the, we do a lot of hackathons and so on and so forth. I think we're seeing a sort of continuing sort of, I'm not sure it's exactly exponential because I try to be mathematically precise on that term, but we're certainly seeing very fast growth in the raw developers using the raw tools. But what I think is even more exciting is we're starting to embed that tooling in other things that we do in HPE software. Because what are my view is that everything we do where we have domain expertise, which is like IT operations, application development, security, and governance and compliance for data are all going to get changed fundamentally by analytics. So in the world of the application developer, for example, we have a great product application life cycle manager, which is going to try to embed predictive analytics to collect all the data that an application spins out as it moves through a DevOps cycle, test results, dependencies, prior performance of that particular development team, number of bugs, et cetera, and then synthesizes that to produce a prediction as to whether that code is likely to be successful in production. And that's a completely new way of thinking. No one thought of applying analytics to DevOps in that way. So we've got the tooling to do that. This is embedding analytics at the core. You talk about that. So I wrote down security, ITSM, DevOps, but just beyond just core DevOps workflow whether the product code is working or not. And then you gave some other examples as well. That's what you mean. So that's happening inside of HPE, an ecosystem. So my thesis is that we should use our own analytics platform in areas where we have deep domain expertise, which are those four areas. And outside of that, I'm looking to other people and to be their tool provider, rather than try to end myself in 50 different industries with all these vertical experiences. That's the point of view we've taken. I think it's the right one, particularly where we're coming from. As I say, others are going to take a different approach. We've seen Facebook, Microsoft, and Google all open source their machine learning libraries. Is this a threat to you when the markets that you're strong in are you susceptible to competitors open sourcing their products? Or is that actually an advantage to you? I think it's an advantage, but I don't think it's specifically about open source. I do think that some APIs are going to become very consumer orientated. And at that point, the ability to compete is actually going to be somewhat limited. For example, I've been thinking about voice. And we've got some really good voice technology that when I look at the amount of resource that Apple or Microsoft is throwing at the voice problem with Siri and Katana and so on, I sort of have to take a pause and say, maybe those consumer technologies will eventually migrate into the enterprise world. And that's the case. Good for them, we'll use them. And we'll go and focus on stuff that's really specific to deep enterprise business problems. But open source in itself doesn't sort of concern me. In fact, internally within HPE software, we're engaged in a project right now to change the whole way we do development to be intrinsically open source, to use open source techniques and tooling inside. Not necessarily to make our products open source, but to use the whole concept of an enterprise GitHub, upstream requirements gathering and so on for the way we develop our own products. And that makes it easier if we then choose to make them actual open source when we take them to customers. Is HPE as active in the open source and the broader open source ecosystem as you need to be? I think we could be more active, but I think it starts with getting our own house in order because one of the things I'm trying to do inside HPE software is encourage much more sharing of code between the various divisions. And like many large companies, we have what I call the reverse of the not invented here syndrome, which is that if they can find a piece of code that we wrote ourselves, it must be useless. Therefore, they want to buy it from outside. It's a congenital problem in the development industry. So what we're doing is to say, okay, every project we do is an open source project internally. It's all up on enterprise GitHub. If you have requirements, put them in the upstream. If you want to contribute code, put it into the repository and we'll put it on to mainstream. And I think it's a completely different way of thinking about code generation. Now we're part of the way there. We haven't completed it yet, but I think he's getting a lot more excitement inside about how we share code and how we move this thing forward. And HPE and now HPE has done a better job. A decade ago, HPE was completely inactive in open source and you participated in many, many open source projects now. So I mean- Yeah, we do, but I think we're also clear-eyed about it. I mean, at the end of the day, enterprise customers look for support for open source and you get a lot of real open source and you get what I call faux open source where the open source really is the attractor to sell a big support contract, which over time becomes really no different to buying license. So you just got to be careful and you navigate through those in a wide open. Right, you're selective and where you see an opportunity to take OpenStack, for example, HPE has provided some support there, obviously. It makes sense, right? We have, but I was talking to a customer this morning, in fact, here who was bemoaning the fact that one of the challenges of something like OpenStack is that there's a release train going through and they're not necessarily ready to upgrade to every latest release. And therefore, if they're not careful, they end up with multiple releases to support with very complex interdependencies. And so they look for someone at an enterprise level to solve that problem for you. You're actually seeing some open source companies now begin to back off on their release schedules because they were pummeling customers with updates. No, exactly. In OpenStack's classic, I mean, they come out with a new kernel, Mintaka, Liberty, et cetera, every six months. And often, code that works on the prior one doesn't work to the future one. And so it's not just the open stack environment you have to upgrade, it's often a lot of the apps and functions that use OpenStack. So it's something I think we have to think about as an industry, we'll care for. Well, an HPE strategy, HPE now HPE strategy, Visa V, public cloud has shifted and now you're sort of embracing certainly Azure and the Vertica Group anyway, Amazon, AWS. You're listening to the customers we're saying, we're doing this, you're either going to come with us. But our fundamental thesis is the world's going to emerge to be multi-cloud. Sure. And I think maybe three or four years ago, we felt they'd be a, quote, a cloud winner. And that's why we put so much money into OpenStack. But right now, I think people are going to make dynamic choices between AWS, Azure, Google, as they move into the enterprise space, their own private clouds, be they on OpenStack or VMware, et cetera, et cetera. And I think what we need to do is be really the abstraction there if we can above that. Is that battle over, the IAAS battle? Well, it's no clear winner. I mean, there's lots of winners. Right, everybody sort of got their piece of the pie, but is it just the pie's going to get bigger and their shares will expand more? I think so, and what we focused on, and I don't know whether it's going to get proven out in the market, but I think it's conceptually the right approach, which is that people are going to want or enterprises are going to want a single orchestration set of tooling that can then go and deploy workloads into whatever cloud they choose. Now, a lot of customers say, I love that idea, not convinced you can pull it off. Yeah, sure. Actually, I think we pretty much close to pulling it off. That other customers just say, you know, at the end of the day, if I'm going to deploy an AWS, I'll use their tools. I want to do an Azure, I'll use their tools. So to close that loop doing what the airline fairsites are doing to the airlines, essentially disintermediating the cloud provider. And a lot of people are trying to do it right now. And I think obviously, AWS, Azure are pushing back because they've got their own tooling and they want to actually get you to use their tooling. It's going to be very interesting to see that. Is it going to be add business value to do it that way or is it going to be convenient to do inter-clouding? So I want to get your point of view on the software industry. We've seen a slow motion collapse in infrastructure software pricing for the last decade, open source cloud, where you sort of know the reasons why. And there's this sort of flight to SaaS. What does all that mean? You're seeing some public companies try to pivot to a SaaS model and go from on-prem to a ratable model. You guys are facing that yourselves. It's a big challenge. You're seeing mega acquisitions. You're seeing the Hadoop business and maybe not pan out the way a lot of people thought. What's your perspective on the software industry? So how long do you have? We have all day. I know, I know. I think there's a lot of stuff going on. I'm not quite sure where to start. I think there will be some consolidation, particularly on the infrastructure side. I have no doubt about that. But I think that we've actually used a really nice slide that I use internally with my leadership team and actually slides you to the customers just to go, here what's our point of view of what's happening? And it starts by developers are increasingly important. Developers are driving infrastructure decisions at a level that never, ever happened before. And then from that, there's some implications. I mean, one is that SaaS as a delivery mechanism is becoming very, very powerful. Secondly, the line of businesses are typically making choices in a way that they never did in the past. In the past, it was an IT department that made the decision very often now, some new groups being set up and they go and do their own thing. And they probably will, in any cases, not choose their IT department. They'll go to AWS, they've got to respond to that. You've got to respond to the fact that open source is becoming very credible. And you can't just dismiss it, you've got to cooperate with it, you've got to see it as a good thing. And actually the Vertica team did a great job on this because I think two years ago, they'd have said Hadoop's a competitor. Now they go, delighted you have Hadoop, we can turbocharge the performance of Hadoop with things like Vertica SQL on Hadoop. So we've got to acknowledge that. I think there's a lot of new technologies emerging like microservices, containers and so on. You've got to be at the forefront of doing that. But you know, at the bottom of this chart goes, this one thing goes right across, but whatever happens, analytics is eating everything. So which is why we're making this big play on if you think about HPE software, what do we do? We bring machine learning and deep analytics to those worlds that we have deep domain expertise, security, IT operations management, the application developer, and IT management and compliance. That's what we do. So that big on analytics, make that your secret sauce, your core IP, work with the rest of all these trends that you lucidated very nicely. And software's still a great business. No, it's a great business. But the thing about the software business, which makes it both terrifying and exciting, is there's no barrier to entry. I mean, 10 years ago you had to buy a server, at least. So you had to go and spend 10 grand to get yourself going. Now you don't even have to do that. And so why? You use your AWS account, you're up and running. And so people, but companies now raise still, Looker raised almost 100 million, why? Because they have to compete in the marketplace and it's branding, it's marketing. It's distribution. But once you get going. But it's not the technology anymore, right? But there's no standing still here. I mean, there's no looker. There's no going, hey, we got a nice franchise here. Let's exploit it. I hear this word sometimes analysts use, your software is really sticky. And I go, I wish I had an index of stickiness. Because software is not sticky. I mean, maybe some application software. I mean, once people put SAP in, they're sort of done for the rest of their no knives. But once you get into our world, software is always displacable. There's always someone out there with a new open source tool or a new product or some new idea. And often it's plain fashion. I joke that I tried to join an industry where everybody wore white coats and walked around with clipboards. And really I've joined the fashion industry. And being on the front end of that fashion curve for software is really hard. Quite what do you speak about? Big data, clearly you're wearing bell-bottom trousers from 10 years ago. People's worlds moved on, it's now cognitive. Trying to keep ahead of that is really interesting. It's just fascinating. Customers, of course, are driving this too because they're afraid of lock-in and they don't want to be tied to a vendor anymore. And through using APIs and open source and industry standards, industry is giving them exactly what they want. But that's shooting yourself in the foot too. It has, but lock-in always reappears somewhere else. I mean, I can get to the endless quest. Data, to me, is now the big concern. It's what the cloud companies are using against each other. As you're locked into Amazon, now you're locked into Azure. And if you put all your data into one environment, it's actually, people talk glibly about petabytes of data as though it's like a diskette. You can carry it from one place to another. It's sort of not. And once you've got a petabyte of data loaded in a repository somewhere, that's actually likely where you're going to put your apps in the future. Well, we haven't talked about combinations, right? And your CTO was up talking today about how we're leaving the data in place. The epiphany of a dupe, right? Was bringing five megabytes of code to petabyte of data. So that's clear, data has gravity. It really does. I mean, and what we try to do with our APIs is make them sort of encapsulated in the sense that they don't need to access necessarily the full data set. You basically, you create the JSON as it were to go in with all the data that the API needs. The API executes gives you a result. It's really quite a simplistic approach, but I think it gives you that isolation. Because once you start putting your fingers into the data, you've got a slight danger that someone, hey, this is an interesting API, why don't I try it? Oh, I just hit 50 petabytes sequentially. HP has had a, pre-HPE, had a long legacy of excellence with systems management. We haven't really talked about what you're doing with predictive analytics in that area, but is that a ripe area for you to help organizations manage? It is, I mean, we're working on a couple of things in that area. We're working on a thing called operations analytics, which really takes the analytics framework and applies it to all the log files that get produced in a data center to help people define problems more quickly and then resolve them more quickly. And I think it's actually showing some success. And then our product service anywhere, I think is a great product for trying to transform the service management business by using analytics. In other words, let's stop creating service management tickets and let's use the technology to help people solve their own problems and get solutions to what they need to do without any intervention. And there's a third thing we're working on right now, which I think is a lot of fun. I know it's becoming quite fashionable, but I think we're again, quite a way ahead of it, which is intelligent chatbots. So that actually, as you're working through, we were looking at the way in which IT operations professionals do their actual job. And we create these beautiful screens and we put huge amounts of energy into the GUIs and the charts. And we find most of the time they work through chat channels. They've got Slack, a Slack chat channel going and they're ignoring the screen. So we're trying to now put intelligent agents within the Slack chat channels that can actually interface to the backend software. So that if you have an IT problem, you set up a Slack channel, the Slack channel is talking to real people, but also to automated bots that are creating the incident, collecting the log files, providing data on it and so on. It's really quite exciting. I know a lot of people are talking about it. All the things that can be automated so you can focus on things that humans can do. No, exactly. That's great. But also using a mechanism they like, which is chat channels. Yeah, of course. Yeah. Robert, always a pleasure to have you on. Thanks so much for coming by. Thank you for your time. Our community loves hearing from you. You're a clear thinker and really appreciate your insights. I really appreciate it. Thank you. All right, keep right there. We'll be back with our next guest right after this short break. This is The Cube.