 Everybody, we're back. Welcome to day three of HP Discover. This is theCUBE. I'm Dave Vellante. theCUBE is a live mobile studio. We go out to events. We land in, we extract the signal from the noise. We broadcast all day long. We've been here for three days now at HP Discover. This is HP's big European event. It mirrors the event that Hewlett Packard has in Las Vegas in June. HP Discover, it's the big customer event. In fact, HP does quite a bit of business in Europe, I would say in some ways this event has actually a little bit even more energy, believe it or not, than what you saw in Las Vegas. So big European contingent, obviously a lot of diversity and a lot of energy. So we're going to kick the day off today. Two great guests, Shilpa Luanda is here. She's the vice president at Vertica and Brian Weiss, VP at Autonomy, both of course HP companies. Folks, welcome to theCUBE. Good to see you again. Thanks, David. It's great to be here. Thanks for having us. So Shilpa, let's start with you. First of all, thanks for staying a little later. I know that a lot of your colleagues are hitting the road, but we're hanging tough here in theCUBE. So thanks for hanging with us. But give us the quick update from your perspective. What's going on at the show for you and for Vertica? So it's been a great show for us. We announced Vertica 7 a few weeks ago and we're just starting to see people's reactions to it and it's been just unbelievable for us. You might have seen that yesterday we had the Facebook CIO on stage with George Garifa talking about what impact Vertica's had on their business. And so it's been just a great show with lots of good customers talking on our behalf, as well as a lot of good customer interest in Vertica 7, especially with FlexZone, which enables us to do not just structured data anymore, but also structured and semi-structured data. And so it really is a huge leap for us. Yeah, it was a great testimonial from the Facebook guys. It was quite an amazing demo. For those of you who didn't see it, it's actually worth going back and checking out the replay. But the gentleman from Facebook put up what we thought was a map, right? But it was the points of light of all the users of Facebook and all their connections. And it essentially formed a map and you could very clearly see the map of the world and the continents. And he said, there was no map when we put this together. This is just points of light. It was pretty inspiring. Now, Brian, from the autonomy standpoint, we're starting to see this sort of monolithic, massive chunk of software being broken into pieces and web services created that HP is beginning to leverage across the organization. That's got to be really exciting. First of all, that you can actually do that with the architecture. And second of all, that it can now find that thousand points of light, I guess pun intended, across the organization. So give us the quick autonomy update. Well, you know, this is one of the benefits that we have coming into HP because you can take the resources of a company like HP and a platform like Idle, which really understands human information, right? And it helps us get insight out of that data in a way that most analytical tools can't do. But as you point out, the problem is that most companies have to invest in the platform. They've got to install it, et cetera. And what we've done over the past year is invest heavily in making it possible now for developers to access Idle on demand. So we're hosting it. And instead of having to put up software and index data and run your applications out, you can just ask the question. So for example, something like image recognition or sentiment analysis, for example. If I'm a coder, I can take that block of code and say, look, I'm going to forward you this image. Tell me where the faces are. And I can build that right into my applications right now. So the functions inside Idle can now be exposed very quickly and easily. We've got some great examples where we're building apps in a day based on a lot of these robust functions. Yeah, so when I look at HP's software strategy, we heard George Kadeefa yesterday and he sort of laid out the portfolio and the capabilities. It seems to me that the big growth area, I mean, it's pretty obvious, is the whole big data platform. However you want to sort of define that. And that's really what Haven is all about. But there's this big white space that we talk about all the time and sort of an algorithmic bond. There's not a lot of big data apps. And big data apps are hard. And why is that? Well, you need domain expertise, you need data science, you need technology, you need tools to build apps. And they've been lacking, haven't they? Well, I think that's true. I think what's helping my experience with this is that the problems that we're really focused on in big data are the ones that are on the extremes. It's the intense amount of machine generated structured data and how do you do that? You can't really do that with normal relational database tools. It can't happen. Or human information, what does it mean? How do you get the sentiment? How do you get that understanding that you and I get very easily from reading or looking at something? But the computer doesn't understand. So we live at those extremes and that's really where the most value is and we're putting a lot of effort into making a platform that you can develop. So Haven, right? Let's just break it down. So it stands for Hadoop, Autonomy, Vertica, Enterprise Security, and N. The N is for N apps. N apps. And then if you add enterprise services and enterprise group to that, you're in heaven. Right? So I'm in here first. Okay, so the idea is I'm in application development heaven for some of these emerging big data apps. Talk a little bit more about Haven platform. Shilpa, just from an architectural standpoint, what should we know about it? Yeah, so Haven basically gives you all the building blocks you need to essentially analyze all of your information. So it's not just structured data or human information, but it gives you the tools to really put them together. So to give you an example, let's say you're a telco and you're trying to find out which of the customers that matter to you, right? So there's a lot of information you can collect from your call detail records and so on. And a lot of that is structured or semi-structured information. Vertica can do great with that. And we have, you know, I would say seven out of 10 telcos in the US are our customers. So now what if you have call logs from your call center and you want to basically find out what are people really talking about your products when they call your customer support? Like, are they happy or are they experiencing dropped calls and things like that? That sentiment that you need, that's human information. When people call you, like you want to identify things like what was the inflection of this person's voice? What were the words they dropped or, you know, are they happy customers? You really want to focus on your unhappy customers and those are the people you want to send your emotions and so on. So this churn analysis is a major problem for telcos. And that's not a Vertica sweet spot. And that's not a Vertica only solution. It's a common, you need to put, you know, the human data together with the machine data here to really get you that extra insight. And these, this processing, all of your information is I think what's going to be the edge. Once everybody starts doing big data, then what keeps one company different from another one? It's what you can do with all the data. I mean the point is to be able to get insight, right? Insight out of information and be able to do that. You have to be able to understand it in human terms and be able to handle a massive volume. So it's great to know that you've got, you know, you get a 10 million calls between 10 and 11 o'clock and you've got a lot of data points about every phone call. But were they mad? What do they talk about? You can't get that data point without understanding and analyzing the actual call itself. And of course we can't listen to these calls. The computer has to do it for us. So if I were a head of application development at a telco, let's say pre-haven, and I wanted to solve this problem, I got relational databases, I maybe have some column stores, I maybe have some way to analyze unstructured data. How would I go about actually building those apps and how would haven change my world? Yeah, so previously, I mean, I think this is the kind of app that I don't think people normally do today. Yeah, it's too hard to just say, this is like, you don't even know it's possible. Wouldn't it be great if? So essentially they would do it in silos. And then try to put the information together and squint through it and. So you might be able to find, you know, like you might be able to do some analysis on your call center logs to find, you know, how many customers are calling or something like that. But that's metadata, right? That's not leveraging the data in your human information, right? So now, on the other hand, you know, Vadeka can do a lot of analysis on the machine data on the call logs and on the call detail records type of analysis. But that doesn't again give you the human, you know, so the combination is, I don't, I think those are the applications that so far haven't been built. And that's what Haven gives you. And look in that equation, what idle autonomous technology doing is telling you the ideas that are in that phone call. What did you and I talk about? It's telling me the main concepts, it's telling me the things that are related to them and it's telling me whether I was mad or not in fairly subtle details. And there's a lot of data points that come out of that. And so now I can use those two together and build an analytical app that says, these guys are mad about this topic and this particular geography and most of them are getting drop calls here. Those sort of things are now visible immediately. So, so Haven is an out of the box platform. It comes with a Hadoop distribution of my choice. Is that right? Or maybe it's Apache, right, Hadoop. And then I get Vertica, I get Autonomy, I get my security piece, and then I get development tools. Even if you don't need all of them. So our focus is really, look at your use case and figure out which blocks do you want to use to build. So for example, there's another example could be Vertica and ArcSight together for a security analytics application. So you might have, you might just buy those two components if you need those other, only two Vertica and Hadoop is another combination that's very common. So what I'm trying to get to is technically or maybe not even so much technically, but what do those building blocks give me? It's from in terms of developer productivity that I wouldn't have otherwise. In other words, they work together. What does that mean? Architecturally or technically, and then ultimately productively? So essentially gives you APIs, right? So it gives you APIs as well as it gives you sort of, think of it as reference implementation. So we have the developer portal as examples of how you would put these two things together. What are the APIs you would use to put these things together? So the other thing you get David is connectivity. So in order to be able to get the data, whether it's human information or structured data or logs coming off of your ArcSight, for example, you need to have connectors. So we have connectors from all the libraries, whether they come from autonomy or ArcSight that actually allow you to get to that. In fact, when you're using Hadoop, you're using, in this framework, you can be using the autonomy extraction to be able to put the data into it. So in fact, the ETL process itself uses the toolkits that we have. Everything from the attraction, everything from the connectors to the extraction to what you do with that data. Okay, I think to your point, the idea is people are coming and saying, start with the use case. What kind of data do you have and what would you love to get out of it? So for example, in the security use case, you've got Vertica, which is handling all those massive logs, right? All the events. But then we also take idle and we couple that on top of it and say, if there's any human information in this event, what's it about? So I can track the fact that at five o'clock I sent a document out of the environment. That's in the log. Here's the question for you. What was the document about and do I care? Was that sensitive IP, or was it just a recipe, or was I sending a photo to the shelter of my kids? Now idle can come in and say, I can enrich that insight into what's going on in your environment, not just what happened, what did it mean? And so that toolkits available now and we can actually couple both idle and arc side together and deliver it out to the field. Okay, so I can have my developers spend time focusing on whether it's implementing the policy or driving revenue or whatever it is and not have to worry about the integration. So you're essentially giving me programmable software infrastructure right out of the, I called it out of the box, but it's not coming in a box, but yeah. Okay, great. That's good. Now let's go through some examples. You guys have announced the digital marketing hub. That's one example. Let's go through some. Yeah, so we talked about the digital marketing hub. This is a framework which is hosted as a service that we can provide and it allows marketers to take multiple different data points about their customer. It might come from their call center, it might come from their customer databases, it might come from their marketing platforms or whatever it might be, multiple different touch points and aggregate that information and allow you to run campaigns in real time. So historically what you'd have to do with marketing is get information from my various applications whether it's Adobe or Marketo or whatever it might be and find out where the customer's touching, what are we doing, how is our campaign working and you have to come back and analyze it and then you have to come back and adjust the campaign and then you have to see if that one's working. So there's long hang time to be able to test and analyze and discover whether you spent money in the right place. Well now all that information is aggregated in one view and we can actually then take it and drive what you're learning about your customers in real time into the campaign itself directly. So I can get campaigns to launch automatically based on dynamic segmentation of my customer brace now. And by the way, this runs on idle and Vertica's on the back end of it. So this is a Haven-based application that we at HP have delivered to the market. So it's built on Haven, Haven's the platform. Digital Marketing Hub is the sort of application that you've built and it supports other applications like Marketo, you mentioned Adobe. Well see, those are data sources, right? So I know a lot about my customer based on any number of different applications which tell me, my call center for example, right? My touch points to you, but normally those are separated. And in order to figure out what I want to sell to and who my segment is and how I'm going to campaign to them, I usually need to do a manual aggregation of it. And now I can do that in real time. And how would people do it prior to Digital Marketing Hub? They'd do it in- Well you'd collect all this and then you'd sit down and put some spreadsheets together and say, you know what? That campaign didn't work very well. We probably didn't target it to the right people and here's what we know about the manual exercise. So it's Excel, Hal, and pivot tables and the like. You know this is the thing that's really interesting and is concept of connected intelligence. People always talk about I want to get value of my data. I want to get insight out of my data. What we're also seeing people focus on is I want to get value out of different data types. I want to see across this one, this one, this one and this one and not, the big data is actually going deep on one particular type of data. It's what you get when you go across all of them. As an insight. And that, you know, we're- It's eliminating silos, right? So another example is the operations analytics product. It's a great example of a use of, you know, different log analysis technology that we have in different products in HP. But Vertica is the engine that provides that analytics. So the data goes into Vertica. But what that allows you to do things like take your network operations data together with your security data and do correlations across those two domains. That even in an IT organization, sometimes those data is owned by different parts of that organization. So they tend to be in silos. If you put the two things together, then you can actually almost think of that data as an addition to your business data, right? Because you might find interesting correlations that you might not find just with your data warehouse that you like, why did I get a spike in sales here? Or if you're a network operator, like why did I suddenly get this spike in my network, like I have no idea what was the problem. Did I do something wrong? And you might find that, oh, people just, you know, bought more of my product because it was raining outside. And so they just came in and drove and bought, you know, hamburgers or something like that. And so, but unless you have the business, the data warehouse or whatever it is, your business sales data correlated with your network data, you would never know, right? So these are problems that sort of cross domains across the business data, machine data. And then if you add human data to it, that's when you get insights that are competitively valuable. So I heard at least two use cases there. One was more sort of what's happening with my sales. The other was what's happening with my IT infrastructure. So are those common use cases that you expect or things that you've talked to customers about? Let's take the latter. Yeah, so I'll give you another example. So think about, you know, you're doing, security is a problem for everybody. All CSOs are thinking about, you know, how to keep my enterprise secure. But there are these threats that really require, so your security software will handle things that are events that happen in a short period of time. But then there might be behaviors you need to identify over long periods of time. So for example, Shilpa is an engineer and she, you know, logs into these systems. But one fine day, Shilpa logged into a sales system. Right? That behavior you can only identify myself as an identity over a long period of time. If you were able to, you know, correlate my activities and establish an identity for me and associate it with my behavior, that type of analysis requires an engine like Vertica, right? So because your same products will do that because that's not what they're meant for. And so these are the kinds of things that you can do by putting, you know, the power of Haven. You were mentioning machine data before. So we've been doing a lot of work in SiliconANGLE Wikibon on the whole, whether you call it internet of things and we did this big piece of work with GE recently on the industrial internet. John Furrier hosted a panel with Jeffrey ML. It was really interesting to hear folks from, you know, transportation, airlines, oil and gas, just talk about the types of data that's being generated by machines. And of course you start thinking about, well, where's that data going to go? How's it going to be analyzed? What kind of conversations are you having with customers in that regard? Are you having them at this point in time and what are they asking you about? And how do you see HP participating in that whole trend? So, as you know, we recently announced FlexZone, which is Vertica's ability to ingest semi-structured data. What we find is a lot of this data, sensor data, machine data, things like that. Even though people think of it as unstructured, there's a lot of structure in that data. It's semi-structured and we find that, you know, making it possible for people to easily ingest and analyze that data is a huge leap for a lot of people because, you know, this data stays dark. You know, a lot of this data that just people throw away. Like there's incredible amount of machine log data that people throw away because there's, you know, they don't know what to do with it. Or they might keep it for some period of time, just for troubleshooting or something. But it's the long-term retention and being able to find insights over time. And across many people. A lot of that data is deadly boring. Like, you're going to have, you know, billion clicks and whatever these machines are doing and you care about the billionth and one that's a little bit different. Because you got to keep the other ones in order to be able to understand the difference. And so you're collecting a lot of noise in order to be able to get that little bit of signal. And, you know, normal relational technologies don't, just can't handle the volume. They're not for the right job, right for the job. So what's the potential of that space? Is the potential there to really start to take waste out of industries? Whether it's to- Speed to insight. Okay, speed to insight, but, so like you said, Brian, there's- Yes, you take the data exhaust. You take the data exhaust. Exhaust, like all these machines. But as you point out, there's so much- Yeah, use that to find trends that are like across populations and things like that, right? Which is not something that, you know, it's like your refrigerator, the next generation of it would have, you know, four, eight hundred sensors that'll tell GE or whoever the company that makes them all about like, what is this person's usage pattern? And you know, what is, and that, if you can correlate that with their energy use, then there's just so much, so if you think about it, you know, I would say the next sort of frontier for analytics would be like data, products that are really built using analytics, right? So it's not just about understanding your own business or your own customers. That's the Web 2.0 model, right? Understand your customers, but like make your products like collect analytics about themselves and then use that to improve the product experience for your customer. That's like, I think the next level of where you're seeing, you know, the internet of things and all that. I'll tell you, I think that customers out there are starting to understand and really realize that you can get a computer to understand human information. And normally, I think that's an outside of the box idea for people. The idea that I can have a computer listen to this conversation, actually listen to all of the conversations in the room. A computer, okay, we can have machines. Tell me what people are talking about. Things that you would ordinarily need people to do. Like, look, if I'm, for example, if you're a lawyer and you gotta find relevant documents for a case, the judge does not say, find me the one with the word bridge in it. He says, find me the relevant documents. Now you can pay people to do that or you can ask a machine to say, by the way, read all of these and tell me the main ideas or listen to all these phone calls and tell me what's in them. And that's where we get into sentiment analysis, social analysis, really valuable for customers like NASCAR, right? All that data is coming in and there's the tweets and how fast they're coming in and when they're happening. What are people saying? Yeah, so this is what I wanted to get to is you've mentioned, Brian, that so much of this information is deadly boring. So, and it's just such a huge volume. A human just can't sift through that, no way. So what you're putting forth is a vision where machines start making decisions based on maybe human injected policies, you know? And maybe a human's watching to make sure something's not going wrong. But the machines are actually affecting whether it's the utilization of a turbine or what gets packed on which freighter or whatever it is to increase utilization. So what happens to the humans? You know, there's the old bromide, you can't take humans out of the equation, humans are the last mile. The humans just do more productive things. What do you guys envision there? Well, two things I'll say, there's automation. So to your point, the computer can actually do things and make decisions. Say for example, you see this in fraud detection all the time, right? If, whether it's something I've written or something I've done, if it triggers a fraud alert, I can block it, right? And the funny thing about, for example, insider trading, when people are going to do that and they're going to write an email about insider trading, they don't use the phrase insider trading. They say something else. And so you have to analyze the subtlety of that, but also you can block that. Or the other one is insecurity, video. If you're analyzing video, and maybe we have our technology doing this in the field from drones, what's important and what's not? You've got thousands and thousands of hours of noise, but the analyst needs to see the one event which matters. And they cannot sit there and watch thousands. They need the truck-shaped object is going to a place where it's not, so you're right, it'll give you, it'll make you more efficient and it'll bring the signal out of the noise so you can either automate a process or being much more valuable and efficient with your time. Insider trading example is interesting. I was just reading an article, I forget what magazine it was in recently. And they were just saying that, that they're not saying, I'm doing some insider trading, they're talking about takeout, I want Thai food today, whatever it is. And when they see that pattern occur, some anomaly about talking about food, in a consistent pattern, and then they identify it with another person, and then that's how they're identifying trading fraud, which is fascinating. There's no way a human would ever be able to do that. Yeah, I'll give you another example of, that impacts like me, this device I'm wearing, Fitbit, right? Yeah. This is why analytics touches me personally, right? Like, I walk extra just to get the five dots at the end of the day, every day. And it's because it tells me, hey, you only have a thousand more steps to go, and then you'll do it, right? And so, just think of the impact that these devices can have on people individually, not just people understanding, okay, you know what? People socially together, you know. If you just tell people a little more data, I think we all become, our lives get better. We were at, I'll tell you a quick story, we were at Hadoop Summit in June, and we were interviewing, I forget his last name, Skye something, I remember his first name, his first name was Skye, and that actually was his given name. He's out there somewhere watching probably live. So he was an Olympic athlete, a velodrome athlete, and he was very upset, of course, with all the doping that was going on. And you know, he was a clean athlete. And he worked with, I think it was maybe Datamere or something, I can't remember the company, to use the data from a Fitbit to try to identify temperatures when he's sleeping, you know, food intake, calories, and everything else. And he were able to dramatically increase performance through data. So that's a fascinating example. The other day I was at a Christmas event, and I was talking just, you know, talking to somebody, and they're a researcher in education. And one of the interesting examples they were looking at is if you give a child or a student, let's say a high school student, information about, you ask them a simple question at the end of a test saying, what resources did you use in filling out these answers? And then you do a survey across all the students and give the students back analytics to show that if they used all that, so for example, did they ask their parents questions? Did they go use the internet? Did they use Google? Did they use Wikipedia? Like what resources did they use? And so if you can show them a positive correlation between the resources they used and their scores on the test, the students apparently go look for more resources. So even like children will make use of analytics if you give them to it without, they won't even know that it's analytics, but that's the power of data, right? I also think there's a great example from the CIO of Facebook and Shilpa and I were just talking about this before we came in here is what they're doing there and you can tell us a little more about it, but the fact that they can get live, almost real time information about who's buying what and where when it ordinarily took them a day, the really interesting point he's made is that we can build a whole different class of applications now because in real time I know what's happening. So they're going to be able to go back in and deliver because of the speed of the insight a whole different class of applications at Facebook because of that. If you have to operate in a batch, way today to get an answer, I can't build an application that's going to react to that and give you a different price or give you a different offer or optimize by marketing. I can't do that. And the exponential factor there, as he said, was the relationships, because it's not just an individual, you can then begin to market to that individual's social graph. The speed of what they can do with that and that was, I mean, what's going to make a whole different class and a whole universe of applications? So it's early days. Obviously you guys are really, this is sort of just getting started here. I mean, you announced Haven, now you're starting to improve the platform. What do you guys expect that we should be watching? Brian, we'll start with you, just as observers. What should we be observing as the uptake in this market? What are they going to give you? The key parameters and indicators that we're really starting to take off and see momentum here. Well, I think you're going to see very rapidly extremely high value use cases coming out publicly. I mean, the Facebook one's a good example, but places where we're providing insights that traditional technologies couldn't do, whether it's coupling an insight into the data with the security event itself. And so you'll start to see those kinds of uses come out of our customer. They're digging into them right now around more and harder, difficult types of information, understand audio and video. So you will start to see those come out and I think there will be some that are mind-blowing. And that uptake is, it's happening faster than I would've expected. Problems that you really wouldn't even think about solving before because it was too hard. That's what you're going after. Anything you'd add to that? The second thing I think that you will see and you're already seeing in this conference a lot is partner activity. So we've seen an unbelievable response from our strategic partners, our service partners, our resellers, technology partners, and really the whole ecosystem around Haven. And so between the developers and the partners, I think we really believe that we have something here. No other company has that whole gamut of tools to handle really all the data. And it's nice to see the uptake by developers as well. So we introduced idle demand. On demand we've got Haven developer tools. The uptake by the folks who want to build applications based on this level of insight is really high. Now you've been asked to SDK, right? So big opportunity for HP obviously in the software group is to go after the developers. We talked to Robert Young-Johns about the potential of having a developer conference. Really excited about that. So anyway, congratulations for being in the hottest space in the company. You must be really excited about that. And it's a pleasure for us to be sort of watching that progress. So Shilpa and Brian, thanks very much for coming on theCUBE. My pleasure, David. All right, keep it right there. Everybody will be back with John Furrier with our next guest right after this. This is theCUBE. We're live from HP Discover Barcelona.