>> Announcer: Live from San Jose, California, it's the Cube, covering Big Data Silicon Valley 2017. (upbeat music) >> Hello and welcome to the Cube's special coverage of Big Data SV, Big Data in Silicon Valley in conjunction with Strata + Hadoop. I'm John Furrier with George Gilbert, with Mickey Bonn and Peter Burns as well. We'll be doing interviews all day today and tomorrow, here in Silicon Valley in San Jose. Our next guest is Amit Walia who's the Executive Vice President and Chief Product Officer of Informatica. Kicking of the day one of our coverage. Great to see you. Thanks for joining us on our kick off. >> Good to be here with you, John. >> So obviously big data. this is like the eighth year of us covering, what was once Hadoop World, now it's Strata + Hadoop, Big Data SV. We also do Big Data NYC with the Cube and it's been an interesting transformation over the past eight years. This year has been really really hot with you're starting to see Big Data starting to get a clear line of sight of where it's going. So I want to get your thoughts, Amit, on where the view of the marketplace is from your standpoint. Obviously Informatica's got a big place in the enterprise. And the real trends on how the enterprises are taking analytics and specifically with the cloud. You got the AI looming, all buzzed up on AI. That really seized, people had to get their arms around that. And you see IoT. Intel announced an acquisition, $15 billion for autonomous vehicles, which is essentially data. What's your views? >> Amit: Well I think it's a great question. 10 years have happened since Hadoop started right? I think what has happened as we see is that today what enterprises are trying to encapsulate is what they call digital transformation. What does it mean? I mean think about it, digital transformation for enterprises, it means three unique things. They're transforming their business models to serve their customers better, they're transforming their operational models for their own execution internally, if I'm a manufacturing or an execution-oriented company. The third one is basically making sure that their offerings are also tailored to their customers. And in that context, if you think about it, it's all a data-driven world. Because it's data that helps customers be more insightful, be more actionable, and be a lot more prepared for the future. And that covers the things that you said. Look, that's where Hadoop came into play with big data. But today the three things that organizations are catered around big data is just a lot of data right? How do I bring actionable insights out of it? So in that context, ML and AI are going to play a meaningful role. Because to me as you talk about IoT, IoT is the big game changer of big data becoming big or huge data if I may for a minute. So machine learning, AI, self-service analytics is a part of that, and the third one would be big data and Hadoop going to cloud. That's going to be very fast. >> John: And so the enterprises now are also transforming, so this digital transformation, as you point out, is absolutely real, it's happening. And you start to see a lot more focus on the business models of companies where it's not just analytics as a IT function, it's been talked about for a while, but now it's really more relevant because you're starting to see impactful applications. >> Exactly. >> So with cloud and (chuckles) the new IoT stuff you start to say okay apps matter. And so the data becomes super important. How is that changing the enterprises' readiness in terms of how they're consuming cloud and data and what not? What's you're view on that? Because you guys are deep in this. >> Amit: Yep. >> What's the enterprises' orientation these days? >> So slight nuance to that, as an answer. I think what organizations have realized is that today two things happened that never happened in the last 20 years. Massive fragmentation of the persistence layer, you see Hadoop itself fragmented the whole database layer. And a massive fragmentation of the app layer. So there are 3,000 enterprise size apps today. So just think about it, you're not restricted to one app. So what customers and enterprises are realizing is that, the data layer is where you need to organize yourself. So you need to own the data layer, you cannot just be in the app layer and the database layer because you got to be understanding your data. Because you could be anywhere and everywhere. And the best example I give in the world of cloud is, you don't own anything, you rent it. So what do you own? You own the darn data. So in that context, enterprise readiness as you came to, becomes very important. So understanding and owning your data is the critical secret sauce. And that's where companies are getting disrupted. So the new guys are leveraging data, which by the way the legacy companies had, but they couldn't figure it out. >> What is that? This is important. I want to just double-click on that. Because you mentioned the data layer, what's the playbook? Because that's like the number one question that I get. >> Mm-hmm. >> On Cube interviews or off camera is that okay, I want to have a data strategy. Now that's empty in its statement, but what is the playbook? I mean, is it architecture? Because the data is the strategic advantage. >> Amit: Yes. >> What are they doing? What's the architecture? What are some of the things that enterprises do? Now obviously they care about service level agreements and having potentially multicloud, for instance, as a key thing. But what is that playbook for this data layer? >> That's a very good question, sir. Enterprise readiness has a couple of dimensions. One you said is that there will be hybrid doesn't mean a ground cloud multicloud. I mean you're going to be in multi SAS apps, multi platform apps, multi databases in the cloud. So there is a hybrid world over there. Second is that organizations need to figure out a data platform of their own. Because ultimately what they care for is that, do I have a full view of my customer? Do I have a full view of the products that I'm selling and how they are servicing my customers? That can only happen if you have what I call a meta-data driven data platform. Third one is, boy oh boy, you talked about self-service analytics, you need to know answers today. Having analytics be more self-serving for the business user, not necessarily the IT user, and then leveraging AI to make all these things a lot more powerful. Otherwise, you're going to be spending, what? Hours and hours doing statistical analysis, and you won't be able to get to it given the scale and size of data models. And SLAs will play a big role in the world of cloud. >> Just to follow up on that, so it sounds like you've got the self-service analytics to help essentially explore and visualize. >> Amit: Mm-hmm. >> You've got the data governance and cataloging and lineage to make sure it is high quality and navigable, and then you want to operationalize it once you've built the models. But there's this tension between I want what made the data lake great, which was just dump it all in there so we have this one central place, but all the governance stuff on top of that is sort of just well, we got to organize it anyway. >> Yeah. >> How do you resolve that tension? >> That is a very good question. And that's where enterprises kind of woke up to. So a good example I'll give you, what everybody wanted to make a data lake. I mean if you remember two years ago, 80% of the data lakes fell apart and the reason was for the fact that you just said is that people made the data lake a data swamp if I may. Just dump a lot of data into my loop cluster, and life will be great. But the thing is that, and what customers of large enterprises realized is they became system integrators of their own. I got to bring data, catalog it, prepare it, surface it. So the belief of customers now is that, I need a place to go where basically it can easily bring in all the data, meta-data driven catalog, so I can use AI and ML to surface that data. So it's very easy at the preparation layer for my analysts to go around and play with data and then I can visualize anything. But it's all integrated out of the box, then each layer, each component being self-integrated, then it falls apart very quickly when you want to, to your question, at an enterprise level operationalize it. Large enterprises care about two things. Is it operationalizable? And is it scalable? That's where this could fall apart. And that's what our belief is. And that's where governance happens behind the scenes. You're not doing anything. Security of your data, governance of their data is driven through the catalog. You don't even feel it. It's there. >> I never liked the data lakes term. Dave Vellante knows I've always been kind of against, even from day one, 'cause data's more fluid, I call it a data ocean, but to your point, I want to get on that point because I think data lakes is one dimension, right? >> Yeah. >> And we talked about this at Informatica World, last year I think. And this year it's May 15th. >> Yes. >> I think your event is coming up, but you guys introduced meta-data intelligence. >> Yep. >> So there was, the old model was throw it centralized, do some data governance, data management, fence it out, call, make some queries, get some reports. I'm over simplifying but it was like, it was like a side function. You're getting at now is making that data valuable. >> Amit: Yep. >> So if it's in a lake or it's stored, you never know when the data's going to be relevant, so you have to have it addressable. Could you just talk about where this meta-data intelligence is going? Because you mentioned machine learning and AI. 'Cause this seems to be what everyone is talking about. In real time, how do I make the data really valuable when I need it? And what's the secret sauce that you guys have, specifically, to make that happen? >> So that, to contextualize that question, think about it. So if you. What you don't want to do is keep make everything manual. Our belief is that the intelligence around data has to be at the meta-data level, right? Across the enterprise, which is why, when we invested in the catalog, I used the word, "It's the google of data for the enterprise." No place in an enterprise you can go search for all your data, and given that the fast, rapid-changing sources of data, think about IoT, as you talked about, John. Or think about your customer data, for you and me may come from a new source tomorrow. Do you want the analyst to figure out where the data is coming from? Or the machine learning or AI to contextualize and tell you, you know what, I just discovered a great new source for where John is going to go shop. Do you want to put that as a part of analytics to give him an offer? That's where the organizing principle for data sits. The catalog and all the meta-data, which is where ML and AI will converge to give the analyst self-discovery of data sets, recommendations like in Amazon environment, recommendations like Facebook, find other people or other common data that's like a Facebook or a LinkedIn, that is where everything is going, and that's why we are putting all our efforts on AI. >> So you're saying, you want to abstract the way the complexity of where the data sits? So that the analyst or app can interface with that? >> That's exactly right. Because to me, those are the areas that are changing so rapidly, let that be. You can pick whatever data sets based on what you want, you can pick whichever app you want to use, wherever you want to go, or wherever your business wants to go. You can pick whichever analytical tool you like, but you want to be able to take all of those tools but be able to figure out what data is there, and that should change all the time. >> I'm trying to ask you a lot while you're here. What's going to be the theme this year at Informatica World? How do you take it to the next level? Can you just give us a teaser of what we might expect this year? 'Cause this seems to be the hottest trend. >> This is, so first, at Informatica World this year, we will be unveiling our whole new strategy, branding, and messaging, there's a whole amount of push on that one. But the two things that will be focused a lot on is, one is around that intelligent data platform. Which is basically what I'm talking about. The organizing principle of every enterprise for the next decade, and within that, where AI is going to play a meaningful role for people to spring forward, discover things, self-service, and be able to create sense from this mountains of data that's going to sit around us. But we won't even know what to do. >> All right, so what do you guys have in the product, just want to drill into this dynamic you just mentioned, which is new data sources. With IoT, this is going to completely make it more complex. You never know what data's going to be coming off the cars, the wearables, the smart cities. You have all these new killer use-cases that are going to be transformational. How do you guys handle that, and what's the secret sauce of? 'Cause that seems to be the big challenge, okay, I'm used to dealing with data, its structure, whether it's schemas, now we got unstructured. So okay, now I got new data coming in very fast, I don't even know when or where it's going to come in, so I have to be ready for these new data. What is the Informatica solution there? >> So in terms of taking data from any source, that's never been a challenge for us, because Informatica, one of the bread and butter for us is that we connect and bring data from any potential source on the planet, that's what we do. >> John: And you automate that? >> We automate that process, so any potential new source of data, whether it's IoT, unstructured, semi-structured, log, we connect to that. What I think the key is, where we are heavily invested, once you've brought all that. By the way, you can use Kafka Cues for that, you can use back-streaming, all of that stuff you could do. Question is, how do you make sense out of it? I can get all the data, dump it in a Kafka Cue, and then I take it to do some processing on Spark. But the intelligence is where all the Informatica secret sauce is, right? The meta-data, the transformations, that's what we are invested in, but in terms of connecting anything to everything? That we do for a living, we have done that for one quarter of a century, and we keep doing it. >> I mean, I love having a chat with you, Amit, you're a product guy, and we love product guys, 'cause they can give us a little teaser on the roadmap, but I got to ask you the question, with all this automation, you know, the big buzz out in the world is, "Oh machine learning and AI is replacing jobs." So where is the shift going to be, because you can almost connect the dots and say, "Okay, you're going to put some people out of work, "some developer, some automation, "maybe the systems management layer or wherever." Where are those jobs shifting to? Because you could almost say, "Okay, if you're going to abstract away and automate, "who loses their job?" Who gets shifted and what are those new opportunities, because you could almost say that if you automate in, that should create a new developer class. So one gets replaced, one gets created possibly. Your thoughts on this personnel transformation? >> Yeah, I think, I think what we see is that value creation will change. So the jobs will go to the new value. New areas where value is created. A great example of that is, look at developers today, right. Absolutely, I think they did a terrific job in making sure that the Hadoop ecosystem got legitimized, right? But in my opinion, where enterprise scalability comes, enterprises don't want lots of different things to be integrated and just plumbed together. They want things to work out of the box, which is why, you know, software works for them. But what happens is that they want that development community to go work on what I call value-added areas of the stack. So think about it, in connected car, they're working with lots of customers on the connected car issue, right? They don't want developers to work on the plumbing. They want us to kind of give that out of the box, because SLA is operational scale, and enterprise scalability matters, but in terms of the top-layer analytics, to make sure we can make sense out of it, that's what they're, that's where they want innovation. So what you will see is that, I don't think the jobs will go in vapor, but I do think the jobs will get migrated to a different part of the stack, which today it has not been, but that's, you know, we live in Silicon Valley, that's a natural evolution we see, so I think that will happen. In general in the larger industry, again I'd say, look, driverless cars, I don't think they've driven away jobs. What they've done is created a new class of people who work. So I do think that will be a big change. >> Yeah there's a fallacy there. I mean with the ATM argument was ATM's are going to replace tellers, yet more branches opened up. >> That's exactly it. >> So therefore creating new jobs. I want to get to the quick question, I know George has a question, but I want to get on the cost of ownership, because one of the things that's been criticized in some of these emerging areas, like Hadoop and Open Stack, for instance, just to pick two random examples. It's great, looks good, you know, all peace and love. An industry's being created, legitimized, but the cost of ownership has been critical to get that done, it's been expensive, talent, to find talent and deploying it was hard. We heard that on the Cube many times. How does the cost of ownership equation change? As you go after these more value, as developers and businesses go after these more value-creating activities in the Stack? >> See look, I always say, there is no free lunch. Nothing is free. And customers realize that, that open source, if you completely wanted to, to your point, as enterprises wanted to completely scale out and create an end-to-end operational infrastructure, open source ends up being pretty expensive. For all the reasons, right, because you throw in a lot of developers, and it's not necessarily scalable, so what we're seeing right now is that enterprises, as they have figured that this works for me, but when they want to go scale it out, they want to go back to what I call a software provider, who has the scale, who has the supportability, who also has the ability to react to changes and also for them to make sure that they get the comfort that it will work. So to me, that's where they find it cheaper. Just building it, experimenting with that, it's cheaper here, but scaling it out is cheaper with a software provider, so we see a lot of our customers when we start a little bit experimenting to developers, downloading something, works great, but would I really want to take it across Nordstrom or a JP Morgan or a Morgan Stanley. I need security, I need scalability, I need somebody to call to, at that point on those equations become very important. >> And that's where the out of box experience comes in, where you have the automation, that kind of. >> Exactly. >> Does that ease up some of the cost of ownership? >> Exactly, and the talent is a big issue, right? See we live in Silicon Valley, so we. By the way, Silicon Valley hiring talent is hard. Just think about it, if you go to Kansas City, hiring a scholar developer, that's a rare breed. So just, when I go around the globe and talk to customers, they don't see that talent at all that we here just somehow take for granted. They don't, so it's hard for them to kind of put their energy behind it. >> Let me ask. More on the meta-data layer. There's an analogy that's come up from the IIoT world where they're building these digital twins, and it's not just GE. IBM's talking about it, and actually, we've seen more and more vendors where the digital twin is this, it's a digital representation now of some physical object. But you could think of it as meta-data, you know, for a physical object, and it gets richer over time. So my question is, meta-data in the old data warehouse world, was we want one representation of the customer. But now it's, there's a customer representation for a prospect, and one for an account, and one for, you know, in warranty, and one for field service. Is that, how does that change what you offer? >> That's a very very good question. Because that's where the meta-data becomes so much more important because its manifestation is changing. I'll give you a great example, take Transamerica, Transamerica is a customer of ours leveraging big data at scale, and what they're doing is that, to your question, they have existing customers who have insurance through them. But they're looking for white space analysis, who could be potential opportunities? Two distinct ones, and within that, they're looking at relationships. I know you, John, you have Transamerica, could you be an influencer with me? Or within your family, extended family. I'm a friend, but what about a family member that you've declared out there on social media? So they are doing all that stuff in the context of a data lake. How are they doing it? So in that context, think about that complexity of the job, pumping data into a lake won't solve it for them, but that's a necessary first step. The second step is where all of that meta-data through ML and AI, starts giving them that relationship graph. To say, you know what, John in itself has this white space opportunity for you, but John is related to me in one way, him and me are connected on Facebook. John's related to you a little bit more differently, he has a stronger bond with you, and within his family, he has different strong bonds. So that's John's relationship graph. Leverage him, if he has been a good customer of yours. All of that stuff is now at the meta-data level, not just the monolithic meta-data, relationship graph. His relationship graph of what he has bought from you, so that you can just see that discovery becomes a very important element. Do you want to do that in different places? You want to do that in one place. I may be in a cloud environment, I may be on prem, so that's where when I say that meta-data becomes the organized principle, that's where it becomes real. >> Just a quick follow-up on that, then. It doesn't seem obvious that every end customer of yours, not the consumer but the buyer of the software, would have enough data to start building that graph. >> I don't think, to me, what happened was, the word big data, I thought got massively abused. A lot of Hadoop customers are not necessarily big data customers. I know a lot of banking customers, enterprise banking, whose data volumes will surprise you, but they're using Hadoop. What they want is intelligence. That's why I keep saying that the meta-data part, they are more interested in a deeper understanding of the data. A great example is, if John. I had a customer, who basically had a big bank. Rich net worth customer. In their will, the daughter was listed. When the daughter went to school, by the way, went to the bank branch in that city, she had no idea, she walked up, she basically wanted to open an account. Three more friends in the line. Manager comes out because at that point, the teller said, "This is somebody you should take special care of." Boom, she goes in a special cabin, the other friends are standing in a line. Think of the customer service perception, you just created a new millennia right? That's important. >> Well this brings up the interesting comment. The whole graph thing, we love, but this brings back the neural network trend. Which is a concept that's been around for a long long time, but now it's front and center. I remember talking to Diane Green who runs Google Cloud, she was saying that you couldn't hire neural network, they couldn't get jobs 15 years ago. Now you can't hire enough of them. So that brings up the ML conversation. So, I want to take that to a question and ask about the data lake, 'cause you guys have announced a new cloud data lake. >> Yes. >> So it sounds like, from what you're saying, is you're going beyond the data lake. So talk about what that is. Because data lake, people get, you throw stuff into a lake. And hopefully it doesn't become a swamp. How are you guys going beyond just the basic concept of a data lake with your new cloud data lake? >> Yeah, so, data lake. If you remember last year, actually at Strata San Jose we chatted, and we had announced the data lake because we realized customers, to your point John, as you said, were struggling on how to even build a data lake, and they were all over the place, and they were failing. And we announced the first data lake there, and then in Strata New York, basically we brought the meta-data ML part to the data lake. And then obviously right now we're taking it to the cloud, and what we see in the world of data lake is that customers ask for three things. First, they want the prebuilt integrated solution. Data can come in, but I want the intelligence of meta-data and I want data preparation baked in. I don't want to have three different tools that I will go around, so out of the box. But we also saw, as they become successful with our customers, they want to scale up, scale down. Cloud is just a great place to go. You can basically put a data lake out there, by the way in the context of data, a lot of new data sources are in the cloud, so it's easy for them to scale in and out in the cloud, experiment there and all that stuff. Also you know Amazon, we supported Amazon Kinesis, all of these new sources and technologies in the world of cloud are allowing experimentation in the data lake, so that allowed our customers to basically get ahead of the curve very quickly. So in some ways, cloud allowed customers to do things a lot faster, better, and cheaper. So that's what we basically put in the hands of our customers. Now that they are feeling comfortable, they can do a secured and governed data lake without feeling that it's still not self-served. They want to put it in the cloud and be a lot more faster and cheaper about it. >> John: And more analytics on it. >> More analytics. And now, because our ML, our AI, the meta-data part, connects cloud, ground, everything. So they have an organizing principle, whatever they put wherever, they can still get intelligence out of it. >> Amit, we got to break, but I want to get one final comment for you to kind of end the segment, and it's been fun watching you guys work over the past couple years. And I want to get your perspective because the product decisions always have kind of a time table to them, it's not like you made this up last night because it's trendy, but you guys have made some good product choices. It seems like the wind's at your back right now at Informatica. What, specifically, are bets that you guys made a couple years ago that are now bearing fruit? Can you just take a minute to end the segment, share some of those product bets. Because it's not always that obvious to make those product bets years earlier, seems to be a tail wind for you. You agree, and can you share some of those bets? >> I think you said it rightly, product bets are hard, right? Because you got to see three, four years ahead. The one big bet that we made is that we saw, as I said to you, the decoupling of the data layer. So we realized that, look, the app layer's getting fragmented. The cloud platforms are getting fragmented. Databases are getting fragmented. That that whole old monolithic architecture is getting fundamentally blown up, and the customers will be in a multi, multi, multi spread out hybrid world. Data is the organizing principle, so three years ago, we bet on the intelligent data platform. And we said that the intelligent data platform will be intelligent because of the meta-data driven layer, and at that point, AI was nowhere in sight. We put ML in that picture, and obviously, AI has moved, so the bet on the data platform. Second bet that, in that data platform, it'll all be AI, ML driven meta-data intelligence. And the third one is, we bet big on cloud. Big data we had already bet big on, by the way. >> John: You were already there. >> We knew the cloud. Big data will move to the cloud far more rapidly than the old technology moved to the cloud. So we saw that coming. We saw the (mumbles) wave coming. We worked so closely with AWS and the Azure team. With Google now, as well. So we saw three things, and that's what we bet. And you can see the rich offerings we have, the rich partnerships we have, and the rich customers that are live in those platforms. >> And the market's right on your doorstep. I mean, AI is hot, ML, you're seeing all this stuff converge with IoT. >> So those were, I think, forward-looking bets that paid out for us. (chuckles) And but there's so much more to do, and so much more upside for all of us right now. >> A lot more work to do. Amit, thank you for coming on, sharing your insight. Again, you guys got in good pole position in the market, and again it's right on your doorstep, so congratulations. This is the Cube, I'm John Furrier with George Gilbert. With more coverage in Silicon Valley for Big Data SV and Strata + Hadoop after this short break.