 Live from San Jose, California, in the heart of Silicon Valley, it's theCUBE. Covering Hadoop Summit 2016, brought to you by Hortonworks. Now, here are your hosts, John Furrier and George Gilbert. Welcome to theCUBE, great to see you. Pleasure to be here, John. Thank you for coming on and sharing the data with us on what's going on. But you're up on stage giving the keynote today and great keynote, very inspirational because it wasn't the speeds and feeds of how many bits are moving, IOT moving the data to the edge. You had some good concepts in there, but really pointing out the value of data, the impact that data has on society. You're talking about things that matter to people because data now is everywhere. Data is the new currency, there's valuation behind it. LinkedIn was sold for $26 billion to your company. So you saw some value in their data, obviously data is a premium. So share your thoughts because whether it's a LinkedIn for $26 billion or bringing data to a third world country, bringing internet access, collaboration, what is the value of data in your mind? Great question. You know, I have the saying, insights are worth pennies, decisions and actions are worth dollars, data is useless. And let me illustrate that with a small example. In India, and I spoke about this in the keynote, we have the state of anthropology predicting when students will drop out of school using machine learning. And when they can predict a student will drop out, they can actually take action to prevent that dropout. And it's that action that makes a difference in the lives of people. Now there are 440 million children in India and only 50% attend school regularly. And when machine learning can retain a child in a school and let the child realize his full potential as a human being, we have made a difference. And it's really that part that we need to post, the gap between data and taking intelligent action. And that's now being empowered with machine learning and analytics. And machine learning and analytics now getting deeply embedded in databases, in real time databases, so that applications can be built that are very powerful, empowered with data and predictive intelligence. And that's- And you know, it's amazing. We were talking earlier in an earlier segment with Mervage, we were just kind of kicking around and riffing on the idea that, this industry is a lot like children in pre-teens. And you mentioned the children saw go there. You know, it's kind of awkward. They're dating for the first time. A lot of play dates, a lot of car pools, whatever metaphor. But there's a lot of adults kind of conversation so that we've seen the maturization of the industry. But AI and machine learning bring out a learning aspect to machines I want to get your thoughts on. Because you mentioned as children go to school, they're learning and as a parent of four, you want your kids to be ready for the real world to be prepared to process the unknown. And that's the world we live in. So machine learning and AI, this is the vanguard of the learning machine. So talk about that aspect to the industry because it's not just about AI and learning machines. It's about how do you prepare for these unknown scenarios? Like a person going out in the real world. Right. And so in the past, we used to believe mathematical equations could describe most of the world. The physics, in physics, mathematics was unreasonably effective. Economists had this physics envy, so they tried to model everything, but didn't succeed so well. But now what is happening is when mathematics is not that effective, when you have tremendous amount of data, you can be unreasonably effective. And it's not just about data. It's about algorithms that operate on the data and can really extract useful information that can actually drive an action. And that is the transformation we are all undergoing. So then a number of things are coming together to make that happen. So why has AI so dramatically changed in the last few years compared to 30, 40, 50 years ago when it was started? It's all the power of data. There is data on every kind of behavior of every creature on the planet, every human being, and on so many scenarios. So you can now learn from the data and then you can become truly intelligent. And you can model the phenomenon so you can be predictive. You have, well, you also have unstructured data. You cannot lock in a database. So no schema kind of restrictions. And you have compute. Yeah, enormous amount of compute. So you have the perfect storm to try to do those ontologies that you couldn't figure out in the past or whatever technology. And when you have enormous amount of data, you can integrate all of that data in the cloud with enormous amount of compute. And that allows you to create. So, you know, Microsoft has got a lot of action going on. Certainly we talked about the LinkedIn thing as earlier on. Clearly the enterprise is so strong for Microsoft. They're the adults in the conversation. Certainly not a teen at all. But the thing about Microsoft is that they have a large enterprise. But as the enterprise becomes more consumerization oriented, the experience of the users are changing. So can you share your thoughts on the kinds of innovations that you guys are looking at with data? I mean, is there algorithms for algorithms? Is there compilers for compilers? I mean, you have a whole new paradigm shift in computer science, in vectoring into an environment that has all this legacy stuff. So you guys have to balance well. Share your thoughts on Microsoft's innovation strategy on data. Great question. I ask audiences very often, how many people wear a tailored shirt? And very few hands go up. But see, when I was a kid in India like 40 years ago, we went to textile stores, got clothed by the yard, went to a tailor, got clothes tailor made. Except what changed in 40 years? You know what changed in 40 years is really mass manufacturing of clothing became so automated. So it was possible to now create ready-made clothes of all sizes, shapes, even designer clothing. Distributed through department stores. And then you could from that vast selection find what you wanted. The same thing is happening with data and analytics today. Now the traditional way of doing analytics is like tailored in 40 years. So first there's a general purpose, kind of like all-purpose analytics, moving to much more specialized prepackaged apps. So it becomes like apps. And like apps in an app store. George, you got a question? Oh, well I just wanted to ask on that, apps that for 50 years we had, we would design the data, the buckets, the data model. We knew exactly how it fit. And if it didn't fit, it was wrong. And we put the process, the business process on top. Now it seems like we've potentially flipped that on its head where the application emerges from the data and the data is always changing. Is the applications can learn from it. But even before that, we are packaging up intelligence into cloud-hosted APIs. For example, face detection APIs, speech recognition APIs, translation APIs, OCR APIs. The list is actually very long. So on the cloud, on Microsoft Azure today, you will see cognitive APIs for a large variety of tasks and that they're ever increasing. And so when you have these finished APIs, application development becomes very simple. You just glue those applications together, those APIs together, and you get a very powerful application all written in the cloud, supported with SLAs of the cloud and backed by a company like Microsoft. And that creates unreasonable speed in developing intelligent applications. That is clearly the way, no more perimeter in the enterprise. Docker containers are on the horizon. So you know where I'm going with this, security. Okay, I love APIs, we're an API economy. A lot of Swiss cheese, a lot of holes potentially. So what is the strategy that you guys see for customers? Because there are some techniques and if you could take a minute to share some of the best practices in security because it's okay to have no perimeter. Otherwise you can't unlock the openness of the API. So what's the best practice of this? So this is again one of the areas where I think Microsoft differentiates itself with its Azure Active Directory for role-based access control. So the Active Directory is the way most enterprises run their Office 365 deployments and user authentication and access control. It seamlessly integrate with Microsoft Azure. So you have identity and verification. And then we have encryption support. We have all types of other things that we have laid out. Like for example, SQL databases in the cloud has something called threat detection. It's intelligently scanning all types of the SQL that is running against it and alerting on things that the customer should know about. So many of these things get built into the cloud and provides a comprehensive layer of security that a typical enterprise running on their own in their own data center will find it incredibly hard to manage and match in its competency. And that's the difference the cloud vendors are providing. So you've got this, what's turning into a database that not only is really good at the analytics and the transactions and the speed and the volume, but you've also got this whole in the Hadoop ecosystem where we've got sort of data at rest at volume. We're beginning to get data in motion at speed with Spark streaming and Flank and others. So we don't have anything in the center to pull it all together to say, based on this latest information, I need to make a decision and operationalize that. And that's a great question because that's one of the areas we focused on in a big way with SQL Server 2016, operational analytics. Let me explain. So SQL Server 2016 has in-memory OLTP transactions. You can support up to 12 data bytes of main memory in one server, by the way. And at the same time, you have real-time, updateable column stores for real-time analytics. And then we integrated R, which is open source, with capabilities of machine learning, deep into the database so that you can have intelligent algorithms running next to aggregations created in real-time off of real-time transactions that are coming in. And so you can have real-time scoring of models happening in those systems. And then this all comes packaged in a database which has great security, which has great capabilities for high availability and replication and all the things that database engineers build over the years. And so here's the transformation that happened. It's not just a database. It's an intelligence base, meaning intelligent models built out of R are managed in a database system. It's a management system that now extends to predictive intelligence as well. So when you bring all of that together, that's when you have the ability to build extremely powerful applications, operational analytics applications that can take in real-time data, can do real-time analytics, and drive real-time decisions, such as, for example, in quantitative trading in financial markets where customers are using this type of capability from us. So, Joseph, I want to talk to you about analytics, kind of where it's come from, where's it going. You can look at analytics to see what happened in the past, and you can use predictive analytics to look at what's in the future. But the moment right now that's happening, real-time, is really the hottest area right now. So what's your vision, and what are you guys doing to capture analytics in the moment? Is it streaming? Is it the machine learning? And what's different about Microsoft? What are you guys doing to solve that problem? Because together, that's the holy trinity of analytics. Past, present, future. Absolutely. So, in the cloud, for example, in the Cortana Intelligent Suite, we have this great service called Azure Stream Analytics. What that allows you to do is to put a standing query in the flow of data. And because it's a SQL query, it's actually very easy to describe. So you can do complex event modeling. You don't have to write Java code and compile it. It's just a SQL query that's standing in the flow. And it does aggregates in real-time as the data flows through. And those aggregates can go to dashboards, like Power BI dashboards. Aggregates can go to call machine learning APIs and get results back. So you can do things like finding anomalies in a data stream or machines about to fail or do predictive analytics from the streaming data. And that is an incredibly powerful capability. And then, second, in the cloud again, on our HDInside Hadoop service, we have laid on Spark. We are a big fan of Spark as well. And Spark supports streaming analytics. So you can now combine there on the HDInside service, Spark streaming with batch analytics, with powerful machine learning capabilities from our server. All of that deployed in the cloud as a service. My final question, I know we've got to go on time, is that what is the big partnering strategy you guys have? Obviously, partnering is in your DNA. How do you view the landscape right now? Obviously, HPE, we saw some partnerships there. You talk about some of your partnerships. Absolutely. You know, Microsoft has, it's DNA has been a platform company. And platform companies thrive on ecosystems and partnerships. So we are, for example, driving great partnerships with Hortonworks, with all of the Hadoop vendors, with Databricks, with Red Hat, with any number of partners in the open source world, so that we can actually create value at that platform layer, incredible value at the platform layer for application developers. That's what we're trying. Just shows us, Suresh, thank you for coming on theCUBE. Really appreciate you. Thank you for sharing the data and the insights here on theCUBE. Hopefully, that'll render well in the analytics that gets powered with all the stats. Thank you for coming on theCUBE sharing. I'm John Furrier, George Gilbert. You're watching theCUBE. We are live from Silicon Valley at Hadoop Summit 2016. We wrap by Eric Bermura for this short break. You're watching theCUBE. Thank you.