 Live from Las Vegas, it's theCUBE. Covering IBM Think 2018, brought to you by IBM. Welcome back to theCUBE. We are live on day one of the inaugural IBM Think 2018 event. I'm Lisa Martin with Dave Vellante, and we are in sunny Vegas at the Mandalay Bay, excited to welcome to theCUBE one of the IBM Fellows that make her welcome to theCUBE. Thank you so much. So you are not only an IBM Fellow, but you're also an IBM Analytics technical leadership team chair. Tell us about your role on that technical leadership team. What are some of the things that you're helping to drive? And maybe even give us some of the customer feedback that you're helping to infiltrate into IBM's technical direction. Okay, so basically technical leadership team is the group of top technical leaders in the IBM Analytics Group. And, you know, we are kind of chartered by evaluating the new technologies, providing the guidance to our business leaders into what to invest, what to de-invest, listening to our customer requirements, listening to how the customer is actually using the technology and making sure that IBM is timely there when it's needed. And also a very important element of the technical leadership team is also to promote the innovation, innovative activities, particularly kind of grassroots activities, meaning helping our technical leaders across the analytics to encourage them to come up with innovation, to present the ideas so that to follow up on those, to potentially turn them into project and so on. So that's it. And guide them or just sort of send them off to discover. As a matter of fact, we should be probably mostly sounding bored. So not necessarily, you know, that this is coming from top down, but trying to encourage them, trying to incite them, trying to kind of, you know, make the innovative activity interesting and also at the same time make sure that they see that there's something coming out of it. It's not just, you know, they are coming up and then nothing's happening, but you know, trying also to turn that into the reality by working with our business developers, which by the way, who, by the way, that control the resources, right? So in order to do something like that. How about, how much of it is guiding folks who want to go down a certain path that maybe you know has been attempted before in that particular way and say, you know what, probably better to go elsewhere or do you let them go and make the same mistake? And is there any of that? Like don't go down that, don't go through that door. Well, as you can imagine, it's a human to attempt to say, well you know, I've already tried, already done, but you know, we are really trying not to do that. We are trying not to do that, trying to have open mind, because in this industry in which we are, there's always new set of opportunities and new conditions. And even if you are going to talk about, like our current topic about like a fast data and so on, I believe that many of these things have been around already. We just didn't know how to actually, how to help, how to support something like that. But now with the new set of technologies, we can actually do that. So let's get into the fast data. I mean, you know, it wasn't too long ago, we just asked earlier guests, what inning are we at in IoT? Said the third inning. It wasn't long ago, we were in the third inning of a dupe and everything was batched and then all of a sudden big data changed, everything became streaming, real time, fast data. What do you mean by fast data? You know, what is it? What's the state of fast data inside IBM? Well, thank you for that question because I also wanted, when I was preparing a bit of this interview, of course I wanted first to, that we are all on the same page in terms of what fast data actually means, right? And like, and there's of course in our industry, it's a full of hype and misunderstanding and everything else. And like many other things and concepts, actually it's a not fundamentally new thing. It's just the fact that the current state of technology and the enhancements in the technology allow us to do something that we couldn't do before. So the requirements for the fast data value proposition were always there. But right now technology allows us actually to derive the real time insight out of the data, irrespective of that data volume, variety, velocity. And when I just said that three Vs, it sounds like a big data, right? And as a matter of fact, there is a pretty large intersection with big data, but there's a huge difference. And a huge difference that's typically big data is really associated with data at rest, while the fast data is really associated with data in motion. So the examples of that particular patterns are all over the place. I mean, you can think of like a click stream stuff. You can think about ticker financial data, right? You can think about manufacturing IoT data, sensors, logs, and the spectrum of industries that take advantage of that are all over the place. From financial and retail, from manufacturing, from utility, all the way to advertising, to agriculture and everything else. So I like, for example, very often when I talk about fast data, people first drop immediately into, let's say, you know, these are YouTube streaming or this is, you know, Facebook, Twitter, kind of postings and everything. But while this is true, and certainly there are business cases built on something like that, what interests me more are the use cases like, for example, Airbus, right? With 10,000 sensors in each of the wings, producing like a seven terabytes of information per day, which by the way, cannot be just a dump somewhere like before and then do some batch processing on it, but you actually have to process that data right there when it happens, that millisecond, because, you know, the ramifications are pretty serious about that, right? Or take, for example, opportunity in the utility industry, like in a power electricity, where the distributors, manufacturers, really entice people to put the smart metering in place so that they can basically measure the consumption of the power, electricity power, basically on an hourly basis. And instead of giving you once a yearly kind of bill of what it is, to know that all the time, what is the consumption to react on spikes, to avoid blackouts, and come up with a totally new set of business models in terms of, you know, offering some special incentive for spending or not spending, adding additional manufacturers, I mean, fantastic, you kind of set of use cases. I mean, I believe that Gartner said that by 2020, like 80% of the businesses will have some sort of kind of situational awareness application, which is not a word of basically, you know, using this kind of capability of event-driven messaging. And I agree with that 100%. So it's data, fast data, it's data that is analyzed in real time. Right. Such that you can affect an outcome. Right. Before, before what, before something bad happens, before you lose the buyer, before all over the place, you know, before fraud happens in financials, right? Before manufacturing lines breaks, right? Before, you know, airplane, or something happens with the airplane. And so there are many, many, many examples of something like that, right? And when we talk about that, I mean, what we need to understand again, even the technologies that are needed in order to deliver fast data, value propositions, are kind of known technologies. I mean, what you really need, you need a very scalable tab sub messaging system like Kafka, for example, right? In order to acquire the data. Then you need a system which is typically streaming system, streams, and you have tons of the offerings in the open source space, like, you know, Apache Spark streaming, you have a storm, you have a fling, Apache fling product, as well as you have our IBM stream capabilities for really the kind of enterprise called your service delivery. And then very importantly, and this is something that I hope people have time to talk today, is you also need to be able to basically absorb that data, and not only do the analytics on the fly, but also to store that data and combine that with the analytics, with the data that is historical. And typically for that, what if you read what people are kind of suggesting what to do, you have also lots of open source technologies that can do that like a samdra, like some HDFS based systems and so on. But what I'm saying is all of them come with this kind of complexity that yes, you kind of lend data somewhere, but then you need to put it somewhere else in order to do the analytics. And basically you are introducing the latency between data production and data consumption. And this is why I believe that the technology like DB2 events store that we announced yesterday is actually something that will come very, very interestingly, a very powerful part of the whole fast data story. So let's talk about that a little bit more. As fast data as a term, and thank you for clarifying what it means again, isn't new, but to your point, as technology is evolving, it's opening up new opportunities, much like it sounds like kind of the innovation lab that you have with an IBM. There might be, Dave was asking, ideas that people bring that aren't new, maybe they were tried before, but maybe now there's new enabling technologies. Tell us about how is IBM enabling organizations of whether their fast-paced, innovative startups to enterprise organizations, not create that sort of latency and actually achieve the business benefits that fast data can help them achieve today with today's or rather technologies that you're announcing at the show. Right, so again, let's go through these stages that I said that every fast data technology and project and solution should really probably have. As I said, first of all, you need to have some messaging from systems, and I believe that the systems like Kafka, absolutely enough for something like that. Then you need a system that's going to take this data of that fire hose coming from the Kafka, which is stream, stream technology. And as I said, lots of technologies in open source, but IBM Stream as a technology is something that has also hundreds of different, basically models, whether predictive analytics, whether it's prescriptive analytics, whether machine learning, basically kind of AI elements, text to speech that you can apply on the data on the wire with the wire speed. So you need that kind of enterprise quality of service in terms of applying the analytics on the data that is streaming. And then we come to the DB2 event store, basically repository for that fire hose data, where you need to put this data in the format in which you can basically immediately, without any latency between data creation, data consumption, do the analytics on it. That's what we did with our DB2 event store. So not only that we can ingest like a millions of events per second, literally millions and millions events per second, but we can also store that in a basically open format, which is tremendous value. Remember, any database system, basically in the past stores data in its own format. So you have to use that system that created data in order to consume that data. I agree with you. What DB2 event store does is actually it ingests that data, but it puts it into the format that you can use any kind of open source product, like for example, Spark analytics to do the analytics on the data. You can use Spark machine learning libraries to do immediately kind of machine learning modeling as well as scoring on that data. So I believe that that particular element of event store coupled with a tremendous capability to acquire data is what makes a really differentiation. It does that how through a set of APIs that allows it to be read and consumed. So basically when the data is coming off the hose, you know, of the streams or something like that, what event store actually does, it puts the data, it's basically in memory database, right? It puts the data in memory. Something else that's been around forever. Exactly, something else, you know. You just need, we just have more of it, right? And guess what, you know, if it is in memory, it's going to be faster than if it is on disk. What a surprise. So of course, you know, when the, puts the data into the memory and immediately makes it basically available for querying if you need this data that just came in. But then kind of asynchronously offloads the data into basically Apache Sparky format into the columnar store, basically allowing very powerful analytical capabilities immediately on that data. And again, if you like, you can go through event store to query the data, but you don't have to. You can basically use any kind of the tool like Spark, like Python or Anaconda stack to go after the data and do the analytics on it, to build the models on it, and so on. And that asynchronous transformation is fast? Asynchronous transformation is such that it gives you this data, which we now call historical data, in a little bit, basically in a minute. Okay. So it's kind of like a minute. So very reasonably low latency. But what's very important to understand that actually the union of that data and the data that is in the memory on this one, which we by the way make transparent, can give you 100% what we call kind of almost transactional consistency of your queries against the data that is kind of coming in. So it's really not a hybrid kind of store of the memory, in memory, very fast log, because we're also logging this data in order to have it for high availability across multiple things, because this is highly scalable. I mean, it's highly what we call web scale kind of application database. And then parquet format for the open source storing of the data for historic analysis. Let's, in our last 30 seconds or so, give us an example. I know this was just announced, but maybe a customer genericize in terms of the business benefits that one of the beta customers is achieving, leveraging this technology. So in order for customers really to take advantage of all that, as I said, what I would suggest to customers to do, first of all, to understand where these situational awareness applications actually make sense to them. Where the data is coming in firehoses, not that, you know, traditional transactional capabilities, but through the firehose, where does it come? And then apply these technologies, as I just said, you know, a position of the data, streaming on the wire analytics, and then even DB2 events store as a store of the data. For all that's what you also need, just to tell you, you also need kind of messaging runtime, which typically produce like, for example, ACA technology. And that's why we have also, we have entered also in partnership with the Lightman in order to deliver the entire kind of experience for customers that want to build applications that run on a fast data. So maybe enabling customers to become more proactive, maybe predictive eventually? To enable customers to take advantage of this tremendously business-relevant data, that is data is coming in the, is it a clickstream? Is it financial data? Is it IoT data? And to combine it with the assets that they already have coming from transactions. So that's a powerful combination, that basically they can build totally brand new business models, as well as enhance existing ones to something that is going to, you know, improve productivity, for example, or improve the customer satisfaction, or broaden the customer segments and so on and so on. Thank you so much for coming on theCUBE and sharing the insight of the announcements and it's a pretty cool day between you and an IBM fellow. Yeah, that's- That's pretty good for a Monday. It's Monday, isn't it? Thank you so much. Not easy becoming an IBM fellow. So congratulations on that. Thank you so much. And thanks again. Thank you for having me. Absolutely our pleasure for Dave Vellante. I'm Lisa Martin. We are live at Mandalay Bay in Las Vegas. Nice sunny day today, where we are on our first day of three days of coverage at IBM Think 2018. Check out our CUBE conversations on thecube.net. Head over to siliconangle.com to find our articles on everything we've done so far at this event and other events and what we'll be doing for the next few days. But stick around, Dave and I are going to be right back with our next guest after a short break.