 Live from the MGM Grand Convention Center in Las Vegas, Nevada, it's theCube at Splunk.com 2014. Brought to you by headline sponsor, Splunk. Here are your hosts, John Furrier and Jeff Kelly. Okay, welcome back everyone. We are live in Las Vegas, Nevada here for Splunk 2014.conference. The hashtag is Splunk.com, C-O-N-F. This is theCube. We go out to the events, our next guest is Greg Rebeck, who's the Director of Engineering at New York Airbreak. Welcome to theCube. Thank you. So Jeff Kelly and I have been discussing data. We love data. But it's all been about security of the keynotes. We got the DevOps angle covered. But one thing we haven't gotten to yet is the internet of things. Because data is everywhere on the keynotes. I love the clients on the floor. It's on the walls everywhere. Data is everywhere and the theme that came out of this afternoon's interviews was analytics for everybody. So I want to get your take on the reality of internet of things. I mean, where are we? I mean, and certainly it's hyped up as often can see, but where's the meat on the bone for internet of things? So the meat on the bone is bringing all the data back to a central location and then presenting it, like Splunk said, analytics for everybody. So to our developers, they dig into the details. They set alerts so they can be proactive rather than reactive. And then we provide dashboards to executives, not only to our internal stakeholders, but to our external stakeholders at different levels. We provide dashboards to executives at the railroads, to their operating practices, all the way down to their mechanical departments for preventative maintenance. So these dashboards, real time? Yes, pretty much. Okay, so talk about what before Splunk. It's kind of like BS, before Splunk and after Splunk, right? It's like, you know, so before Splunk, right? What was the, what it looked like? Was there like reporting, was it old data warehouse that didn't get, was it any in place? So we lived in spreadsheet health. So I had high value resources, senior level system engineers, meticulously combing through spreadsheets. If I wanted to report, it would take three days to get. One of the things that our software does relies on is compliance. If an engineer or the locomotive is not using it, they're not going to get results. So we, in order to help the customer build a business case, we show different levels of compliance and different return on investments. So in order to say, what's the return at 65% and what's the return at 80% and building that bridge would take weeks. Now I can do it instantaneously. And what's the impact on the business? So I mean, obviously, I mean, that's definitely sounds like a BS environment. Pun intended, I'm not going to get it. I'm going to copyright that Splunk. You can't use that on plug-ins royalties. So after Splunk, talk about the consequences. I mean, what was the impact? I mean, because you can start to quantify that. I'm sure the resources just add up their value. You've got any material business impact that you guys have seen? So we're not only able to build business cases for our customers very, very quickly. We went from having a nine to 12 month deployment cycle down to 90 days. Wow. So like you get kudos for management, like you get a raise, come on, what happened? No, so... Was there like people like celebrating? I mean, no, I know people. Absolutely. And you know, we went from turning down business now to actually proactively going out and getting business. So you just top line, it's revenue coming in. Right, that's cool. So let's take a step back. So you manufacture, build, design, sell air brakes for trains, for locomotives, correct? So your customers are shipping companies or... The Class One Railroad, so Union Pacific, Norfolk Southern, CSX. And then internationally, we do a valet. Over in Australia, we do Rio Tinto, BHP. So they have options where they can source their air brakes from different vendors. So you're obviously, those are your customers. How is Splunk helping you attract and retain business? What are you able to provide to your customers through using tools like Splunk, using data essentially, to add value to your customers? So even though we have one customer, a railroad, there's really maybe five or six internal customers. You have operations, you have mechanics, you have IT, you have executive leadership. They're all siloed. And they all have their own information. We happen to be in a unique position where we interact with all of them. We need to interact with mechanical to get our product on the locomotive. We interact with IT, so that we get the dispatching information to feed all the metadata to our software. So we become an aggregate across all those silos. And then we can provide value back to each of those silos individually and to the railroad executive team about their operations, bringing it all together. So you can provide that data as a service, really, that you provide? Or the glue to provide decision-based? So I bring this out because I love the contracts. I mean, you're building a very industrial product, air brakes, but at the same time, you're a very cutting-edge data-led organization, essentially using data as a line of business for you. Yep. In fact, one of the things that one of my engineers just discovered was since we modeled the braking efficiency of each car on the locomotive, there's actually a formula that correlates to the thickness of the brake pad based on how much that efficiency is. So we called up our division in Kansas City that makes those brake pads. They figured out the formula, and now we can start looking at predicting brake pad deterioration in real time. So that allows you to reach out to your customers and do proactive maintenance, so that- Yeah, we can do proactive maintenance. We can predict inventory levels and ship brake pads to the right locations rather than to a central location and distribute. And that just saves, I would imagine, millions of dollars when you can predict and take a proactive approach versus having a train or have to come offline because of a problem. Absolutely. The existing model is the trains are just scheduled every 90 days, whether it needs it or not, it comes in for an overhaul, and then it goes on its way. This allows that asset to stay on the railroad as long as possible, making money, and then only come in when needed. Now, talk about the efficiencies in terms of just fuel efficiency. We talked a little bit beforehand. I'd love you to share that kind of anecdote. I mean, you mentioned your use of diesel fuel is, I think, bigger than, I think you said, all the Navy. I mean, just talk about your use of fuel and how you can actually find efficiencies there. Yep, so what our product does is it takes in all the physics of the locomotive, does some optimization, and comes up with different driving strategies. And one of those driving strategies is optimized for fuel efficiency. And that efficiency usually ranges anywhere from 5% to 8%. And a single class one railroad will use more diesel fuel in a year than the entire U.S. Navy. So even a 1% change in reduction results in tens to hundreds of millions of dollars. And again, you can, I'm guessing, and that is an area where your customers are looking to you to really add that value. They, as I said, they probably look at, committed bid, where are we going to source our air brakes from? If one vendor can say, look, we're going to help you be more efficient and save on fuel chargers, that is just easier ROI calculation for the customer. Absolutely. Fuel is our number one business case that we go with. I mean, we do other business cases where we reduce the wear and tear on the locomotives, but fuel is absolutely number one. So I'm not sure how long you've been with New York air brake, but can you walk us through kind of the evolution of the company in terms of getting to this more data-driven approach to serving your customers? I'm sure at some point it was not quite so focused on using data in this way. Was it a big shift for your company, for your industry? Absolutely. The railroad is a very old industry and New York air brake itself is now celebrating its 125th year in the industry. So we tend to move a little bit slower on the technology side of things than the rest of the world. So it was around 2010 when the leader products started getting traction in the industry and we started realizing we had all this data. We know all the trains that run across the US. We know what they're carrying, how much they weigh, where they're going, everything. And there was a lot of interesting things we could do with that. And so from there, we took that information and we're starting to build now vertical products on top of our core units. And what did it look like inside the organization? Did it require buying from the very top of the organization? What really drove that transformation? Absolutely. Everybody kind of felt like there was something there, but they didn't know what it was. So we went out and we started asking, and once our customer said, what are your top 10 problems that you face? Then we drilled down and said, okay, how are you solving them today? What questions could you have answered that would help you get there? And how do you measure, existingly with the metrics you have, that these solutions that you have in place are working or not? We took that information back and then started building metrics and analytics that could help answer those questions that they didn't have answers to and then help measure to make sure that the solutions that are provided or that we do provide in the future are actually bringing value. So talk a little bit about Splunk a little more specifically. So one of the things we heard in the keynote this morning got resolved when the CEO of Splunk kind of, he put up a slide where he kind of compared the old world of kind of the enterprise data warehouse and BI, which is more rigid and you had to know the questions ahead of time versus this new approach that Splunk and some of the other more modern tools to do for other things. Enable works, unstructured data, you don't have another question ahead of time, you don't have a schema on read, that kind of thing. How, walk us through the reality in your organization. I'm guessing you've been around for 120 plus years so I'm sure starting 20, 30 years ago you started working on data warehousing, BI, that kind of thing. What's the environment like now between some of those older approaches that God pre kind of highlighted versus some of the new approaches and things that Splunk are nailing? So there's very much a relationship in your data and the old way was you had to establish that relationship basically get engaged, get married to it, everything was well defined and the longer you were together the harder was the divorce, the more ugly it got, right? Now it's kind of like speed dating. I can try something, I can bring new data in, I can basically try different things out without a commitment and that's definitely accelerated our pace and the ability to service our customers. Before it was do all this research, get a commitment, build long, big business cases up to executive leadership, get approval and by that time you may or may not have been the market leader. Very slow, very multi-year cycles. Now we're looking at quarterly cycles. We want to test something, we can just bring in the data, test something out, put a proof of concept together, serve it to our customer and then get feedback and see whether it's good. So basically it's like going into the tunnel to use the locomotive example. You don't really know, you can do all that investment, all that energy, time, I mean it's muscle work too. It's a heavy lift, right? And you don't even know what the hell is going to come out of it, right? So you've got all that prep work. So what you're saying is you're now in the mode of you can do some heavy lifting, like a light workout if you will, get some real taste of where it's going and then iterate quickly. And that's like a night and day situation. I mean that must boost the morale up a little bit. Absolutely, all the engineers are very excited. They're not afraid to ask questions and then dig in where before they're like, oh do I really want to invest all my time and effort into it because it became very overhead intensive. Now it's, let's be creative. So Jeff's doing a bunch of research right now around the role of developers in big data. So I want to get your take on this because this hits the point that we've been talking about, what we love is when you get developers exposed to the data and they can taste it, there's a little heroin action going on, a little bit of addiction. The creativity comes out which is kind of cool because they're close to the front lines now, they're certainly close to the front lines. But what does that do for the organization and does that work? And are developers now thinking like data's part of the development process? Meaning what's it like now in this environment for you guys as a great case study because are the developers like, give me the data. Are they like data, data hordes, are they hungry for the data? I mean what's going on with the role of the developer particularly? So it brings a new energy to the organization. So development has always been part science and then part art. And we've always kind of shoved the art part back in the corner and just said, hey, here's your tickets, go crank these out and stay back there in the closet. Now they get to express that creativeness and they interact with each other much more and that boosts them around, boosts productivity, it's seven o'clock and people are still hanging around the office rather than 430 and staring at their watch. So would you consider data to be a key part of that? I mean the data pipeline? Absolutely. It's an enabler to give the developers that creativity because it gives them the facts to come back to the leadership team and say, hey look, look what I discovered, this is what I can do and this is the value it brings. So before I head back to Jeff, Jeff wants to jump in. We fight for the microphone when we get it. We get some good topics going here. So I got to ask you, I was getting to Jeff, what's the coolest thing that you've done with the data in the after Splunk scenario, right? So before Splunk you had the manual spreadsheet held, now you're in an awesome environment, Liberation, happy, everyone's having a drinking beer, partying, what's the coolest thing you guys have done? I think the coolest thing is correlating the information coming from the train from our models and then correlating to real physical assets like those brake pads. Now being able to look at and say exactly, this brake pad's going to be three quarters of an inch thick because it experienced this much braking effort. I mean, that's really cool to correlate. Something that you're modeled to see in the physical result and being very accurate. And then you guys should put resources behind that big time. Yep, and we can look at, there's other problems in the railroads. One of the problems is hot bearing detection. And that's been a problem for 20 plus years. There's nobody been able to solve because they've all been looking at those bearings. We can now take a step back, look at other things that maybe influence those bearings, build a correlation and then solve that problem. I'm going to see the headline now. Big data stops the runaway train, literally with the brakes. That's awesome, you guys are awesome. Jeff, go ahead. That's why you're the founding and editor of Silicon Angle and a great media. You can come up with headlines like that on the fly. It's New York post legs. Basically, it's very good. Can't go wrong, best headlines in the business. Daily news in the post. That's true, most entertaining for sure. Just want to take a step back. John mentioned the data pipeline and it just kind of jogged something in my memory from the keynotes this morning. A gentleman from Coca-Cola was up there talking about, in his case, using the cloud, but essentially automating a data pipeline. So that data is being fed into, in his case, what do you call it, a data lake? So whatever you want to call it, how do you go about making sure that all the data sources you need are brought together in a place where you can access them and do the kind of analytics that you've been talking about? And how do you make that, do you automate that process? How do you go about doing it? So we do automate that process. So all the locomotives have a wireless connection to a back office, to a communication server. And we need that not only to get the event logs, but also to send things like speed restrictions down to locomotives, so if there's track work going on or what the train to make up is, et cetera. But those logs get awkward to the communication server and then they get sent back to our office and then they get processed. So one of the things that Splunk is helping us do now is before that analysis used to take two to three days because it would have to basically redo the entire run. And that was called playback. Create a DVR style, what the engineer saw on his screen, what all the forces were, what was all the characteristics. Now with Splunk, we can kind of do that near real time and we don't have to wait days, we wait hours. So I was gonna say, so what is the value of that? And then you just explained it, it's the time to insight, really. Yep, and it's very important whenever there's an incident on the railroad such as a break-in tube, that is very important to the railroad that they get that information back of what half of what occurred. Was it the driver's fault? Did they have a faulty coupler or a knuckle? Was it a clement failure? They need that analysis right away so that they can address the situation. So last question for me, and Wikibon, as John mentioned, we're doing a lot of research around talking to big data practitioners and try to identify what are the key characteristics of successful big data practitioners. From your opinion, from your perspective, what are some of those key characteristics at your company that has enabled you to really leverage data? And that could be technology, it could be culture, it could be leadership, what are some of those key characteristics at your company? I think the key characteristic is not going in saying, hey, we have big data, let's go tell us what the data can find for us in building false correlations. If you remember back in the 90s, you had the magic eyes, you had to stare at the newspaper and move it out, cross-eyed. Your data kind of looks like that, all jumbled together and you're trying to make a picture out of it. With Splunk, you put on glasses and you see the picture. By bringing all that together and building relationships between the data and what your business needs and using a question-based, problem-based approach to the data, then you'll find success. And that has been very successful for us in understanding what our customers need because they have all the data in a day. We're not getting any more information that they don't already have, but we're providing a tremendous amount of value that they don't have. So if you guys quantified any of the savings, like a 1% savings, I interviewed with the CEO of GE, we interviewed all their top customer at the chairman of United Airlines, I mean a 1% savings to United Airlines because the data that they've got is over a billion dollars. And that's United Airlines, but you guys aren't United Airlines, but you got a significant impact. Any quantification? So on average, we save our customers about a billion dollars a year between fuel, reduction, a billion dollars. One billion. One billion dollars. You're in the big. We're in the big B. All right, awesome. And how much data do you guys have coming through there? It's not massive data. No, so we only have a 10 gig license. So no matter what she tells you, it doesn't matter how big your data is, it's how you use it. Yeah, exactly. Don't come knocking if the locomotive's rocking. So, but yeah, I mean, we hit the big B and that's fuel savings is a big chunk of that. And entering force management, which is basically preventing those break-in tubes. So I got to ask you, are you a Yankees fan? No, I grew up in Cleveland, Ohio, some of the big Indians, Cleveland Browns fans. Just want to make sure, because we're Red Sox fans in Massachusetts. So, you know, we're going to talk about it, because if Dave Vellante was here, we'd be talking about the Yankees because the G year thing, but that's great. I mean, you guys saving a lot of money for your customers and obviously safety's a huge issue. Absolutely. Any reports there? So positive train control, which is basically a act by Congress that said, hey, the technology exists. If a guy's busy texting in a locomotive, blows a red light, the technology exists to stop the locomotive. That's going to roll out next year and that's going to bring a whole new data stream into the organization. We'll know signal states, we'll know. Can't even imagine yet, we'll know. And that's going to help us build some tougher safety products. Great, great stuff. Creativity, I mean, I love the developer's success, love the customer value proposition. I mean, it's really a great case study. Again, we're not making this stuff up, this is real. Big data kind of gets some real tangible numbers. That's to me, Jeff, what we talked about with GE and other companies doing the Internet of Things is fantastic. This Internet of Things is real and it's billions of dollars of value, quantified. Not like pie in the sky. Greg, thanks so much. All right, this is theCUBE live in Las Vegas. We'll be right back. I'm John Furrier here with Jeff Kelly with Wikibon Live in Las Vegas. We'll be right back after this short break.