 Live from the MGM Grand Convention Center in Las Vegas, Nevada, it's The Cube at splunk.conf 2014. Brought to you by headline sponsor, Splunk. Here are your hosts, John Furrier and Jeff Kelly. Hello everyone, welcome back. We are live in Las Vegas for The Splunk Conference. This is The Cube, our flagship program. We go out to the events and expect the signal to noise. I'm John Furrier. With Jeff Kelly, the big data analyst at Wikibon, number one in the industry. Always putting out some great reports. Our next guest is Dave Vivesh, global market counterparty risk support manager at BNP. Welcome to The Cube. Thank you very much, good to be here. One of the things that we were tweeting earlier in our crowd chat was the AWS announcement that teases out the whole cloud play, which we get excited about because we love DevOps. We have our own cloud software and we know all the amazing goodnesses that comes from that. So at the same time it's a cultural shift. So when we were talking to the product guys here at Splunk, it's funny, Wall Street doesn't really understand yet in my opinion what Splunk is because it's such a modern, it's one of these modern companies that come by and it's just like, okay, looks a little different until they have that continued performance. I don't think they'll truly understand it, but you've got big data, you've got tooling, you've got a platform, it's pretty robust. But the cloud is really where the action is. So there are some moves of the cloud, a lot of their customers are. So what's your take on that cultural shift and is the technology in place to combine the on-premise goodness of data into the cloud? I think the cloud poses some really big problems for the banking industry. Its security is their key problem. Another key problem is always performance. Everybody wants to know they're getting the best performance for their money. And there is a very strong opinion in many places that if I buy my piece of metal, then I know it's mine, I've got it. No, I'm not sharing it with anybody else. I'm not losing to anybody else, in fact. And mutualized storage is showing that problem where in a large enterprise, you can find you have a performance hit from someone completely unrelated to you. So I think that clouds are, they're beginning to come in. I know several banks that are using them, but they tend to be private. And you have to have very, very strong metrology around that in order to ensure and to give your internal clients confidence that they're getting what they're paying for. Yeah, it's almost conflicting or fear on one hand and also promise on the other. Customers want what they want now. Mobile apps are really important. So you got to have mobile infrastructure. That's a client services issue. When people want to have the best place and greatest user experience. At the same time, you don't have a perimeter in the cloud. So you don't really can't even permit or security in the data center these days because of the app. So it's kind of interesting, right? So we're in this dilemma. Okay, I want the best of today's modern apps, APIs, notifications, all that goodness, but I got to lock it up. So what's your take on that? I mean, it's not everyone's agenda. It's not everyone's mind. What's your thoughts there? I mean, the banking highlights it straight up because of the data, the money. It's like the bank. It's always been a worry of mine that we put so much money into performance. We buy the best hardware. We buy, we have great systems and the internet in general, as Mark was saying just now in the security keynote, we're in this fetid miasma, as he said. My worry is that with all of the security that we're having to see at home PCs, they're a great example, the virus scanners. You know that your computer has got much more power than you're able to realize because you're having to weigh it down with all of that security. So, and again, that's another reason why I think many environments, especially banks, banks are relatively secure places. We all hope for our own savings and our security. And so they put a lot of infrastructure in place outside to ring fence what's inside. So it's classic trade-offs right now. So you mean straight up, okay, you know, you want security then you have to give up a little bit of that freedom or elegance or in this case, you know, apps. So that's a classic trade-off. Now roadmap, what's your take on the roadmap of that evolution? What's the vision from your standpoint of, you know, those table steak, check box items, check the box. It's like on the airport, go through security. You need to have those minimum things. But is there a roadmap in your mind that you see with DevOps, with, you know, agile programming, infrastructure as code? Is it virtualization? Is it, you see some new tech that you like that gives you some hope and promise? I mean, certainly we're only in kind of worrying there, but now there's a promise side. How do you address that roadmap? Well, I think there's so much cool stuff now available in the virtualized world. Especially in banking, you've never got enough systems. You always want to run lots of new tests. And I know that you can find lots of cool tools, like Delfix for virtualizing databases. There's all sorts of things that we can bring to the table inside the banking industry, inside the private cloud, which will enable us to scale out the hardware on demand. We can run tests. We can build test environments extremely quickly. And I think banks are going to really have to get to grips with the virtualization, understand it, feel comfortable with it. But I think with the DevOps movement, you're also thinking about proving your security from day one and building that into your code and building that into your tests. And for those tests to pass automatically so that when the code gets deployed in hopefully a continual delivery environment, that everything's already there, it's already proven. And, you know, the financial industry, financial sector, if you will, in banking in particular, have been great early adopters of all the latest and greatest in terms of when it's reliable. Fast computers, all the speeds and feeds, big data certainly, arbitrage, you're looking at fraud detection to up and down the benefits that are there. So with that being said, what's next? I mean, how much more performance can you wring out of that? I mean, and what would you share with folks of things that you've learned? Because it's banking and financial early adopters on all the cutting edge stuff. Some say better than the government in terms of taking over the security because they have to protect the money. But, okay, for folks that aren't on the edge like you guys are in terms of the latest and greatest, what advice would you have for other IT guys who are scratching their heads like, okay, I want security too, and I want to balance those trade-offs? Well, we've found Splunk a great help for that because in the banking industry, you're finding that even your developers can hardly get onto the production system. One of the challenges we always have is developers are writing code but they very rarely talk to an end user. That's where DevOps kind of comes in, bring the guys closer together, but then it's also very difficult for a developer to really understand how that application's performing in the live environment. If he can't get onto the machine, typically he won't have all the same monitoring tools as the support guys, but you can extend that out very easily with Splunk. You can give them access to log files without giving them access to the host. You can build dashboards quickly and simply. They can see for themselves how they can extend their own logging to help Splunk, and when you start putting that in front of your business users as well, they can start to see. So you get them closer to the customer, closer to the front lines, if you will, and two, you get them some data to play with as well as like a development source. It's like candy or playground, here, go play and let them saute with that. It's amazing how creative they can be sometimes. I know developers spend all their day creating new wonderful solutions to problems, but when you put the data in front of them as well, then they start to see the value of that, and we've seen that ourselves. Have you seen creativity come out of the engineers? Is normally they're looked at as guys in the back, banging away code. Yeah, I mean, typically you're feeding them jeeras and just getting new releases back. Hopefully they work first time, but no, we're really seeing new ideas, especially with the front end developers when we put Splunk in front of them. We started to get much more rich logging coming out, so they were able to leverage the power of Splunk, be able to understand. One example we had, we got the front end developers and the back end developers, so we have a system where traders can simulate a trade, do a what if. Trader has a limit, he needs to know if the next deal is going to blow his limit or not. Now this front end would have problems. Every system does. And so with getting the developers together with Splunk, they were able to, if an end user has a problem, they click a button and it gives you their session ID. They give us the session ID, we can now do a full front to back story of what went wrong. We can see the journey. How they got there. We see the journey all the way through from the front end for its Tomcat servers, back down into the calculation engines, whole journey through the calculation engines and back out to the client's desktop. And that's huge. Yeah, all you do is just, and you don't even need to index secret, you can just put the session ID in and Splunk will give you everything very simply. So you sort of answered the question I was about to ask, which was, does this require your developers to develop new skills essentially around data analytics that maybe isn't in their sweet spot? But it sounds like Splunk lowers the barrier to manipulating and understanding data in a way that maybe they don't have to become data scientists per se, but they can still understand and get what they need from the data. Is that accurate? Definitely. Because I think once you open the window onto this kind of data, people start to think of new ways of using it. And if I can expose this, what else could I expose? Instead of writing the typical way that a developer will write their log files is often after the event, not just before I'm about to do something, so it can make it pretty difficult for a support guy to work out what's not happening. But also, so, I'm sorry, I lost that thread there. Well, so let's expand a little bit on just the kind of the culture of this kind of data-driven decision-making in application development. So, I mean, do you see this, we just have to start of this kind of DevOps movement, if you will, and what role are tools like Splunk, but other tools as well, that are making it easier, lowering the barriers for people to actually start using data? How important is that going to become in terms of the application developer role? I think one of the key problems I've found is finding a hard, concrete definition of what DevOps is. And I think that shows you how early we are in the DevOps movement. Now, I know that we've got DevOps days and we've got lots of things going on, people getting involved, and I think a lot of us kind of know what DevOps is. We've got the Phoenix Projects book, it's a great book by Gene Kim, and amongst a few others, and you've got the talks of, say, Jess Humble has put together a velocity conference and QCon have seen some good talks there, but I'm really looking forward to the DevOps cookbook coming out to really try and help provide this good foundation of information to show you that you can go to senior management and really show them what DevOps is with the things that you're trying to instill, because DevOps needs to be a kind of top-down approach. You need everyone at the top bought into the whole process because after all, you're trying to build a new culture and you've got to really protect that culture, because one of the problems is you can very easily under extreme pressure of tight deadlines for people to suddenly start going off into their own retreating and not sharing what they're doing and not keeping everybody in the loop. So I think that's something that we really need to look at is how we can keep everybody together in the DevOps culture. That's one of the challenges, I think, is it's not so much on the technologies and it's not so much on what we should be doing. It's keeping the discipline of doing it. When times are good, it's very easy to do these kinds of things, but when times start getting really tough and you're maybe approaching a deadline and you know you've still got more days work to do than there is before the deadline, then how do you keep this going? And so part of that you mentioned is kind of getting that top-down buy-in. How do you, is it easy to translate the benefits of DevOps to a higher level, C-level person who's not necessarily in the trenches every day? Do they intuitively get it? Maybe they don't call it DevOps, but they understand the agility that's required to serve customers with applications that are responsive to their needs? I think when talking to senior management, you've really got to show them what the benefit is going to be for the client. Because ultimately, most senior managers, that's what they're focused on. They know that that's their reputation, that's their own image. When they stand in front of their clients, they're going to be judged on what they're receiving. So that's one of the main selling points and for me, the underpinning thing of DevOps is the inclusivity of everybody. It's getting everybody involved, everybody on the same page, everybody working towards a common goal. And under the old-style models, say in my own experience, we worked in financial institutions where the technology department is in a completely separate business environment. You have two different agendas. You've got the client who wants the functionality, the technology group that wants to improve the technical architecture and make a more technically sound and the long best practices kind of system. Whereas if you're all together in one group, then everyone can have the conversations and understand, well, we can give you this functionality, but at the same time, we need to keep the road on the show by building a sound underpinning to all of this. So you can have misalignment with the old model where you've basically got two different sets of goals for the two different groups for IT in the business and inherently you're going to have tension in that model. The collaboration model, the DevOps model, helps solve that problem, but as you said, actually, one just kind of migrating to that model and getting everybody to change the way they've operated for so long and then anytime you've got collaboration, you're going to, it could get messy. So the way we approached that was not to try and do it wholesale, it started off with an individual project, a well-defined small project and kind of prove how it could work there. So last year, we put together a team, it had members of the business, members of business analyst groups, the developers, support analysts, all working together and we really saw the benefit there because we had tight deadlines, we could see, we were able to shorten the feedback loops, prove things were okay or not before they went to the live environment. Whereas under old models, it's very easy to not have enough time to test and then end up with finding the problems more often in the live environment or at the time you're actually installing it which is an even more worrying situation. Not ideal at all. Dave, really appreciate you coming out, I'll give you the final word. Share the folks out there here at this event here. What's the most impressive thing? Or let me say this, what's the theme? What's the vibe? Share with the folks who aren't here. What is this conference here about this year? Theme, Mojo, just what's happening? I think this morning it was very exciting to see the different use cases of Splunk and what's coming up in the future and it's kind of underpinning the sort of things we've already been doing ourselves which is the analytics for everybody and that's the real thing that I've seen as exciting in our own brief introduction to Splunk over the past year is how everybody finds a use for it, whether it's a developer, a support chat, senior management, a business user. We've trained everybody in our department how to use Splunk. Just very simply, it didn't take long, just about an hour to train them and it's this idea of opening up all of your data to all of the people who should have access to it, obviously not just everybody, but opening it up to everybody and then allowing people to use that creativity and find new ways of slicing our own data. So, yeah. I think the analytics for everyone is awesome and I think your point about developers being part of this process because automation's got to come from somewhere. You've got to abstract away the complexity. You've got to write some code. Definitely, definitely. And it's so often developers can be seen as part of the problem, causing the problems, whereas now they're seen as part of the solution which is great for all of us. You know, I've been called to Visionary, but I wrote a post in 2007. It's the data's the new development kit and I think that's definitely happening right now and the people who put data in the hands of developers really drive that infrastructure's code. That's the future. Splunk's going there. Really appreciate you coming on and giving the insights. You guys are on the cutting edge and you've got to be strong, relevant and also secure. You've got to trade off to protect the money and give the users great apps. It's a tough challenge. You're worrying a lot. So it puts a lot of promise. We're here inside theCUBE, breaking it all down. Jeff Kelly with myself, John Furrier. We'll be right back after this short break.