 Okay, we're back live in Strata. We are siliconangle.com and wikibond.org. We're bringing you independent analysis of the Strata conference. This is our day two, day one of the conference. We've been broadcasting live all day, getting deep, talking to all the people here. We can, all the smartest people in the room we can find to extract a signal from the noise and really digesting the Strata conference. wikibond.org is the best research firm on the planet when it comes to big data and we're the first to put out the real groundbreaking report around the market size of big data and revenues by vendor and you can find that report for free at wikibond, wikibond.org slash big data and you'll find some really interesting analysis there. This market is real, it's growing, it's profitable and it's only going to get bigger. I think they undersized the market, still massive and the $50 billion day was a little bit smaller on the estimate side. I think it's going to be bigger. I'm John Furrier, the founder of siliconangle.com and I'm joined with my co-host. I'm Dave Vellante of wikibond.org. John, I hope you're right, I think the market actually could be larger because it's, big data is everywhere, it's pervasive. We just had pervasive on, but we have a great guest now, Yves de Montchoy who is VP of marketing at Tallinn. Yves, welcome to theCUBE. Well thanks for having me guys, it's great to be here. So Tallinn's kind of has grown out of this concept of master data management. It's really your sweet spot and now we're seeing this in big data but why don't you start by telling us about Tallinn and then we'll get into the big data angle. Sure, so Tallinn is essentially an open source integration vendor. The company was founded about six years ago with the promise of democratizing integration. It was initially positioned as an alternative to the proprietary vendors in the ETL slash data integration space. The idea was to bring data integration technologies to companies of all sizes for all types of projects to prevent them from having to resort to manual coding for their integration projects. If we fast forward to today, the company has grown quite significantly. We have first expanded into other areas of data management, getting into data quality and as you said master data management. We then branched out into application integration and business process integration and last fall we actually released a complete unified integration platform that covers all aspects of integration all using an open source approach. Now, we are here at Stratac Conference because we have been recognizing this big data trend for a long time and actually Tallinn is not a newbie in the Hadoop world. We've been supporting Hadoop for about two years as far as connecting to Hadoop, making it simpler to integrate Hadoop with the rest of the information system. So you started in an open source. How did you come about and why did you start with that open source philosophy? Talk about that a little bit. Well, at the time where Tallinn started, open source was starting to gain some traction in some parts of the IT stack. You had, of course, the operating system, the database, the application servers, but there was no business applications. There was no middleware alternatives to some of the largest players on the market at the time. And our founders, Bertrand Yaron Fabrice Bono, were implementing data integration technologies from our competitors, and they felt that need for something else, for an alternative. They recognized that opportunity. They created Tallinn as an open source integration vendor. Okay, so are there two dimensions of that? They saw the problem of integrating various open source platforms and then taking an open source philosophy to give back to the community. Am I understanding that correctly? Correct, this is the... But, David, it's not only about integrating open source platform, it's really about integrating the different platforms in the information system, whether they are open source or proprietary. And Tallinn is really about building bridges between all the IT systems, databases, applications, software as a service, cloud-based systems, and to integrate by moving data and transforming data around, also to integrate via the processes, via services. So, this notion of data integration, master data management, it's all about breaking down these silos, like you said, building bridges. It's absolutely bridges between silos of data, making sure that all applications are consistent between one another, and also that the different integration processes in the organization are consistent with one another, so it's at several levels really. So a lot of the promise of enterprise data warehouses, for example, and many business intelligence efforts have this single version of the truth. Now we have all this big data explosion, and Hadoop comes along, gives us access to all this new data. Does it pull us away from that goal of single data, single version of the truth? It doesn't. It's actually a very interesting complement to that single version of the truth. It's bringing more data into that single version of the truth. You are not restricted to only what you were able to bring into the data warehouse, but you have access to a lot more different types of data that will be sitting alongside the traditional data in the data warehouse, or that will even in some cases maybe replacing the traditional enterprise data warehouse. So as your vision, you're creating this data management hub, this data integrated master data management hub, is that? That's exactly what it is, and the biggest challenge for that is to bring all the relevant data into this data management hub. And for that, you need of course connectivity to all the systems where the data resides. How many website is some great customers? You know, huge list of customers actually, Experian City, Monster.com, talk about some of your Thomas Cook, talk about some of your customers, I mean, some great names on there. Well, so first of all, you're right, we got lots of customers, and being an open source vendor, we are also a commercial company, and we have about 3,500 commercial customers, people who have bought the enterprise versions of our solution. And some of them are mid-sized businesses who are just replacing manual code with, I would say, enterprise-grade integration solution. Some of them have embraced the entire data integration, sorry, the entire integration stack, data integration, application integration, and business process integration. Actually, a number of our customers have been using our big data and our Hadoop integration solution for some time. You looked at some logos on our website, Dave, and there is an example pops to mind of one of our customers, a bank, which is actually using Hadoop as an auxiliary engine to process data before loading it into their enterprise data warehouse. What they are getting is just, I would say, mind-blowing performance out of an ETL engine that is relying on Hadoop as a transformation engine based on code that is generated by talent for big data to drive those transformations inside Hadoop. They are essentially bringing a lot more data into their enterprise data warehouse that they would have been to bring with traditional ETL processes. So, Eve, is it an approach that you have an open source version of your solution, you give it away for free, allow people to download it, play with it, and then you sell an enterprise addition on top of that? Is that right? That's exactly the model that's called the open core model, yes. Yeah, and I mean, that's a common approach in this business. Some people are maybe taking a somewhat different approach and maybe focusing on services, for example, but so talk about how it's working for you. When you originally conceived of that model, there must have been a lot of discussion and debate about that, especially a couple of years ago when this was so new. So how has that progressed and how is it working? Well, the key in building an open core model and making it work is to find the balance between what you give to the community and what you are selling. What you give to the community is what's going to provide real value in the open source version and ensure that you're getting sufficient adoption and sufficient use by the community to maintain the viability of the open source version over time and to grow a community. What you sell, it's what's going to sustain the business and it's going to pay for the salaries of the engineers who are building both the commercial edition and the open source edition. Of course, we are getting contributions from the community to our core, but we need to have engineers who are driving that process. Otherwise, the product will just slow down and probably cease to exist at some point. So it's not completely unheard of, but it's somewhat rare that we have guests from outside of Paris on the Cube. I know there's some technology and plenty of innovation going on, but talk about that a little bit, the geographic location, the talent pool, the distance from places like Silicon Valley, what's all going on there? Well, that's an interesting question and it's true that talent started as a French company. Our two founders are French people, but the initial assumption was that one of them would relocate to the Silicon Valley and Bertrand, our CEO, actually moved to the Silicon Valley within the first six months of talent's existence. He's based in Los Altos and he lives here with his family, very happy and I don't think he's going to move back to France. So talent is truly a multinational company. We have staff on three continents, we have about 400 people and it's about 1.3 in North America, 1.3 in Europe and 1.3 in Asia. And when we launched the first version of Tan Adoption Studio, remember it was in October 2006, we had no marketing budget, we were not doing any promotion, immediately 40% of our downloads were coming from North America. So with open source you don't choose where the product is going to be used, you just put it out there, word of mouth plays its role and traction and adoption is going to come naturally. Then based on where adoption is coming and of course the size of the business in a given country, you're going to choose where you deploy your business operations and today I mean, no surprise, our business operations are the strongest in the US, in France, in Germany, in the UK, in Italy and in Japan and that's where frankly a lot of IT companies have their operations also. Right, okay, so you've got Silver Lake, Baldurton Capital, some other private equity firms and so where are you in terms of the company and its funding, it looks like you're well funded? We are extremely well funded. We've raised approximately $62 million in a few rounds of funding. 62 million. 62 million, yes. Our latest investor, Silver Lake as you pointed out, is actually one of the largest VC funds in North America, IT funds in North America. Baldurton Capital was a very interesting investment for us because the general partner that joined talents board for that investment was Bernard Liotto, was the founder of Business Object, another extremely successful software company and frankly we are very proud to follow in his footsteps. Yeah, and there's some good action that Baldurton has, they mentioned MySQL and others, so good. Okay, so you've got plenty of capital, you're not looking for around 62 million, John, that's some serious. If they raised a couple more million they could almost be like Yammer. Yammer, right. Yammer, Yammer, damn, they're loaded right now. Maybe they're a big data player. So how much do you have left on the bank, good amount? Oh, we have sufficient to continue to fund operations or investments, absolutely. Great, well congratulations on the funding. As I say, you can never go bankrupt if you have money in the bank. That's a good business model. Talk about your deal with Hortonworks and how you guys are going to market with those guys. Obviously we've been covering them since their launch. I kind of fell on my sword, as they say, when they were announced, I kind of gave them a little bit of a hard time, but they handled themselves well with coming out and competing with Cloudera. They really were classy about it. There was a little bit of skirmish in the communities around who contributes more code. That kind of died quickly when everyone realized that the market's growing so fast. And competition's good when the code is being donated into open source. So that's positive. I kind of gave them a lot of kudos for that. So tell us about your deal with Hortonworks and what does that mean? So John, what we announced today jointly with Hortonworks were a couple of things, actually. The first one was that we are going to release Talent Open Studio for big data under an Apache software license. So we have experience with the Apache license. We already have technology under Apache specifically in the application integration world, but in data integration historically we've been using the GPL v2 license. Now, obviously, Hadoop is Apache. Most of the big data stack is Apache. So it makes a lot of sense to have integration for big data under Apache license. So we made that decision and that's something that we are acting upon. The other thing we announced today jointly with Hortonworks was that Talent Open Studio for big data would become the data integration component of the Hortonworks data platform. So HDP, which already includes, of course, the core Hadoop component, MapReduce, Scoop, Hive, Pig, et cetera, is going to include in there. So ZooKeeper is part of HDP as far as I understand, but that's something that's related to the management of the components, not to integration itself. So let's relevant as far as integration is concerned, which is why I didn't mention it. So HDP is now going to include Talent Open Studio for big data as a way to bridge Hadoop with the rest of enterprise IT. So get data in and outside of Hadoop and into Teradata or from Salesforce.com or from Oracle or MySQL or whatever. So link Hadoop to the rest of enterprise IT systems and as a way to transform data inside Hadoop without having to write MapReduce code. And frankly, that's a big thing. It's very complex to do MapReduce. Requires a PhD in, I don't know exactly what, probably invented it's PhD yet. Well, of course, you guys are a great product and one of the core issues that we've been talking about on theCUBE here yesterday and today is really two conversations. One is what I call the pure green field Hadoop opportunity that Cloudera is leading with Open, now Hortonworks and others. And then there's the existing legacy enterprise guys who have existing data warehouse business intelligence. So the legacy business intelligence and data warehouse guys, I mean, huge installed base of existing. They're not going to just throw that away. So how to integrate is a big question mark for them. Do they put a wrapper around it? Do they fence it around? Do they, how do they integrate? So what's your advice to the folks in the business intelligence data warehousing world around the integration challenges and the roadmap to successfully bridge those two worlds together? Because there are predictive and real time analytic benefits from the new world. Well, frankly, I see two interesting avenues for this kind of situation. The first one is to extend the existing enterprise data warehouse with a Hadoop cluster and to, I would say, have some data, traditional structure data in the enterprise data warehouse and have the less structure, the semi structure data into the Hadoop cluster and to use those in combination, which means that you need to have BI and analytics tool that are able to cover those two worlds. And we have great partners, for example, Jaspersoft, which has this kind of capability. The other one is, the other possibilities I was mentioning earlier is to use the Hadoop cluster as like an ETL engine on steroid. Get much more data, much more, I would say, aggregated data into your data warehouse by using Hadoop as the pre-processing engine, but also the cleansing engine. And all of that stuff, obviously, can be driven by code-generated by Tanandopan Studio for Big Data. And what is the challenges in terms of dollars involved? I mean, just give me order of magnitude, operational budget that someone should consider when looking at that integration task of going in and really bridging the Hadoop piece of it. Is it millions? Is it, you know, proof of concepts? We talked about scaling of what size these proof of concepts will be, and what are you seeing out there in terms of the budgets inside? It should start with a small Hadoop cluster, or you are certainly not talking about millions, you're talking about maybe tens of thousands to deploy and run that cluster. And, you know, Tanandopan Studio for Big Data is free downloadable under an Apache license, so it's not going to add to the cost of the infrastructure. The ramp up is actually extremely quick. All you need to, as long as you understand your sources and your targets, it's a drag and drop graphical design that doesn't require any deep expertise, any coding or whatnot. So we are clearly not talking about huge amounts of money for getting started. After that, it's a matter of how much you're going to have to scale your Hadoop cluster to execute those very complex transformations, I think. Okay, well, hey, congratulations on all the cash you have in the bank and the success you're having. Obviously, hot area integration is a really important part of it. Congratulations on your Hortonworks deal, and thanks for coming on theCUBE. All right. And tell everyone in France that the siliconangle.com and the Wikibon Cube is the hottest thing you've seen in Big Data. I'd be very happy. You actually get a decent amount of traffic from France, as I'm sure Mark Hopkins knows, and John, you probably looked at the logs. Must be in the furrier last name. Fudier, it's a little bit of French, that's taking a claim, you know. I'd be happy. The French side of me. I've spread the world. Yeah, I had a really weird background. French and Irish, so it's going to really mix very well, but, you know. Hey, it works. All right. Okay, thank you very much for coming on theCUBE. Talon, great company. Merci. Merci. Merci. Au revoir. See you later. Thank you. Okay. We'll be right back.