 Okay, we're back live here at Stratoconference in Silicon Valley in Santa Clara, California. I'm John Furrier, the founder of SiliconANGLE.com. This is theCUBE, our flagship program. We'll go out to the events, extract the signal from the noise, talk to the thought leaders, CEOs, developers, whoever is extracting the signal from the noise and also startups. And we're excited to have a startup right here just recently funded and hiring and doing a lot of stuff in a very progressive area in real time, streaming, age streaming. Again, I'm John Furrier, I'm joined with my co-host. I'm Dave Vellante of Wikibon.org. And as you know, we love to bring you new and exciting and maybe not so well-known startups and technology innovators. And we're here with a company called H-Streaming. Yana Ulig is the co-founder and chief executive officer. Yana, welcome to theCUBE. Thank you so much. So the buzz this week has been, this week and even the last Stratas, been all about bringing SQL-like capabilities to Hadoop, to do BI in real time. Now that's kind of a misnomer. And the world that you live in, how you define real time is dramatically different than what I just described. So talk about that a little bit and then we'll get into, tell us more about H-Streaming. Okay, so let me first tell you what we are doing and then I compare it to what you just said. What we do is we help companies to convert big data into value. So we take data when it's still streaming, when it's still in flight, as it gets generated. And we put it through our strategy engine to unhelp companies to derive some action out of data. Before it's stored. Before it stores. Okay. As it streams in. All right, so tell us more. Our audience I'm sure is intrigued as I was when I first about this. Is like, really you do that? How do you do that? What do you do that for? Who does that? So tell us more. Okay, we do it on Hadoop. We do it on Hadoop. We take the data as it gets generated. We push it through our strategy engine so that customers can be doing, repricing their products, catching the right customers with advertising, trading, they can do all sorts of different things or just triggering in real time business processes as the data is coming in. Because as you know, the data is coming in such a volume and velocity is just impossible for anybody to be sorting through it and looking for it to do with it. And big data is getting stale very fast. So when it's really critical that they take action on the data as it comes in. So streaming is obviously a big buzzword. I'll see people that in the inside the big data world know. It's hard to harness real time. We see Impala, we see EMC trying to push that same message in others. And you see companies out there like Twitter using it. And there's open source like Storm. So talk about what people are interested in about streaming. What's different about what you guys are doing around real time data? And how does that compare to some other approaches? Storing it, building a corpus, doing data mining, doing data warehousing, doing the normal business intelligence. What are you guys doing that's different? And what's that value proposition for the market? Okay, the other products that you mentioned except of Storm are doing a real time query. And that is very complimentary to what we are doing because companies usually start first digging into historic data, looking for insights. And the faster they can and the more interactively they can get to an insight, the better for them. But once you have an insight, the insight by itself doesn't have any value per se. It needs to get converted into an action so that you can do something with it. And this is where we come in play. So we are totally complimentary to the products that you mentioned. Everybody's pretty much trying to speed up cavaries to be more interactive and being more real time. What problems specifically are you guys solving? We are solving all sorts of different problems. I mean, we have customers that are doing a lot of video analytics that needs to be in real time. For example, in security. You know, in airports you have a thousands of camera looking and say somebody goes through a secure area. You want to know whether this is a security intrusion or not. So we can roll back the time and look what that person was doing in the airport half an hour ago, whether there was some behaviors that were awkward. We can match the data in real time with other data sources and then derive some intelligence to say, oh, that was a mistake, somebody just got lost. Or that is a true security intrusion somebody needs to get dispatched to. So that's an example, Yana, where you're actually taking data that has been stored and analyzing it faster than alternatives. Is that right? No, no, no, you're looking at the camera as it is happening. Okay, so what you said, you roll back 30 minutes. So it's, it's, uh- Yes, because we annotate- And comparing it to- We annotate all the videos that it's coming in and we hold it in memory. We span a window of timeframe over what is happening. So what we are doing is we are looking for patterns in real time. But sometimes that patterns, you know, you're looking for event one and event two and event three to happen at the same time that makes a meaning. So sometimes that event happened a long time ago, but the last one triggers that, wow, now it means something. So can you compare what you're doing with something like, for instance, that's been around for a while, complex event processing, which has, of course, been big in the financial services business. How are you similar? How are you different? Capability-wise, we are very similar what they are doing, but we are bringing in a totally new context because their bread and butter is just finance. It's made for small data. It's not made for built data. It can do, it's built for finance, pretty much. And what we are doing, we are extending the usage cases that anybody can do it because the big data is bringing in new problems because it's coming in so fast and furious you really have to capture the value before it gets stale. And that is a new paradigm that in the small data was not the case. So you're CEP for big data, essentially. That's right, and it's built all in Hadoop. So talk more about the tech behind it and what you had to do to develop this capability. We are coming from the analyst position that it needs to be easy for the analysts to work with that. So we built everything inside of Hadoop so that it looks like the normal environment that analysts used to work with. Now who's we? So you're a husband of team that started the company, right? That's right, and we have another partner. We're three founders. Okay, go ahead. So you just recently got funded by Atlas Ventures, generally specifically Chris Lynch is joining the board, is that right? So that's interesting too because you're a West Coast company. John, we've had Chris Lynch on before and he's always talking about East Coast, East Coast, East Coast, but he's made an exception. That's right, he never invests in a East Coast company. He doesn't like coming West. He's out exploring the West Coast. Chris, congratulations, welcome to the party, pal, as we say. But you know, seriously, Chris is smart. I mean, we know Chris is a great investor. He's aggressive, he's smart. He's got a good nose for the business, but I think what's more important, Dave, is I think what Chris Lynch is doing is consistent with what the top VCs are doing. If you look at the consumer web, Fred Wilson, Rich Leventhoff, other VCs are in all, we're in all the early deals on the consumer side, Twitter, Zynga. Those were West Coast VCs, East Coast VCs who invested in those guys. So what made you go with an East Coast early investor? There's plenty of opportunities out here for investors. Why an East Coast investor like Alice? Chris has unmatched a breath of experience, operational experience from Vertica, and what he sees in age streaming is a big parallel to what he did with Vertica. Scale it out, get many customers, and he's always saying there is a lot of usage cases that they had at Vertica that they just couldn't do because the latency's requirements were too short, and he believes that every customer that has an MPP database will want to buy our resolution. So that resonated with you, and that resonated with you in terms of your choice of investors. I mean, he's bringing a lot of experience on the table. So the thing that Hadoop brings to the table, and you and I were talking about this last night, is that Hadoop can bring value to the data warehousing side, but some people who are on the more innovative side would call that cheap data warehousing just by bolting on Hadoop into data warehousing. So I want to ask you, and this brings the conversation back to some of the core conversations, are you moving it around the network, does data move around the network, and how many transformations, but more importantly, latency versus throughput. Just what's your personal opinion as you look at the landscape of different solutions out there, there's pressure for low latency, and also at the same time, scalability and throughput? How do you guys look at that in terms of your solution compared to others? I mean, we have an amazing throughput as well as latency. I mean, we did test our system on 1,000 nodes, so we scale out and through one node and push up to two million events per second. So it's very scalable, and we do latencies in milliseconds. So I think you can have it both. So Asli, we're hearing, and we've always said, Dave, and actually I think we were at Oracle Open World, we said, when talking about software-led infrastructure and flash memory, the emphasis was on benchmarks. Now in big data, you're starting to see benchmarks come out, not always kind of the use cases that people would predict. Are they customer workloads? So benchmarks are interesting. So at the EMC Green Plum announcement, they're a hundred times faster than Impala and Hadoop. What's your take on benchmarks? Because benchmarks can be rigged, some say, but yeah, people are impressed by benchmarks. What do you think about that statistic a hundred times faster than Hadoop? Besides not believing it or believing it, what do you believe in? Too slow. It should be a thousand times faster. I think it's bringing a lot of power to the analysts so that they can get to the inside much faster and this is playing to our game as well. I think it's good for everybody if the ecosystem is bringing faster, better, and more user-friendly products in the market. So let's talk more about use cases. Obviously, ad serving is one. Talk about that a little bit and others that might apply. Okay, we have customers that are, we're in the mobile advertising space. So when you open your phone, you, an app, there is a space for an ad. It goes into the exchange so you have 100 milliseconds to fill that. This is what currently companies are already doing. What we can do is we can in the 100 milliseconds also do a lot of real-time analytics so that we can do a better targeting. So we can do behavioral analytics on what is the customer doing, how it is usually moving around. We can do capping, retargeting, and many other things, and that brings up the revenues for both the advertiser as well as for the app owner. So that you can move that hit needle higher up than you can with existing technology. It's actually dramatically changing the market of mobile advertising by doing better targeting. So you mentioned you've got customers in mobile advertising. So how long have you been shipping the product? Give us the sort of status there. We have a customer in production. We made some announcement around mobile advertising. Good job. Yep, so okay, so you're in production. Yes. Congratulations. Thank you. That's great for a small company like yours. And this is your first round of financing? Is that right? That's right. Okay, so really just getting started. So what kind of people are you looking for to build your team around? We are looking just for the top notch. I mean, we pushed it so far with three people. We have awards, award-winning technology by Gardner and CRN and other companies. We also have a shipping product, et cetera. So we are looking for salespeople for marketing that can get the bus out, as well as for top-notch engineer that would fit in a German engineering culture that we have. What do you think about the cloud? So we're going to do a segment tomorrow and cloud means big data. Obviously streaming is one of those things where there's a lot of event-based analytics and also different kind of a methodology in dealing with data versus just straight storing it to a disk and then working on it. Amazon has got a lot of traction in cloud. Elastic MapReduce is getting some traction because of ease of use. How does your solution fit in the cloud? Does it fit in the cloud? And what's your take on using the cloud for Hadoop-like environments? We work on any Hadoop distribution. So we have a partnership with everybody on the market. We are certified in Cloudera. We worked on EMC that you mentioned. We worked on MapR. We worked on every distribution, data stacks, etc., that are on the market. And we also work on AWS. Any results, any clients there on AWS? We have to say like most of our customers want an in-house installation and do not want to be working in cloud. Explain why. I mean, I know there's obvious reasons, but I'd like you to hear. Because most of the customers that really have a big data and have a need to making business decisions really fast are large organizations. Then they just don't trust their data out in the cloud. Well, right. Who does? It's okay, John, thank you so much for coming on the queue. Great to have startups. Congratulations on your financing husband and wife team. We'd love to see that in Silicon Valley because that's not always endorsed. You know, we see don't usually invest in husband and wife teams. Let's just go VMware. Well, it's the exception, not the rule, we know. I mean, I know, I've heard that before. Again, congratulations. You're hiring, got some seed funding. Good luck and you're in a hot area. Okay, hstreaming.com. Check them out. Growing San Francisco based startup doing a hot startup in the area of real time and streaming. That's fast data. That's something that we like to see. And Chris Lynch is a big investor with Atlas Ventures funding them. So we'll be right back with our next guest after the short break. Thank you.