 Hi, I'm Alex Williams from Silicon Angle, here live at the Stratoconference. I just love that ad, it's so catchy. I just want to dance, I mean, every time I hear that, we'll disco here on theCUBE. Here with the two fine gentlemen from Informatica, would you guys please introduce yourself and we'll get started with a few questions. Sure, my name is Hasib Badani, I'm the VP of products at Infaneta Systems. I'm Rob Ramarow, co-founder and CTO of Infaneta Systems. Great, what I'd like to do is just get a framework for the market right now. Tell me about where we are in the marketplace with WAN and big traffic, I think is what you call it. Where are we in the market? How is it compared to like a year ago? Sure, so we like to think of ourselves as the van optimization company for big traffic. We've been around for three years and in three years a lot has changed, so certainly in the last year, a lot of things have changed, particularly around how people are thinking about big data and how just a kind of non-sexy thing such as replication are making a big impact on enterprise vans. So where the market used to be five years ago, even a year ago, where a gigabit was a big link between data centers, now 10 gig is its table stakes and a lot of our customers are talking about multi 10 gig, 40 gig links between their data centers. And in order to address these kinds of van links, the approach you need from a van optimization perspective is entirely different. So that's what our focus is at infinite. So that really is essentially big traffic. That's exactly right. And what is the correlation between big traffic and big data? So in a number of cases, and you know what, I'll be honest with you, a year and a half ago, I didn't even know what big data is and I've heard of it before. Some of our customers who are large users of big data, they started looking at using multiple resources in different data centers for big data users, for Hadoop in particular. And they talked to us about moving data between sites to process it in larger clusters. And when they pulled us in to understand better how van optimization can help with big data, that's when we started looking at things like latency and how that impacts big data processes and the need for bandwidth between data centers and how that impacts big data end to end. So let's get a little bit into that. Maybe we can talk a little bit about the technical challenges. Virtualization has become mainstream. And so companies are trying now to figure out how they can actually now start to avoid these latency issues, because the latency issues are now more of a problem than anything else. And it costs a lot of money when you have latency issues. Tell me about the technological challenges around that. Now, as I said just mentioned, on one hand, we have a huge amount of data going between data centers. And on the other hand, any physical distance has to be taken into account when talking about latency. So you can go faster than light. So there is some inbuilt latency as soon as you have separated data centers. So now what happens is when you try to reduce the amount of traffic by doing something to that, like deduplication or compression or whatever, in that process you are introducing more latency into the end to end system. So what we are trying to do is essentially make sure that we do not increase the latency between data centers in any appreciable way at all. In fact, our products introduce about 50 microseconds of latency on each side. So for a total of about 100 microseconds, as opposed to any number of milliseconds and so on, that other competitors products introduce into the system. So I think of things like, you know, in my world, I think of things like memcached, right? Where, you know, you cannot, you can automate it to some respects that experience. Is there a similarity into what we're talking about here? Well, you know, yeah, go ahead. In a way, you can think of what we are doing like memcached, introducing extremely small amount of latencies, but at the same time, reducing the amount of traffic substantially so that you are able to send much more traffic without introducing any significant delays into that. And in the process, actually reducing the delay end to end because the amount of traffic being sent is far less than what you would have sent without us in the middle. So in fact, we are adding a lot more value in terms of reducing the overall latency end to end and making applications more productive and efficient. And you talked about the virtualization, so a lot of people talk about VM ability, doing VM motions across data centers. If you read the literature from VM motion, they talk about an end to end latency budget of say five milliseconds. If you want to do VM motion between any distances. Now, the challenge with that is if bandwidth is the problem that you're facing, because if you're doing many, many VM motion flows, what you're going to run into is, if you use any traditional van optimization systems, they will add a few milliseconds of latency in order to process that traffic. Now if your end to end budget was five milliseconds, if you had another four or five, that kills the VM motion. You can't do VM motion anymore. So our product actually is the only product that can actually work in a VM motion environment or even for that matter, for synchronous replication environments, which is something that van optimization has never looked at before because of the latency challenges that come with van optimization, particularly deduplication and compression. So deduplication and depression, and you're talking about that. Can you explain your process for that? Okay. So again, looking at the existing approaches to deduplication, basically they are meant for 90s, early 90s type systems, where you have single core IO band which is extremely expensive and so on, so they don't scale. Okay, that's the first thing. So you are hardly looking at maybe one gigabit per second type rates they can process. So if your link is 100 megabit link, it's okay to use their kinds of products. But when you go beyond that, when you're talking about one gigabit links, 10 gigabit links and so on, the amount of traffic you ought to process is now at least four or five times more than what the link can carry because you are reducing it. Okay. Which means we are talking about 40 gig, that kind of amount of capacity in the system to process the data. Okay. Now again with the existing systems, the existing technologies, that's not possible. So what we have done is we have started ground up to solve this problem, make it very scalable, at the same time reduce the latencies and also keep the reduction overall due duplication you get at the same level as the other guys can get. So we can get into more detail on exactly how we do it but essentially that's what we accomplish. Great. Okay. So where do you see the market going over the next year? What is happening in the marketplace now? And why is an event like this important for you? So particularly because of big data and what it does to the two band links, right? We have a number of customers actually telling us, you know, the 10 gig box that we build, that's great but a couple years out, 10 gig is actually not enough. So many of our media customers in the financial services industry, a number of our customers who are using big data as an internal resource for multiple of their internal clients, they're looking at 10 gig as, you know, at something that was a 2010 thing. In 2012 and onwards, they're going to be looking at much larger links. So they're actually forcing our hand and asking us to build bigger and bigger systems. So, you know, things like big data, you know, they are forcing our roadmap in some sense and you know, that's very important to us to track and understand, you know, where do we need to go? And the financial services market tells a lot about where the rest of the markets go. Exactly right, yeah. Tell me about the, what you're finding interesting here. I mean, one of the things that seems to be that I can try to deduce from what we're talking about here is like predictive analytics start to play a role in your service to some extent because the more data you have and you can't start analyzing that and start to develop efficiencies by understanding the patterns around that data. So is that something that you're looking at right now and predictive analytics going to be important for your technology as you go forward? Yes. Yeah, actually predictive analytics is one area and also the advancements coming through SSD type technologies where the bandwidth has increased substantially and now we are able to do inter data center processing for huge amounts of data. So those are the areas where we are very much interested in working with some customers and some other partners and so on to get our solution to deal with those problems directly. So who are you seeing as the customers then for this solution? So we spend a lot of time with financial services so my boss and I, we spent a lot of time in New York for sure, right? We've had a lot of pull in the federal sector so we have a number of intel agencies who are our customers which is surprising for a small company in the early days and in addition we're working with a number of internet centric companies who have a lot of bandwidth and they're really frankly, if you have a latency issue you can keep throwing bandwidth at it it doesn't really solve the problem. Is services a big component for what you do? We try to sell systems that are very easy to manage being a small company but certainly we provide a lot of very close handed support to our customers. Who are some of your technology partners? So right now we're focused on storage partners actually so we're working very closely with EMC to get certified and then we're going to go down the line of the storage guys, the EMCs and the NetApps and so on get certified with all of them and work with them in the field because that seems to be where we see a lot of pull in that direction. How about a provider like FusionIO? Certainly, as Rob mentioned, we're looking at a lot of SSD technologies for our system so we're working with somebody like FusionIO we haven't gotten to FusionIO yet but all in good time. So we have cash in that respect it might be of interest too. Sure. Right, right. That seems to be really kind of important for your company going forward to be able to take advantage of these very powerful storage environments. Right, and you know, I will say the thing that runs on the thing that's really cool about the technology is that it does not assume anything about the underlying hardware. So today if we use memory or tomorrow we use SSDs it's an extremely scalable system. So today we use it for van optimization and there are so many other users of this deduplication technology that we haven't even started looking at and that's what's really exciting about this company. Well, terrific. Well, thank you very much you guys for coming and joining us today. We're going to be right back with more interviews here on theCUBE. I'm Alex Williams, a Silicon Angle. We'll see you soon.