 I'm John Furrier, the founder of SiliconANGLE.com, and this is theCUBE, our flagship program. We've got the events. I'm sure you can see them from the noise. I'm my co-host. I'm Dave Vellante of Wikibon.org. Check out Wikibon. We just released a new report on software-led infrastructure. Of course, software defined is all the rage. We've defined that. We got a ton of information there on Flash, which is relevant to our next topic. We've got Ashish Gupta, who's the director of marketing, a violin memory who's a hot company, doing an IPO, all the buzz in Silicon Valley is around violin. Ashish, welcome to theCUBE. Thank you. Yeah, so it's good to have you on here. So you're a big partner of Dell's. You got a presence in the booth here. You guys move a lot of product in partnership with companies like Dell and IBM and even HP. And so, talk about, give us a quick update on violin, what's going on here at Dell World, and then we'll get into it. Absolutely, so, violin's been obviously president in a lot of events outside. We've announced a partnership back in August with VMware. Since then, we've had our presence officer at Oracle Open World, at Sapphire and SAP. And now we're Dell as another strategic partner of ours. Clearly, the idea is that the position that violin's in is it's a unique position. We obviously go after the traditional storage market, but our partnership on the compute side is absolutely relevant to the new generation of storage that's being brought out to the market, that's being led by violin in that space. The idea that Flash is going to create a wave in the storage industry is absolutely foundational to everything that violin does. Again, if you recall what we're bringing about to the market is essentially storage at the speed of memory. If you look at what Flash does, it's almost memory-like without all the drawbacks of memory. We've got essentially persistency, economics, and shared nature that we can bring in. And if you build a system from the ground up like we have, you start bringing some compelling economics to the market. And that's the reason why working with Dell, and other compute partners of ours, really brings a fundamental difference to the customer base. So the question that comes to mind, actually, you know, if we love Flash, we'll cover you guys, Fusion I.O., Solid Fire, another startup, Aero, Spike doing some in-memory stuff, kind of the same related trend. It's a mega trend, and you guys are out there leading the charge. Michael Dell talks about the data center in the future, his modern infrastructure. How do you guys fit into that? I mean, obviously, caching layers, and people talk about old Flash as caching layer, but it's just more than that. You just drill down, and just share with the folks, what's, where you fit in this modern infrastructure, and are you just a caching layer? Because that's really what people think of when they don't really do their homework on Flash. That's a great, great point, John. The idea of Flash as a caching layer is actually a bit of a misnomer. And you think about really what the data center is today. It's got a lot of components that are really taking it to the next level. We've talked about data center transformation over the last five, six years, and certainly on the compute and networking side, that's born true. The storage element is sort of lagged behind everything else, and Flash has sort of brought in a little bit of a spice and excitement in that market. However, if you don't do things right, it kind of leads it to that special place called Cache today, and we'll just re-center workloads, et cetera. We feel that that's sort of short changing what Flash can do. It's low-hanging fruit. It's low-hanging fruit, and works for particular market segments, and for people that want to protect an existing business. Clearly, we're not hampered by any anchors of a project expanding this market that we're trying to go after and sustain our ecosystem around. We want to go out and disrupt that, and that's really what Flash is doing for us, is essentially allowing us to really take that media and really bring it out to market as a new tier of storage, if you will. And the idea is that memory and memory computing that we started talking about is interesting, but if you look at what people want to do, it often causes applications, other environments to re-architect themselves to fit in that ecosystem. What we're saying is you can get memory-like performance with storage-like economics and storage-like infrastructure environments. So you bring in that same application that you're doing today, but get the memory boost if you build an infrastructure like ours, which is a foundationally re-architected solution that's taking flashes of media and building a new generation of storage around it. If you add that into your infrastructure, what you end up getting, and like Michael Dell was talking about, is a true Silicon data center of the future. And that means that the last vestige of non-silicon environment of just storage really can be brought into the 21st century or 22nd century. So you're essentially saying, if I'm referring correctly, that you're developing software on top of that architecture to exploit the new technology, the persistent memory-like technology. Can you describe, first of all, is that correct? So I'd say if you look at what violin has done, we're a nice hybrid between a software and hardware company, right? A lot of our IP is actually hardware accelerated. So if you look at the system, we actually have a custom controller that's on the actual array. And there's actually 64 of these controllers on the array. A lot of the IP that we do is embedded in that controller. And so what that does is essentially allows us to go out and scale and bring our software IP that we have accelerated hardware across in the entire array. So we are a software company in that we have got a lot of innovation, but we actually hardware accelerated. That's the only way you can actually bring this whole memory-like infrastructure. So compare and contrast that to somebody I'll just say it, bolting on Flash to a traditional sand controller. Specifically, how are you different in terms of the value that you deliver to customers? So there's two ways that people add Flash today, right? If you add it to the controller array layer, you're primarily doing it from a caching perspective. The cache gets warmed up, your service reads through that environment. If you add it to the SSD environments with their standard legacy array with the controller and then an SSD shelf underneath it, you get a certain level of boost. Absolutely no problem there. However, the legacy infrastructure is designed with spinning disk in mind. The buses that those controllers attach to have a certain bandwidth associated with them. Those buses get completely loaded up when you bring in a shelf in there with an SSD or a bunch of SSDs in it. Meaning the controller is completely saturated the moment an SSD gets introduced to the market. So when you're buying a spinning disk environment and putting SSDs in it, that controller is being, I'd say, inoptimally utilized at that point in time. That's where we come in. We basically change that economic model. We were designed from the ground up. And so what we do is bring in an array like ours which is completely designed for flash. And what we do is because it's designed from the ground up give you latencies that are equivalent or almost in memory like environment. So you get microsecond latencies that are sustained resulting in really fast response time. So when we're trying to get real-time analytics done that's where the latency is coming to play. Now when you're trying to fit into an ecosystem you want to be able to be storage-like. So we are a traditional stand environment where we give fiber channel in Finneban high scuziba 10 gigaton and traditional shared storage infrastructure like look and feel. So you get memory and storage combined into one. Without disruption essentially an application sees a single level store essentially. Absolutely. So you mentioned analytics. Talk about what you do or do you do anything for something like a use case in HBase, for example. So if you look at what people are trying on HBase or any big data environments, et cetera, it's really about getting access to information fast. And the idea that what Violet is trying to address is really around two parts. Most of it is around latency which is basically compute. If you look at compute utilization or CPU utilization it's looking at really the IO wait times that's a big challenge today. Utilization is 100% while in reality the compute's actually doing a lot less work. The processing time may be 30, 40%. Most of the other time it's waiting for IO. And so from a pure big data analytics perspective the faster we can service those requests the better it is which is going back to latency. That microsecond latency we're talking about that's exactly where we fit in perfectly. The other part of it is how much of it can you do in parallel? And that's the IOPS discussion. And so when we look at what we do our three U form factor delivers a million IO's. That means there's enough parallel streams we can handle in addition to the latencies allowing people to process a lot more data. So if many customers are trying to do a whole bunch of real-time analytics in our environment primarily because there's real value to it and they're doing it without disrupting their environment. So how are customers getting value out of this? What's the business case? I mean Flash is more expensive than disk but some people are gobbling it up, you know? Yeah. Help us understand that and where the business case is specifically. So you have a great point Dave. The economics is a funny thing. If you look at traditional spinning disk environments yes it's a dollar per gig conversation but you look at where traditional Flash is deployed or Flash is deployed in traditional environments it's been the caching concept then it's been this expensive thing you bring in to solve a particular problem. The way we look at it is that it's an application issue. We have SLAs that customers are trying to achieve and meet and when you look at that environment and you say okay I've got a requirement to deploy a thousand desktops or desktops that have to have equivalent or better than real desktop environments in a virtualized infrastructure. That changes the game. Now you're looking at operational efficiencies in the back end and saying okay I want to give an end user a desktop like experience that is better or equivalent than the current desktop in a virtual environment. That means that IO's become critical and so you're moving the conversation from a dollar per gig to a dollar per IO environment. If you go on that route we're absolutely the most economical solution that's out there. The second part of the value is that if you look at what the value of Flash or legacy storage brings in or doesn't bring Flash changes the game from an application economics perspective. We can come in and reduce the compute resources that you need to run the same application. In fact in some cases we can reduce the actual application license requirements you have. So you have ROI that you can get all the way through from compute to application site that can be essentially calculated and added on to the OPEX value that we come in by reducing the legacy spinning disk infrastructure that you have to frontend the IO requirements. Yeah and start to deliver bottom line productivity. Alright Ashish we're out of time but thanks very much for coming. Violent memory systems one of the hottest startups in Silicon Valley right now and really dominating the Flash business. Still private, going public soon as everyone's been speculating and talking about. Don Basile, rock star CEO. We've been following his career for some time and congratulations on all your success and Flash is the future. It's the no brainer. It is part of the modern infrastructure. We're covering it. We love Flash. I'm John Furrier with Dave Vellante. We'll be right back after this short break. Thank you.