 Live from San Jose, California, it's theCUBE, happy adaptive flash launch brought to you by Nimble Storage. Now here are your hosts, John Furrier and Stu Miniman. Hey, welcome back everyone. This is theCUBE Live in Silicon Valley in San Jose. This is theCUBE where we go out and extract a signal from the noise. We're here for the exclusive launch for Nimble Storage. I'm John Furrier with Stu Miniman. Our next guest is Shuresh Vasudevan, CEO of Nimble Storage. Welcome back to theCUBE. Good to see you. Thank you very much. It's a real pleasure to be here. An exclusive product launch, really intimate but really powerful presentation. You're up on stage. You want to get into the details quickly but I want to talk about the culture of your company because your success story, you're growing fast, you've got great numbers, great product technology, which we'll talk about, but you're building a durable company. So talk about the culture of Nimble Storage because this is really the Silicon Valley story. Indeed, very much so. I think when I describe what Nimble is trying to do, I always describe it as one side of the coin is about the business that we're building and the other side of the coin is about the organization we're building. And so there are many attributes to that organization. First and foremost is the set of values that are important. We've got a great place to focus on that. Some of the precepts that we've adopted are things like we would rather not hire a superstar. That person does not know what to be collaborative with everything else. For us, having no jurisdiction is part of that. And communication is really important. And it's a durable company and you have a marketplace that values you. It's very, it iterates very quickly. It's about the performance. Unlike the private market right now where there's a lot of valuation, fluctuation, there's not a lot of market trading going on. Obviously it's a private market, but the valuations are massive huge. Talk about that and does that confuse customers? I have to say, first off, the variety of choices that a customer has is never been created, whether it's storage, whether it's security. But in storage specifically, there are 40 to 50 startups in just the last three or four years, more than I remember at any point in time in my two decade long career in storage. So that's the first thing. So it is certainly confusing. The one thing that I would also say is we're going through a phase where increasingly of the 50, there'll be five surviving. I don't believe there'll be 50 successful storage startups. What was confusing as well is over the last year and a half, the market was valuing growth over everything else. And so you started to see a phenomenon of companies investing in a manner where it was growth at all costs. I think slowly the markets are coming back to what you need is a disciplined business model where you're delivering hyper growth, combined with improving operating leverage, combined with a sustainable business model, which we've always outdeared. And so to some degree, I think we're seeing benefits of that. Let's talk about where you guys are right now, the product launch here today. You guys are classically skated. You are where the puck is right now. You've got a great market entry, the mainstream adoption of flash, where the pressure for performance and agile and adaptive capability is really critical. Now you have a bigger opportunity. So let's talk about what you launched here today in Silicon Valley. Indeed. So specifically, the announcement itself had two components to what we launched, right? So if I step back from that, let me start with that architecture. Architecture was always informed by the belief that performance and capacity are both important, and it depends very much on the workflow you're addressing. The customer that has pointed to workflows will want some applications to have absolutely flash-like agencies all the time. Other applications to have this cost structure all the time, and many in the middle that require a blend of performance and capacity. So the architecture we built was about saying depending on the workflow, you should flexibly deliver any of those attributes. What we announced today was a high-end version of our platform that needs a small, tiny appliance with 12 touch drives, 4 solid-scale drives, 4 flash drives. So you can deliver up to 125,000 IOPS. When you take 4 of those systems and combine them with a scale-up cluster, you get half a million IOPS. That is, performance rivaling that of mainstream flash arrays in the marketplace today. At the other extreme, though, you can also take those same 4 systems and configure them to be a capacity that exceeds the pedabyte like extremely cost-effective price points. So it is the broadest, the breadth of the platform in being able to address any workload is really what we brought to bear today. Sir, first of all, congratulations on the launch. Your tour here is really excited. It's nice to see 4 locations across the U.S. And I wonder if we can unpack a little bit for our watchers. Some of the architectural decision, because obviously, not every product is kind of fit in every environment. The hybrid of flash arrays is what we call kind of a fat middle of the market. And you guys are pushing down with capacity. Up with performance. Question number one is, we're moving right down the road. Where do you guys fit? So you're going to get all the flash arrays? You're not at all moving. You're going to get all the flash arrays. So why don't you need to make all the flash arrays? That's a great question. So we're going to take a couple of minutes So let's go here and find the notion of all the flash arrays. I just want to know what are the two aspects to what a personal watcher may find good. Why not? It is. The one who knows what is right is the one who knows what is good. I don't know. I don't know how many hundred or thousand of my apps are there. I don't know how many systems do you want me to speak to. I just want to make sure that the link is in the description of my app. So let's see what we need. Now, first of all, we have four units to our buffer to deliver the footage from everybody else. The first one that we're going to show you is in delivering random right commands. So the way that we deliver random right commands is very different from everybody else. In this case, it indicates in the front of the screen that the random rights are coming from your system. They're going to use flash to absorb those random rights. And then we're going to, in a pure system, they migrate the data map. In a fashion, it's a single piece of flash. That's one place where we're very different. Instead of using flash to absorb them coming in, what we do is translate random IO into what looks like a flash drive. So the first thing that we're really going to do is deliver one-fifth-thousand random right ios for random right ios. We'll let this arrive, which is more than the right ios performance of flash drives. That should be good. So let's get back into this. We've got a very low-level developer of low-level, say, high-right IO performance using this system. What we now need to worry about is how do we then deliver low-level, mid-lake, and ADM? Because all our rights are now solved with this, we can afford to pick them up. That's a big catch. If you want a percent of your ios, that's a very low-level system that's completely different from the one in flash. But if you have an application where you want all of your ios to come up with your ios, then you can figure out a much larger amount of features. So if I step back, in a nutshell, by delivering high-right IO performance, we make this drive that exceeds that of a flash drive. By having a very good flash that can deliver any experience you want, all the time a flash experience or a good hybrid experience, we address the same needs as a system that has no flash drive. So I guess if I reflect back to you, I think your marketing message is to get the performance of an all-flash drive with the price of a hybrid. So in the future, I mean, could mass-larking be an all-flash drive? Indeed. So I think the way you think about that is there is one fundamental must-have for us to perform a great big flash of persistence here, rather than thinking about it at full distance. The must-have is will the price of flash come down roughly to about three to four X-series of identity drives without compromising the difference? That's the key sort of ambition here. What we've seen in the past is price of flash coming down mostly by trading off endurance, if you will, of right cycles for cost. We can get the MLC-like endurance levels that we're used to today, and a price that drops from the 15 to 20 X premium today to about 3 X premium, then our architecture castle has been designed to say, we don't, we have underneath the single file system umbrella, we can deal with both flash layout and disk layout, so we can evolve to a flash only layout, if you will. Okay. The other thing I wanted to ask about storage efficiency, I mean, we've seen a huge explosion in being able to get just, you know, more data on whatever media that it's on. You guys have thin provisioning, you've got compression. Can you talk a little bit about Ddupe? Indeed. Why is it a compression architecture, you know, really rather than a Ddupe which so many have? Of course. So there are really two inputs into why we think about compression and thin provisioning as adequate and why we've not gone down the deduplication path. The first is to say the biggest benefits from deduplication tend to accrue in three environments, object storage, backups, and server farms where you're cloning the boot images over and over again, right? And our perspective has been we don't address object storage. With thin snapshots, we are solving the same problem that deduplicated backups do. And in the case of server farms, typically you can use zero-copy cloning to deliver the big benefits that dedupe does. In most other environments, dedupe tends to deliver between a 20% to 35% benefit, and so it's not a substantial enough benefit to go after dedupe, with specifically because our system is able to deliver very high performance using low cost drives. So that was the underpinning which is given that we are already using low cost drives, we would have to spend a lot on memory to maintain the dedupe hash tables, and the savings are going to be 20 to 30% so they don't actually cost justify. Now having said that, as you start to evolve towards systems that have very high flash content, deduplication becomes a necessary aspect of the architecture. So for us, it's not a religious argument around is dedupe important or not important. It's a question of if we're going to spend more on memory to facilitate deduplication, more on CPU, at what point does the underlying media cost reduction justify that extra expense? We came from an environment where we understand dedupe really well. Yeah, so if you're looking at just cost and storage in the market, I wonder if we can touch on cloud because you guys are pulling the back-up piece in-house. So you're seeing a lot of even the big storage guys who are starting to use the AWS as a tier or back-up to a secondary site. We talked to Rod earlier and talked about InfoSight, how you are a consumer of cloud, but how does cloud storage fit into it? Yeah, that's a really good comment. Two observations. The first observation I would make is today, cloud service providers are a key segment driving our growth, but when we say cloud service providers, these are either SaaS companies or infrastructure hosting companies that are essentially saying, here's a virtual machine for rent, come run your application. So that's a segment that's growing very quickly for us. In that sense, it's actually driving our growth. There's a segment of the cloud service provider market, companies like AWS, companies like Google, companies like Azure, that are not necessarily deploying our infrastructure and they're taking away from data that a customer might otherwise place on a storage system. Now, typically the way we think about it is most of the data that's going on to public clouds is what I describe as eventually consistent data, so archives, back-ups and so on. It's not necessarily appropriate for transactional data. In transactional data environments, you still need an engineered storage system like ours. So you nailed it by saying, over time should back-ups live in a public cloud? We think so, and is that a direction that at the right time we would pursue? Absolutely. That makes sense for us to consider as the way we would solve the back-up challenge. We'd still want to leverage the way we approach snapshots, but perhaps think of the cloud as an infrastructure that enables us to use it as well. Sir, I'll talk about the excitement. You guys are on stage with a lot of swagger here, confidence. Good swagger, I don't mean that in a negative way. But for customers, how do you boil it down for the customers? How do you take this excitement, share with them kind of why? What is going on? What's so important right now from the customer standpoint? Yeah, absolutely. The first comment I would make is that the one attribute of the company that I hope we never come across as demonstrating is arrogance. Confidence is good. Arrogance is where we draw the line. So hopefully that did not come through in the way that we presented ourselves. No, it did not. From a customer's perspective, I have never believed more that storage architectures have to change because if you're a customer with a data center where you have lots and lots of data, moving to a flash optimized platform is radically going to change the economics of your storage, whether it's the cost of delivering capacity, whether it's the cost of delivering performance, or the ease of data management. And so I would say if you're not contemplating what options are available to you, then you're arguably sacrificing some degree of efficiency. And it's a matter of time before that architectural shift will occur. We happen to think we've built the broadest flash platform and that's what's driving the rapid customer acquisition, but that to a customer I think evaluating your options has never been more necessary than now. So Suresh, so speaking of that, we just came off what IDC called the worst quarter for storage that they've seen in many, many years. You guys have an embryo position with better margins than the competition. So can you speak to kind of storage broadly? Is storage just in a slow decline? Can you maintain that growth and margins as you grow bigger? How does that affect you and the channel and storage as a whole? Absolutely. I think there are really three factors that are going on that speak to a decline in overall storage revenues. The least important of which is often called out by most of the mainstream companies on conference calls, which is that customers are sweating as it's more extending depreciation cycles. That's happening, but I think that's the least significant reason of the reasons why storage is slowing for the larger vendors in particular. The two main reasons for years we've been used to over provisioning storage to deliver performance that applications need. Flash has come around now as an answer to that problem where you no longer have to over provision the amount of storage to deliver the performance that application needs. So by definition I think we're going to see a period of shrinkage where you're not going to have to spend the same on storage to meet performance requirements. And like most things in storage, when thin provisioning came it slowed down storage capacity growth and then started to go up again. So I think the first factor that's affecting the likes of EMC, NetApp and Dell is reduced over provisioning for performance. The second factor that's again affecting some of the vendors more is the fact that eventually consistent customer data, archives, large scale content repositories, a few years ago would live on very large NAS systems. That was the best place to place those archives. Today increasingly object storage systems, whether they are on-premise or in Amazon S3-like solutions if you will, are taking away from data that would otherwise gone to Isilon or NetApp or others. And so that's also causing a shrinkage in revenue. I think those are factors more important than just macroeconomic slowdowns if you will. So Resh, thanks for coming on the queue. Really appreciate it. I know you're super busy. But to end the segment I want you to share in your own words to the folks watching in the technology industry right now. Why is it so important right now, this inflection point and has it compared to others that you've seen and lived through in terms of Cycles? Honestly, it's a little bubbly on the private side. We talked about that earlier. But why is this so it's an exciting time? Yeah, you know, I think if I think about the catalysts of change that go beyond storage, I have never seen a time where I believe databases, rules of building databases are being rewritten, rules of how you solve security problems are being rewritten, rules of how you address storage are being rewritten partly because there are some fundamental enabling catalysts that have never existed before. That is a catalyst that in many ways was really the enterprise manifestation of the internet era. The second big enabler is semiconductor-based persistent storage, flash, or it's follow-ons, right? And so that's the second big enabler that never has existed before. Analytics, data analytics at a scale that's unparalleled. So all of these, when you think about the underlying building blocks, it basically allows you to reimagine how you might solve a storage problem. Reimagine how you might solve a security problem. I think that's partly why enterprises are absolutely unbelievable change in the data center right now. Final word, every company has a culture. Intel's Moore's Law, some people have sales cultures. What is the culture of your company? Yeah, I think there are two words that encapsulate really sort of what the culture of the company is. Efficiency and collaboration. And so they're not sexy words, but let me describe each one. Efficiency, we are all about saying, can you be innovative in using every resource you have, whether it's in our business process, whether it's in our architecture, to the utmost potential? That's really sort of one core theme that drives us. Collaboration is all about saying, we want to build an environment and a community where people as teams deliver far more than any individual is capable of. So it's about sort of hiring no jokes, but retaining talent that's really able to work cross-functionally extremely well. Efficiency and collaboration, you guys are on a great run, and the public market's keeping you honest. And the orders are flying in as you say. I don't want to get you in trouble with the... Well, it's a fun time, it's a great time. Okay, sure. So CEO of Nimble Storage here, exclusive coverage is at SiliconANGLE, Wikibon's theCUBE. We'll be right back, extracting the silver from the noise right after this short break. Thank you.