 Okay, we're back. This is Dave Vellante. I'm here live at the Wikibon offices in Massachusetts. And this is our first ever simulcast. We've got a bi-coastal cube gig going on here. This is the Flash Cube. My colleague and co-host John Furrier is in the brand new Palo Alto studio, SiliconANGLE studio. This is the cube where we extract the signal from the noise. We're bringing you the best guests that we can find. Scott Dietzen is here. He is the CEO of Pure Storage, an up-and-coming hot all-flash array company. Scott, welcome back. Good to be back. David, John, very, very thankful to be on board. Yeah, always good to talk to you. I mean, there's a lot been going on since we last talked, whenever it was. I don't know, it was almost a year ago now. And you're seeing some of the big whales get into the marketplace. We saw EMC make some moves, big announcements, bold statements. IBM just announced that it was investing a billion dollars in flash. It rebranded the TMS acquisition. So what do you make of all this action? We think it feels great, right? I mean, what if the big guys aren't coming after the solution that you're putting forward in the marketplace, then something may well be wrong. So this competition is a great thing for those of us trying to grow our businesses because you need competition to drive growth. And most of all, it's a great thing for customers. Customers are going to have an array of different choices with very different parameters around them in the flash space. So they're gonna have to do some homework before making their purchases. But we couldn't be more excited about the competitive environment and how much fun it's gonna be in the next few years. Yeah, so talk about Pure's differentiation a little bit. Obviously you're going up against now much bigger companies, more well-funded, a lot of cash. How do you differentiate in the marketplace? So when Pure was founded, we felt the key hurdles to flash adoption in the enterprise were two. We're cost and compatibility with existing enterprise applications and workloads. The customers we work with don't want to rewrite their applications. They don't want to change their configurations. They want a flash substitute that can be just slotted in in place of a disk array and yet offer all of the advantages of flash, the performance, the power and space efficiency, and the dramatic simplicity advantages over disk. So we crafted this recipe four years ago now where we used off the shelf MLC flash and where we used very fast data reduction. So global inline deduplication and compression that works sub millisecond and yet shrinks the data that we store on disk so we could get the price point of all flash down to where it was competitive with disk. And that remains a unique differentiator for us in the marketplace, although clearly, with solutions like ExtremeIO from EMC and Flashray coming from NetApp, we expect to see company in the years ahead. Yeah, so you guys have been very aggressive about the data reduction strategy. It's critical and you're pushing that very hard. So what I'm hearing from you, Scott, is you feel as though you've got a lead in that regard relative to, certainly, ExtremeIO is not in the marketplace yet. IBM hasn't done, for instance, the data reduction yet, even though you expect that to come with store-wise. So all that's coming. So what's next for you guys? How are you going to stay ahead of that curve? Time is going to tell how steep these hurdles are. I mean, we've invested dramatically in software. I would say some of the companies that got into the flash space, especially those that got in earlier, had to go off and make hardware investments even assemble a form factor that was 100% flash. And we had the luxury of coming into the market when we could put together off-the-shelf hardware technology and pour all of our energy into software because we see that as the differentiator long-term. So the way we think you went in this marketplace is through cost. We talked about data reduction. One of the things we've done is actually put our data reduction aggregated across hundreds of systems up on our website so that everyone can see what the average data reduction is on a pure storage array and the difference that that can make. What happens next is continuing to grow and advance the solution. So we have features like dedupe aware snapshots available in the marketplace today. That's critical for most business critical deployments of storage. But we still are lacking things like replication for a cross-site disaster recovery. And we've got to build that technology out. There's ever a greater scale of the solution. We're doing 100 terabyte plus deployments today, but we need to get into the half petabyte range for the customer demand that we're seeing going forward. And then there's opportunity to support alternative access protocols and so on. We joke, our CTO jokes, we've got a five year roadmap to keep one of the most talented engineering teams busy that will hopefully allow us to stay ahead in a very competitive market. Yeah, I love the forthrightness, Scott. You're always that way with us on the Cube. Talk a little bit about this whole notion of software defined or what we call Wikibon software led. What makes you guys feel as though that's the future? Number one, and what makes you software defined or software led? Well, you know, the term at least with respect to storage, I do think it's still being sorted out. So I like the renaming you guys have proposed. When some people talk about the term, they talk about storage where you actually put the media into the server infrastructure, and then you sort things like all tolerance and being able to protect the data through mirroring and so on, all with a software stack that runs entirely inside of the server. And that mimics what some of the large data centers have done, what Google and Facebook have done with their architectures. The challenge to that model with a traditional enterprise deployment, as I mentioned earlier, is it doesn't provide native high availability, native non-disruptive upgrade, non-disruptive expansion, features like consistent snapshots and replication out of the box across data centers. These sorts of features are crucial to traditional businesses, and we think ultimately need to be part of that infrastructure. But ultimately the thing, I think it's gonna be hardest of all to pull off in this space is the really fast de-duplication and compression. The challenge with putting all of your storage into the server tier is no individual server sees that much of the data set, whereas if you have a shared storage footprint, you can achieve much higher levels of data reduction because you've got a larger volume of data to reduce. And that's actually how we've been able to get the cost of flash in our storage array down below the cost of flash deployed in the server tier. Well, still providing all of the control you would expect from a software-defined backplane, right? Being able to integrate with technologies like OpenStack so that customers can do all of the provisioning and management that they seek to do out of a software-defined infrastructure, following standard DevOps. But we're still a believer in the appliance model of storage as opposed to just loading up your servers with PCIe flashcards. Yeah, so one of the things that John Furrier and I talk about a lot is Hyperscale. And John, I wonder if you could talk about that a little bit and maybe we can ask Scott about the lessons learned specifically from Hyperscale and how their pure is applying that to the data center, almost in a Hyperscale for the enterprise. Scott, one of the things we talked about earlier on the program here on the FlashAhead FlashCube is scale-out open source, which obviously you are a big proponent of open source and open systems, you know, you've pioneered a lot of the, you know, what we call early, you know, generation, first generation web workloads, right? So the conversation tends to go scale-out open source, open source software on industry standard hardwares, focusing on workloads, essentially the applications. So the question I want to ask you is, obviously on the website, Google builds their own stuff, Facebook assembles their own infrastructure, you know, Apple does their own thing. So these web companies led the way. For an IT enterprise, it's just not that green field. So I'd like to get your perspective is, what's it gonna take for IT enterprise to cross the chasm to get, to start putting their toe in the water of Hyperscale? Specifically, it's just not that easy. I mean, it's pretty easy for Facebook to start from a clean sheet of paper with one application, Facebook. But for an enterprise, it's a little bit complex. You mentioned Ddupe, all these different things. So I'd like you to share your, you know, on one, what it takes for an IT enterprise to really start moving in that direction. Because they all want to kind of go there. They just kind of build that bridge. Could you comment on that? Oh, it's a pretty far reaching topic. So open source, there's a technology stack that continues to grow up that is ultimately challenging a bunch of server software proprietary incumbents. So, you know, I'm a big fan of the big data technology, you know, around Hadoop. And I'm involved in Cloudera as their outside director. And it's been really fun to watch that software change the way businesses approach, you know, analytic problems. So in fact, if you look across, you know, all of the traditional software in the server tier is being challenged by some next generation, typically open source technology or SaaS, where they're using the open source infrastructure to go off and empower it. I would say that there's still a lot of growing up that this stack has to do, however, right? I mean, customers want full elasticity so that they can, you know, add commodity hardware to be able to scale and, you know, right size, the deployment for an application. They want to be able to mix and match private and public cloud and have investment protected across those things. And there's really no, you know, simple easy rules that you can discern other than picking, you know, open source technologies that have received broad adoption. And, you know, and often times being used by these large public open, you know, the public consumer website because they've been proven out. That can give you a distinct advantage there. And a comment, examples of people migrating from IT over as any specific use cases we see pure storage playing a role there. I mean, honestly, there is an emphasis on build out for these applications. So there's investment. So the sweet spots that we've been targeting have been virtualized and database workloads. So three different use cases. Database, it's mostly analytics. So customers that would be running some workload is performance constraint. Typical customer might have an analytics batch job that takes 24 hours to complete and they're running up against, you know, being able to inform business decisions for, you know, next day setting pricing, for example. The solutions when based on disk and move to flash will, you know, run at least 10x faster. You know, we've had people get other workloads down to, you know, to an hour or less and actually consider switching the whole batch environment over to real time. So the, you know, the business is constantly being updated over the course of the business day or they can ask, you know, 10x more interesting questions. The next tier I would describe is, you know, people that are running heavy virtualized workloads where they're pressing the storage really hard. You know, products like Mware and KVM and other virtualized architecture are very random and very intensive which is really expensive on mechanical disk. Stress is the mechanical disk. We've seen by being able to introduce flash storage that customers see three to five fold server consolidation and that includes the enterprise software running on top of those servers whereby they actually save enough much by moving to flash that the storage product is free. And the last workload, you know, we've had great success with virtual desktop infrastructure or VDI. You know, with the emergence of tablets and smartphones there's been a push toward your own device to the workplace. Enterprise likes that model. Ultimately they want to serve their users but they hate giving up the reliability of having a backup of everybody's desktop image in the security of being able to manage that. What all flash storage can bring is all of the great user experience of VDI. In fact, users on top of a flash store like Pure find the user experience is better than a flash laptop with a local SSD even though you're going across the network and yet the enterprise doesn't sacrifice any of that manageability and security. Scott, I wanted to talk a little bit more about the software piece of it and particularly your data management, your storage management, your volume management stack. You mentioned some things like replication that you're trying to build out. It's not trivial as you mentioned. And also I love the deduplication focus. It worked really well for Data Domain and Frank Slutman. He used to pound that really hard and I think it's a simple, easy to understand message for customers but some of the other pieces of the software stack are much more nuanced. And if you look at the landscape within flash, you see a real mix of software capabilities. It's not easy to just, as I say, it's not trivial to build up a data management, volume management, a storage management stack. And you're seeing, for example, violin OEMs from Symantec, Extreme I.O., still working to get its stack together. Who knows, maybe it can borrow some IP from EMC. Same thing for IBM. Very currently, data management, volume management, storage management, light. Talk about where you're at, what you see as evolving in the industry, generally and specifically with Pure. So we mentioned earlier that this is a complicated environment for customers to make choices because if you look at the marketing literature from all of the vendors, there are a lot of dramatic claims. So for storage in particular, customers have to test the solution to be comfortable that they're getting what they're paying for. The good news about storage is you really can and should beat it up and watch what happens while you're doing so. So most of our engagements still start with a proof of concept where the customer puts some performance-tensive workload on the product and then actually observes the data reduction ratios that we're delivering and then can use that number to figure out exactly how much money we're saving them relative to disk. And then they'll try things like while the storage is under heavy load, they'll pull SSDs, they'll pull network cables, they'll unplug power supplies, they'll install new versions of the software, right? True enterprise technology should be able to accommodate all of those use cases non-disruptively. We think of non-disruptive everything in the storage here because everything upstream depends on that storage being available. And this is where customers will see substantial differences in the various solutions. The level of data reduction that's actually afforded in these production systems and then their ability to withstand all sorts of incremental failures and self-heal around those failures in the environment is something we would encourage every customer to try out. That being said, something else we launched, when people hear our story, the typical reaction is, flash at the price of disk with all of my quality of service features implemented as well, that's just too good to be true. And so we move to a model where we provide a, we call it the love your storage guarantee. If we don't deliver on the data reduction, the performance, the quality of service, the simplicity that we advertise through the course of our engagement with a customer, then we'll give them their money back. Have you ever had a deliver on that guarantee? Not yet, but it's such a competitive marketplace. You know, I'm sure we will have to deliver on that at some point in time, but when we look at it, I mean, we look at the customers that are delighted with our product after they get it. It's a no-brainer investment for us to offer that money back guarantee, given how rarely customers don't fall in love with the storage once they get their hands on it. Scott, can you talk about the competition? You mentioned it's very competitive. Obviously, you guys have a different advantage because you're coming up and you're essentially growing startup and heavily funded. Obviously, the business that you're in, you have to be. But the bigger players have to make adjustments. IBM's announcement was very clear. We talked about it earlier. It's a stake in the ground. Much more a rallying by internally for IBM to kind of rally around storage again. And two, it gives public confidence. And the rest is kind of like, I think it just is what it is. EMC went out and decided to buy Extreme I.O. Again, a big move. Some are saying that, you know, internally it wasn't working, so they go outside to get it. How long does it take? Can they just cobble together acquisitions? How much organic growth do you think they need to do? Obviously, modern infrastructure is changing. What is your opinion of the M&A of these big players and what do they need to do in it? EMC specifically Extreme I.O. That's a big acquisition. They've been in the farm on Extreme I.O. Well, so I shouldn't suggest I really understand EMC's strategy in full. But it feels to me like they are building the train track from both directions, right? They're building out the Extreme I.O. solution and investing in that. But they're also continuing to invest in VMAX and VNX, their legacy disk solutions and adding ever more flash to them. The challenge with those solutions is it's harder to go off and do the radically innovative things like full deduplication. Ultimately, we think they're gonna need to do compression too to be competitive with the data savings that we're providing. But EMC is a large organization. They'll be able to make those broad investments. Extreme I.O. is a technology that comes with a bunch of the foundational pieces addressed. I mean, most crucial things are being able to get that fast data reduction married to native implementation of snapshots. If you try to provide snapshots that's not well integrated with deduplication, then snapshots get very expensive from space and reporting perspective. Whereas when they're natively implemented and tied in with that data reduction technology, you can almost implement them with zero overhead. So we think Extreme I.O. has got a good start. In fact of the larger solutions, we would rate it the one that's closest to us in terms of being able to deliver all of that value integrated together. We hope we have a 12 month technology lead and we'd like to grow that lead over time. I think for many of the other vendors, there's a much longer software investment to make. We first shipped a technology that was doing this inline deduplication and compression in business critical deployments three years ago. And we've got a fantastic team. So the challenge will be, this market is developing rapidly. If you're still years away from having a competitive product that is able to deliver flash at the price of disk, it may be too little too late. So Scott, I had two last questions for you. So what sort of milestones, what data points should observers be looking for in terms of pure progress over the next six, nine, 12, 18 months? What are the things that you're really focused on? So I mean, the challenge in front of us is scaling our business as rapidly as we can. I mean, we've mentioned that we're out of the gates faster and growing faster than any company in storage history, at least anyone for which we have a public data. That puts a lot of challenges in front of a team. It's also just fantastic fun. We've got partners coming to us. In some territories, we actually have waiting lists of salespeople that wanna join the company. As soon as we have enough pipeline available, then we'll invite them in. So it's a very gratifying position to be in, but we've got to make great decisions about who we hire. We've got to make great decisions about the partners that we continue to work with. I mean, most of our business is partner-led and or fulfilled. So continuing to attract the best and brightest partners that can make the biggest difference for our business. That's why I sweat, right? The sweat in front of us, the product is working so well. This is all about execution and moving quickly to capture as much of the greater market as we can and make sure that we're showing the advantage of product to these competitors. Excellent. I was gonna ask you, what keeps you awake at night? What worries you? And I think you kind of just answered that, hiring the right people, partnering with the right people, getting ROI out of your partners and your ecosystems. Anything you'd add to that? I do lose some sleep over differentiating and it is a very noisy space. But I think the fact now that we got the big guys coming out of the product, Extreme I.O. and Flash Written from NetApp are both very similar to the product that we launched, the term of the recipe that's being articulated. So that ultimately, it's easier for us to say, here's people, right? This is what the market is converging on as the right solution. Frankly, it's exactly the recipe that we launched Pure Storage with four years ago. I mean, it doesn't get any better in tech to have the big guys come to your playing field and try to win against you then. So shame on us if we're not able to continue our technology lead and prove out the best at delivering this proposition in the industry. But it's gonna be a lot of competitive work to stay ahead of three talented companies. Yeah, excellent, Scott. Well, you're right. It's a lot better than being all alone in an echo chamber. Scott Dietzen, CEO of Pure Storage. Thanks very much for coming in to theCUBE today. Really appreciate it. My pleasure. Always glad to join you guys. All right, everybody. Keep it right there. We'll be right back with this Flash Flash Cube right after this commercial break. We've got more guests coming up. This is Dave Vellante. We're live from Wikibon and Palo Alto.