 extracting the signal from the noise. It's theCUBE, covering VMworld 2015. Brought to you by VMware and its ecosystem sponsors. And now your host, Dave Vellante. Welcome back to San Francisco, everybody. This is theCUBE. We're here for our sixth year at VMworld 2015. We're here in Moscone North this year. For the last five years we've been in Moscone South. If you're at the show, come see us. We're at the street level. We're going to talk about machine learning and big data and analytics, really injecting into the IT operations and infrastructure management. Jerry Melnick is here. He's the COO of a company called Sios. And Tom Matts is here with SanDisk. Gentlemen, welcome to theCUBE. Thank you for having us, Dave. Thanks, Jerry. We first met, we had a really interesting discussion at the Wikibon offices around how do you apply big data analytics to solve some of these nagging problems. I remember the first VMworld I ever came to. I came to the realization that, wow, it's really complicated to understand what's going on inside that system. Talk a little bit about Sios and how you solve that problem. Yeah, that's exactly the problem we're after. I mean, clearly everybody out there, this environment's evolved over the last 10 years. We've taken sort of monitoring tools, what I'd call first generation analytics tools. And what we're doing is really taking it the next step and trying to use and apply, and applying big data and data analytics, sort of the stuff that's been used in biological sciences. They're using financial services to really understand IT operations data and answer some of the tough questions that people haven't been able to answer before. So, Tom, Sandisk has had a really interesting evolution. I mean, everybody obviously knows you, you guys do the sticks, right? But there's so much more to your transformation, a lot of software components, really becoming a major enterprise player. Talk a little bit about your journey and how you got here. Well, I think if you look at Sandisk as a whole, yes, everybody knows us for the USB sticks or what goes in your camera, but it's really the same fundamental technology. And we are a flash company no matter what we do. And it's now moving into the data center in a big way. And in addition to the hardware that you probably are aware of, where we've got PCIe cards, Fusion IO, SAS and SATA SSDs, like you mentioned, we've got software now that makes it easier for customers to adopt flash in that data center. And that's where our flash soft server side caching software fits in. And that's really where you came into the company. So you've got a software heritage in it. So the culture is changing, right? It's becoming everything software defined now. So how are you guys working together to solve this problem, Jerry? Yeah, well, clear, this is a Sandisk host-based caching problem, this sort of perfect example of a use case for advanced analytics. And we've developed a new product, we just announced IOS IQ. And we're showing here and debuting here at the show. And that's an analytics platform that's broadly capturing data across all the dimensions around that infrastructure. And we're doing it in performance, efficiency, reliability areas as well as capacity. In the Sandisk case, and where we complement each other quite well, is how do we help customers who have storage issues identify the workloads that are really suffering and that can be managed by applying Sandisk technology and flash soft technology to implement host-based caches, to reduce latencies, improve throughput, and really get the value out of the system that they have. So flash soft uses the concept of flash as a host-based caching layer, is that right? Can you just elaborate and add a little call to that? Yeah, so the software takes any flash device in the server and uses that to store the hot data that we've identified with our algorithms to accelerate those applications. We are automatically moving that data from the backend storage into that local flash. So as an IT person, you don't have to, it's a hands-free thing. The real challenge with where Jerry steps in and helps out is that while customers look at our benchmarks and our marketing material and say, that's great, they oftentimes want to get a real sense of, hey, how's this going to work with my application? I'm different. My applications are running a little differently, and that's where SIO software can step in and show them kind of behind the scenes but looking at their exact application. How is this going to benefit from server-side caching? Okay, so Jerry, take us through how that works. So you're using all kinds of math and machine learning and other big data algorithms to identify what, where the hotspots are, where the bottlenecks are. Take us through, take a picture for us. So our product basically is a simple download and install software capability that you install and we acquire- No agents. No agents, completely, it's very simple. Actually, I should say simplicity was sort of a core design center that we chose. We said, we're going to do this big data analytics play but you got to have it delivered very consumably so we built a nice user interface that's touch and mobile. We acquire the data broadly across the infrastructure, the virtual machines, we look at CPU memory host performance, we look at all the IOPSA network and storage. We pull that all into a single repository across the infrastructure and then we do an analysis across performance to try to understand is this a storage problem, is it an application problem, is this a storage problem, how in the way, what can we do in the way of recommending improvements by using host-based caching to improve that storage capability? Okay, so you don't remediate, but you suggest where the problem is, how does the remediation occur? Well, this is the, as we talked earlier, the find it and fix it, right? So we find it and we fix it. So in addition to what I talked about with the kind of this initial customer concern of is this going to benefit me, what areas, which particular VMs do I need to accelerate? As you're operating, the environment changes, right? You can have VMs that may have a peak workload at a certain time of day and that's what Jerry's software can find. Our software can change configurations very easily, transparently behind the scenes and the customer, as they see things maybe getting a little bit, okay, this application, this VM's getting out of whack, they can make the changes in our FlashSoft software, in the caching, to fix that and optimize for the right configuration. So when you say make the changes, Tom, a lot of customers, earlier you were saying it's sort of automated and kind of hands off. A lot of customers like knobs to turn, just because they like to play with things. Then over time, they sort of become more comfortable and let the system take control. How does a typical use case work? So this would be a customer, let's say you've got somebody running a host with, I'll simplify it to a single host, but it's a host with 20 virtual machines and they know that 10 of those are where they've got a critical application need. They've had some of their users complaining. You can go in there and identify those 10 and say only those 10 VMs are the ones that can have access to this caching, this local Flash capability. The other 10 will continue to get access to data as they normally would through the backend storage array. So you can kind of start, you could start with as little as one, right? You knew there was one VM and then you can gradually progress and start adding VMs. We also have customers on the other end that just start with everything. We'll accelerate everything, then Jerry's software can tell them how to refine that. How do I get better at this? How do I optimize for what I'm doing? So what do you guys got going on at the show, Jerry? So you've got a booth here. We have a big booth with a touch screen, a 70 inch touch screen demo. We can show you sort of how we operate very quickly. With one touch, you literally can get to the answer of which virtual machines, which disks, which ones and how to configure host-based caching to best optimize your performance. That's only one use case in that whole piece and identifying and fixing a problem may or may not be related to storage performance. So if you're not doing a big data analysis that you're not really taking into account all the other factors. So for instance, maybe it's not a storage issue. Maybe it really is an application that's just out of control, a noisy neighbor or whatnot. So being able to identify sort of the root cause of a problem, is it storage? Will it benefit by host-based caching? Is it really CPU bound? Is really part of that whole problem. So here at the show, we're demonstrating all of that. So describe, I mean, you just announced CYOS IQ a couple months ago, but you started it with VMware. Thinking about the blame pie of performance management problems, what's your data tell you so far? Who's to most to blame? We talk about storage a lot. I would imagine storage is a big culprit, but how does it break down? Yeah, so obviously we're going straight at storage problems right away. So we're collecting a lot of our data there right now today, because 80% of the stuff you see out there, clearly everybody's recognized as storage contention, but applications are very problematic as well. And so understanding what's coming out of the application and how it's impacting the environment, how it's changed is really important. Then of course networking. Those are all introverted, right? So Tom, a lot of people would say, I don't get it, Flash, does it change everything? You just throw a Flash array at it, all Flash array and all of my performance problems go away. What's missing in that analysis? There's a couple angles there that you can look at, and that certainly is a potentially good solution for some customers. If you're talking about a Flash array, you are still potentially talking about network, network traffic being the potential bottleneck there. You're not local to the host and the application, so there's a little bit of latency, and there's an expense there. So what we're offering is kind of a solution that gets to maybe a little bit more of a middle ground, we kind of call it 80-20, where sort of for 20% of the price, you can get 80% of the benefit of going with all Flash. So given that it's just a storage focus on the problem areas, and this has been at least 30 to 50%, maybe even more, a storage show several years, and you have this sort of interesting dynamic where a storage company owns VMware, VMware historically, as it relates to storage, it said, okay, here's a bunch of APIs, ecosystem, go solve the problem. And now you're seeing VMware get more aggressive about things like vSAN, grabbing little pieces of the stack. So Tom, let's start with you. What's your perspective on as an ecosystem player, the white space, the opportunity, VMware's moves, where you guys fit, how you see that evolving from your strategic standpoint? So it's certainly an interesting time to be in this position. Just today, we had an announcement in conjunction with an announcement from VMware of a new API that they're announcing for their 6.0 product called VAIO, vSphere API for IO filtering. So they announced that this morning, and we announced our FlashSoft 4.0 product that will be implementing that particular interface. We were the design partner for them on that particular interface. So you have the SDK, you were first to get it. So we have been working for quite a while with them on this. No comment on that. No comment on that part. But we're pretty excited about that and the benefits that it's going to bring to server side caching moving forward through the 6.0 platform. So we see our relationship with VMware as being very, very close. EMC was another design partner for them on another aspect of the API. So I think they're doing a nice job of working with the ecosystem. So this is leveraging vSphere to a greater extent? Is that what this is all about? Yeah, it's really, it's giving us access through an interface to the capabilities that we need out of the hypervisor to be able to deliver better latency, kind of making the software stack a little bit more efficient for us, letting us deliver clustered solutions, letting us deliver right back capabilities efficiently. So there's a lot of ways that we can take advantage of these. So deeper integration, really. Exactly. So Jerry, from your standpoint, I'm presuming that right now anyway, you don't care. It's like, okay, great, let VMware get as aggressive as it wants in storage or not. We're solving a different problem. Is that fair or do you see sort of, at some point it becomes a strategic fork in the road for you guys? I mean, really, where everything's going in a software defined data center is actually sort of exactly the problems that we're looking at. You have to be able to acquire data. The more you do scale out, you have to be able to acquire data across that infrastructure, accumulate that and really identify where's the problem. Where is it? And as we scale out, that problem just gets larger because it's going to be harder and harder for people to find it. And if you don't apply advanced analytics, if you don't really apply that next generation of data mining, you're really not going to apply the intelligence that you really need to make these systems work. So when VMware first came out, I mean it was, remember the first VMware demo you ever saw, your jaw dropped? You went, wow, this is going to change everything. And it did. And obviously the ascendancy of VMware followed. Now they're a big company, right? Everybody's taking potshots at them. You got sort of OpenStack, you got Docker, you got Amazon coming out. So, and of course, VMware has to expand its TAM. It's a big company, it's got to keep growing. Public company, you know. And so, my question is, with our VMware, where do you see this whole thing going? You talk about scale out. What's your take on VMware? Obviously it's a big opportunity for you. You started there with CYOS IQ, huge customer base. The industry's best customers from an enterprise standpoint. Everybody runs VMware, but there's a crossroads here. What's your take on what's going on here? You know, clearly with some of the announcements today and the hybrid cloud approaches, this is all going to be a variety of resources. We saw that actually in our high availability offering, how people are moving off a physical, virtual into cloud and using a variety of all those. And so VMware clearly is going to be a place where they're doing and want to do exactly that. So for us, it's ideal. We are a platform, VMware certainly is our first target, but it's not exactly where we're going to stay. We have to acquire, in order to be effective and to really get out there, we're going to be acquiring data from every type of platform, whether it's an OpenStack platform, whether it's a Hyper-V platform, et cetera, or applications across those. So the problem is one of data and how to analyze it. At the top, you guys are an arms dealer. So you stole my analysis. I was going to question whether I should use that analogy, but yes. I think it's fair to say, you're really... Battle it away, everybody. Go nuts. As long as you need more storage and you got a performance issue. So talk through that plan a little bit. It's exactly what you said there, right? As all of this IO needs to be funneled through different aspects of the network, different parts of the data center. I mean, it's just going to become a flash world here, right? I mean, flash is going to be everywhere. There's going to be multiple levels of flash, right? So management of data, even in an all-flash world, they're still going to end up being, you know, you're going to need to have your flash that's the highest performance flash. Maybe it's on a dim slot, you know, slot, or maybe you've got a lower cost flash array that you may have as your back end. We see all these trends that are happening and what VMware is doing is nothing but positive. You know, we had an interesting conversation this morning with D Raj, the CEO of Nutanix, and he was almost, my takeaway was, he was saying, you know, the whole DevOps meme, we may have gone just too far. That infrastructure still really matters. It's not like just infrastructure is going to go away and even though their tagline is infrastructure, make it invisible. There's still a real complicated piece, but the roles are changing. You know, the traditional storage admin, you know, and the LUN management aspects are kind of, you know, perceived as less value. People are trying to get, you know, new training and new skill sets there. What do you see Jerry happening at the organizational level in the roles? That's a great question and it's really interesting because given what we're talking about today, we obviously cross those roles. So what we run into are very different kinds of organizations and you can very well identify those organizations who are thinking exactly about this problem and when we come in and talk about what we're doing, resonates very well because they've already taken and put an overlay over those silos that have been operating so many years in the IT organizations and they're thinking about that and then you have sort of, I guess you would call it, more old school environments where it strictly is a silo and it's very difficult, if not impossible, to find the right people to talk to yet. But you know, you know that that has to change in the world that we're going and we're headed. And Tom, are you seeing, I mean, I imagine you, how are the, just the affinity between the Flash and the application development people and how that's sort of changing the way in which they operate? Yeah, I think as we go over these next year or two, you're going to see even more integration between those two groups because Flash, as it becomes more pervasive throughout the environment, all of a sudden you're an application developer, you look at this Flash, you say, hey, if I can find a way to, you know, to write to that particular capability or I know that this Flash capability is going to be here, you know, there's more and more opportunities for optimization, taking advantage of what Flash can truly bring to the picture. All right, last question. We'd love to ask the bumper sticker question. So much has changed in the last five years. So Tom, I'll start with you. The bumper sticker on, you know, VMworld 2015, you know, if you get a good tagline, our guys might put you on Twitter. What's the bumper sticker? Trending topic. Yeah, you'll be trending, all right. VMworld 2015 from SanDisk perspective. Let's see, how about Honk if you're integrating more Flash this year or something like that? Yeah, that's good. Honk if you're integrating more Flash, I love it. All right, Jerry, how about from Syos? I would say, software defined everywhere. SDD, everything. All right gentlemen, thanks so much for coming to theCUBE, it's great to have you. Keep it right there, buddy, we'll be back with our next guest, this is theCUBE. We're live from VMworld 2015 in Moscone. We'll be right back.