 Live from the Mendeley Bay Convention Center in Las Vegas. It's theCUBE, covering VMworld 2016. Brought to you by VMware and its ecosystem sponsors. Now, here are your hosts, John Furrier and Stu Miniman. Hello everyone, welcome back to the VMworld 2016. This is theCUBE, just looking at angles flagship programming, go out to the events and extract the symptoms of noise. I'm John Furrier, Mike, co-host Stu Miniman and our next guest is South Team Veghani who is the co-founder and CTO of Pernex Data. Now with Nutanus, welcome to theCUBE, congratulations. Thanks guys for having me. Glad to be here. You are the community guest voted in by the community as the top guest they wanted to hear from. All right. So let's do it. So let's talk about what do you think of the keynotes today, VSAN, where VMware is going and the industry. Do you think it's a, this multi-cloud thing has legs? I think so. In fact, many vendors have talked about choice, right? And so it was initially about the private cloud and choice within the private cloud. Then there's a whole bunch of vendors about the public cloud. And they are kind of sort of wedded to that theme about the public cloud. But it looks like now given the range of workloads that we are seeing in the enterprise, it seems like there is some legs to the theory at least as of now, that there could be a potential to have both kinds, both flavors first of all. And more importantly, a need for things to seamlessly move across, right? So I think some of my colleagues were talking about it yesterday, right? The next generation, VSAN, will be the VSAN that kind of liberates you from the underlying substrates. Yeah, so Jim, you know, you were at VMware for a number of years. You know, want to get your commentary on kind of storage and software storage. Has that been changing a lot? Obviously, you had what you were doing at Pernex. Nutanix is working on kind of not just HCI, but in many ways, you know, modernizing the stack itself, including some competition against VMware. So can you give us a little bit of a historical view as to how you've seen things change so fast over the last five years and where do you think it's going? The last five years, you know, to tell you guys frankly, you know, I didn't quite expect type of changes we've seen in the last five years. I'd be lying if I said, you know, everything was, you know, exactly according to plan. But, you know, kind of reminding back a few more years is true to your point. As you know, when we started at VMware, this was circa 2002, when we were still trying to convince the world that, you know, consolidating multiple operating systems on a single server is the way to go. And, you know, at that point, and again, quite frankly, just by accident, we realized that, you know, workload mobility is going to be a key thing. In the next generation data center. And so, you know, the real kind of inflection point, if you will, for storage, virtualized storage, if you will, was when we decided to do something to help VMotion. So it wasn't even about, you know, storage roadmap. At that point, it was about a VMotion roadmap. You know, how can we make VMotion possible? How can we make it such that the VM can move without moving its underlying storage because that would take days or weeks back in the day when we are still sitting on 100 megabit ethernet and so on. And so the first thing we realized is we shared storage was key. And so we ended up inventing a cluster of files of some VMFS. That was my pet project. And, you know, still one of my favorite things when I look back. And, you know, that kind of search of kind of a sequence of events, right? Just once we did VMFS, we fell into this whole sand bulge. And, you know, sands did a large to make it very optimal to run virtual machines. But, you know, at the end of the day, we realized that there were just a lot of players. You know, there was a switch administrator. There was a storage admin. There was a server admin. And so that, of course, led to, you know, hyperconvergence and the vSAN and the Nutanixes of the world. And then, of course, there was, in parallel, there was this other thread which was storage was becoming faster than we had ever imagined. And it still kind of keeps going. Probably even faster than the rate at which processors are evolving. And that made it very important for applications to run right next to storage or the other way around. And so, you know, server-side technologies were very important. Yeah, so Satyam, I mean, you created VMFS. I mean, there was, you know, transformational for VMware going forward. You know, we think back to the early days. I mean, I remember when I saw VMotion for the first time, it blew everybody's mind. It was magical, right? We've all talked about it. Still differentiation for VMware. A question for you, and you know, maybe it's a little personal, but is VMFS long in the tooth now? Does there need to be reinvention? You know, what do you think about kind of the VMware stack and where things need to go? I think so, to be honest, you know, it is showing a sign of its age. But, you know, I say this often. In fact, I was speaking at a Vima keynote a few months ago. And the first thing to realize about any technology is that it's going to get disrupted, you know. The worst thing you could do after inventing a technology is getting too focused on that one thing and assuming that that's going to last a decade or two decades. You know, it just doesn't happen, especially in today's world. And so, yes, it is long in the tooth, but you know, back in the day, when it used to be highly relevant, it did a fabulous job, you know. Many people talk, have great stories about how shed storage and VMotion absolutely changed how they run their IT infrastructure. And now it's time to move on, you know. Now it's time to actually focus on storage class memory, server-side media, because, again, devices are just getting too fast for the networks to be a bottleneck. And that's life, you know, nothing surprising. If you ask me. As Stu says, it's always shifting the bottleneck somewhere else as no networking parlance there. But that brings up the point around the data. Right now, to us, at least we were just commenting on the opening about the data is the competitive advantage opportunity. And being open and not proprietary doesn't exist anymore. You use open source, I mean. But using open source doesn't mean you're open. So there's now a nuance around, not proprietary, but lock in spec with being sticky with the data. So we're trying to understand, love to get your thoughts on how you see that because that will truly define multiple clouds, because if Microsoft comes out and says, hey, we're going to create some stickiness with the data, or we're open, technically you could move your data, but moving across the network again is another problem. Yeah, that's true. What needs to be in place in your mind to truly have open cloud for the data and the data area? I guess that's a five-year question, but I'll take a swing at it. That's a great question. Should the arrow forward? I mean, we're speculating, so we're riffing, but go ahead. Also, the first thing about data is, at least if you look back on the past few years, we, at least in the storage ecosystem, have focused a lot on data services around a blob that is otherwise impenetrable, right? So you have a virtual disk or blob storage. You don't quite know what the application is doing inside the blob, but you still try to guarantee some quality of service on that blob, whether it's IOPS, latency, or some amount of replication or snapshotting. But I think now we are all realizing that we need to peek inside that blob. It's all about the applications, but to understand what the application is doing, you need to have that level of transparency. So I think in my very, not so formed opinion yet, I'm also kind of sort of seeing how it evolves. But in my opinion so far, we are going to first kind of go into this phase where a lot of vendors, including possibly Nutanix, is going to try to break into, through that, a weakness around users' data, and actually help people understand what makes up an application, and then try to do some interesting management capabilities around different components, different data components that make up an application. And just to give you a very simple example, a database is not just a blob, right? A database is some table space, there's some lock space, there's some temp space, so, and different types of data types require different service levels, and different types of management, and different life cycle management. So I think we are going to see a lot of vendors who don't quite have the keys to those applications try to break in. So that's, I think, going to be the first phase. But then, I think in the second phase, we are going to see some standardization, right? The infrastructure vendors will work with the app vendors to somehow standardize the application, the infrastructure interface, so that we can do many more kind of application-oriented services, but inside the infrastructure. Yeah, so you bring up, I think, a key point, I want a lot of our audience to understand about what you've been doing at Pernix Data, because there's some people look at, like, you know, not to trivialize it, but, oh, they help performance, and especially with Flash, and I can take something older and make it run a little bit better, but it's the analytics. And we understand the storage industry, we've been talking for many years now, it's not just about storing, but how do I leverage data? As John was saying, you see Nimble Storage, what they're doing, the analytics, a lot of storage and networking companies are really trying to get in there, leverage that, and pull out of it. So can you talk to us? I know you probably can't share, you know, what Nutanix plus Pernix will be, but, you know, what the analytics is, where you're looking at taking that, and at the kind of core to what you've been doing the last couple of years. Well, for sure, and by the way, thanks for recognizing that, right? Because in fact, analytics is a new thing, and I think, you know, when faced with a new thing, everybody tries to jump on the bandwagon, but there are some key requirements to make analytics successful. But anyway, before I go there, you know, this is something that also excites me about the Nutanix Pernix data manage. And, you know, maybe I'll try to paint you a picture, given that we guys go a long way back, is, you know, we, both the companies, I see so many synergies, you know, back in the day, both the companies realized that server side was the place to be, just by virtue of how media was evolving, and then by virtue of the fact that, you know, data centers were becoming this huge conglomeration of servers, and the boundaries between servers were vanishing. And so we made a distributed platform that gave you a scale-out performance. Nutanix did so as well. The only difference was they brought in capacity, we didn't, and there were good reasons to do it, either way. But then both companies realized that a product doesn't make a company. And so we realized that, you know, the next level of value edge, you know, storage is going to be storage, you know, there's only so many ways you can do RAID and snapshotting and so on. So the next level of value edge is indeed analytics, you know, to make people actually to help them understand what's going on in their data center. That was a key behind architect, which was a Pernix data product. That's also the key behind Prism, which is a Nutanix product. And so just to kind of give you a little bit of an idea about what the future holds. You know, I think we have a lot of work that we can bring in all the work that we did around Pernix data architect. We can bring that into the Prism platform. But then there's also the other angle, which is, you know, Prism as a platform so far helps the hyper-converge part of the data center. And so the other question is, hey, can we actually help a much broader part of the data center, the next generation data center, and that broader part of the data center spans legacy infrastructure. It also spans some amount of cloud, right? So then talking about analytics itself, you know, I'll just say one thing, which is, you know, right now we are in this phase where we try to collect a lot of data, but you know, right now the server side seems to be at a huge advantage when it comes to analytics. Because if you are there, just like you saw in the VMware Keynot, you know, there's the network virtualization assets, there's the storage virtualization assets, and then there's of course the compute assets. Other people like Nutanix have it too. And so having all those assets lets you get a holistic view of the data center and feed that into your analytics system as opposed to a storage-only view or a compute-only view or a network-only view. And I think that's going to be a key differentiator, just kind of starting out of the gauge. I think one of the things that excites me about this conversation, one is we're kind of painting a picture about what the future might hold and certainly what you guys have been doing with Nutanix kind of teases out future. But the reality is on the app side, app developer side, this whole data ecosystem, whether it's Hadoop or databases in general, have always been kind of a systems of record model. And now with real-time data, you don't know which data you might need from a different database. So you're seeing the mega trend being that data from every database will need to at least be exposed to some level intelligence. You know, this has really not been worked on. You guys are really cracking the code on this. So it's at the beginning stages of this opening up of metadata, about the metadata, if you will. How early are we? I mean, we're kind of like pre-gaming this thing. I mean, what's your thoughts? I mean, because that's where everyone struggles at these analytic shows with the power of the data. I think, well, maybe I'll explain it through an example is back in the day, I think it was 2005 or 2004, we had VMFS and somebody out in the open, reverse engineered the VMFS on this format and then they made a Java application and it was a VMFS file system driver. It was read only, because they couldn't figure out transactions, but you could at least read data. And so I think that's how early we are, is there's too many applications to worry about, at least in the enterprise. And so we are probably in the phase where we are not going to see too much standardization because we don't even quite understand the scope of the problem, to standardize, right? And so the first thing we'll do is we'll hack. So that's how early we are. That's the creative phase, the hacking. It's true, we're hacking around. It's hard to standardize but you don't understand. So what's the hardest part of this hacking phase right now that you see? Or what's the exciting area? Hard, exciting, it's kind of the same thing and from a hacker standpoint. The hardest part will be, well, I'll answer in the technology sense in the business sense, right? And the business sense, the hardest part will be like it or not, we are all frenemies, right? So when we start adding value beyond a point, people stop being friends. And so we are in this phase where we really need to be friends to understand each other's applications. The infrastructure guy needs to understand the application. The app guy needs to understand infrastructure. You might say, why? Because the app guy might want to orchestrate infrastructure changes, right? To make the app perform better. And so we really need to be in that friend phase but unfortunately as we add more and more value, that's going to break down. So I think that's the kind of key problem I see on the business side. And on the technology side, I see you yourself, the examples you gave, right? How do MongoDB, all those things are moving so fast. We thought Apache Spark was a thing and suddenly now there's this new thing called Flink. And so it's moving so fast. It's like nightclubs. I mean, once trending one day, it closes down and a new one opens up. I mean, but this is open source communities. They move with the trend. So it's fashionable, if you will. To never fight fashion, as they say. Yeah, exactly. But because it's moving so fast, just by the time you think you have figured out the whole life cycle, you have figured out the marriage, the orchestration, the party somewhere else to use your analogy. Yeah, I mean, you bring up a really good point. If you look at a lot of those modern applications, it has different requirements from kind of the storage side and might even have challenges with virtualization. How do you take that into account? How do you think of kind of some of those new applications that might not fit into kind of our traditional buckets? Actually, that's a great question. So, you know, again, honestly speaking, the canvas is yet to be painted. I think we'd be very, very wrong in assuming that everything we've invented over the last 10 years or in the last five years, for that matter, will just, you know, we slap it on and it's just going to work. And in fact, that's the exciting part, right? Is, you know, there is so much potential for innovation here, as long as we have an open mind, as long as we don't kind of just get too married. And so, again, to answer your question slightly more directly, I think, and through an example, for that matter, storage-class memory, right, you know? There is a school of thought which says that, you know, the right way to use storage-class memory is to give it directly to the application. And that's the only way, because it is so damn fast that anything in the middle is going to, you know, be pure overhead. But then there's this other school of thought, which is more infrastructure-centric, that, hey, you know, there are some infrastructure services which are better off done once and perfected once, especially when it comes to data, right? You know, making versions of data, sending it off to across a wide-area network and so on and so forth. You know, all those services are so difficult to build that you don't want to build it a thousand times for a thousand different applications. So I think we are going to see this tension between the two schools of thought. I think, you know, I'm assuming that we'll eventually settle somewhere in the middle. But then the real question is, who is going to get there fastest, right? And again, not to pitch this, but that's the other thing that excited us about the Nutanix-Pernix data manage, right? Is, yeah, we are on the server side. We see applications. We see this ridiculously high-speed media types evolve and maybe we can do something about it. And it could, you know, be an addition to the stack that we already have or it could actually be a disruption to the stack. Satya, final question for you, what we wrap up is, your take on the ecosystem, obviously the VMware has always had a robust ecosystem. Then there's been kind of, when there's always these shifts are happening, the ecosystem will move, won't be long in the tooth as well. We'll have to shift and get disrupted again. But now we're seeing an ecosystem 2.0 going on. So you have a lot of partners with VMware. They have a partnering strategy. How does the ecosystem evolve in your mind, your vision of where the ecosystem might go to, how it might evolve, obviously cloud will be one. Do you see any major shifts in the ecosystem? I do. And some of it is worrisome, especially, you know, because I love this ecosystem. I've spent so much time at VMware. Worrisome in a good way is, you know, the disruptive evolution would be the ecosystem pretty much just, you know, wholesale migrates to maybe potentially a new queen bee, so to speak. And the queen bee could potentially be the Amazon's of the world. But that's also the opportunity that companies like VMware have is, you know, to show us the path, you know, all of us as VMware partners. You know, if we can cooperate enough instead of, you know, facing too much friction between ourselves, then maybe we can all together move into where the kind of puck is at, right? Which is probably the public cloud or the marriage of the private and public cloud. I hope to have Raghu on theCUBE. I saw him last night at one of the press analysts event. And we were talking about that particular point. Mainly also the data aspect of things and becoming less friction. So yeah, we'll keep an eye on it. Satyam, thanks so much for coming on theCUBE. Appreciate the commentary. Great conversation. And congratulations on your sales at Nutana. Great partnership. It looks like it's got a lot of great synergies there. Looking forward to hearing more. This is theCUBE here at VMworld 2016. I'm John Furrier with Stu Miniman. We'll be right back with more after this short break.