 Live from London, England, it's theCUBE, covering .NEXT Conference Europe 2018. Brought to you by Nutanix. Welcome back to London, England. You're watching two days of wall-to-wall coverage from Nutanix.NEXT 2018 Europe. I'm Stu Miniman, my co-host is Ute Piscar. And happy to welcome to the program Tim Isaacs, who's the general manager of Data Services, which really is the core products at Nutanix underneath the hypervisor, if I understand correct. Correct, right. With Nutanix. Tim, thanks so much for joining us. Yeah, absolutely. Thank you for having me. All right, so Tim, this is my seventh .NEXT. And for a lot of them, it's like, okay, where are we with kind of that HCI marketplace? A couple years ago, Nutanix expanded the marketing to Enterprise Cloud and now has, and I got right from two years ago, it was about two products. And today it's more like 14 products. Some organically, some through M&A. But as it was put out, there's Core, which is AOS. It's the HCI, HV, Prism Management, and your stuff is what all of these customers that are attending here are using. I'd love to get a number sometime as to how many Nutanix customers aren't running that in the future. We were hearing how many are doing more than that. But give us kind of the state of your business. Sure, sure. I'd love to get that the, you know, no pun intended core of what Nutanix does. Absolutely. Yeah, so, I mean, I think you were sort of referring to our most recent segmentation, you're talking about Core, then Core is basically AOS, which is synonymous with the HCI. And then obviously that includes AHV built in and then Prism for Management. And then Essentials is several other things relating to operations management, automation, file storage, so on and so forth. So I'll talk about Core and I'll talk about maybe a few things in the Essentials bucket. So Core obviously is all of our customers today, right? It's a layered cake, clearly. So people start with Core, then they move up the stack, if you will, right into Essentials and then many into Enterprise as well. And Core, at its base, is software-defined storage technologies, right? Powering HCI. So what we realized quite early on is, look, you know, HCI is all about virtualization. So virtualized workloads. So you've got virtual machines, they can be desktops, they can be servers, they can be databases. But then there's also the notion of unstructured data, right? So what about all this file storage that I have? What about all this object storage that I have? And what we realized was, well, we have the platform, we have the infrastructure, software-defined storage, and it was simply a matter of expanding that. And if you think about files, it's just another use case on this infrastructure. So we started on our files journey about two years ago and I think you might have seen some announcements today as well as about six months ago where we're getting ready to release our object storage solution. So now if I take stock of the portfolio, what do you got? Buckets, I believe it's called. Buckets, it's called buckets, exactly. What do you think of the name? I don't hate it. Okay, okay. As an analyst, that's probably the most you're going to get out of me. Gotcha, gotcha. Yeah, so if I take stock of the portfolio today, you got Core, right? Which is the Piper Convergence, virtual machines, multiple workloads, and then you got unstructured data via files as well as soon via objects. And then we also provide just generic block storage for anybody who wants to, hey, I got a database, it's running on a bare metal server. Can you give me block storage? I want to consume it and I'm going to run Oracle on it in a rack setting. Yeah, sure, go for it, right? So even though most of our customers are indeed hyper-converged, there are some customers who use us as storage only. And that's okay. If that works for them, great. But the power of the whole thing is now you can consolidate all of your workloads on a single platform. Yeah, one of the things we talked about with Deerage yesterday is when Nutanix launched, there were certain waves that it kind of hit. It seemed to be the right time for things. So, you know, software-defined, before we called it software-defined, what was there, flash as a technology was really coming from a little niche product to broad adoption. A lot has changed in the, you know, about nine years since the solution went on and you had a major file system rewrite in 5.10, I've heard some people call it, think of it almost as AOS 2.0, to get ready for some of the modern things happening from a technology standpoint, as well as the modern applications that will sit on top of it. Deerage has said it's like, you know, the plane flying at 35 miles, 35,000 feet, you know, running at full speed and we're going to change the engine out. So, it gives a little bit of insight as to, you know, what goes into that, what that took and what that prepares Nutanix and your customers for. Absolutely. So, you know, there comes a time in every technology's lifespan where you have to re-architect significantly. And that's because of, you know, things are changing, right, applications are changing, the world is changing, there are a bunch of emerging technologies that come about. And we are at sort of an interesting time where things like memory class storage, NVME, RDMA, all of these things are starting to get mainstream and for good reason, right? They actually deliver a lot of value. So, the file system that developed nine years ago, yes, you can make incremental changes, but there comes a time where you have to say, look, I got to make these big changes. I have to rethink my data metadata structure. And that's what you're seeing. And obviously this will be in phases. Phase one was more about optimizing metadata. Phase two will be about rewriting the file system in a major way. What we're calling block store to basically take advantage of things like memory class storage. And the end result is two things. One is, we'll be able to take better advantage of all of these new technologies. And by doing so, now you're delivering a very different kind of a, if you will, not just an experience, but value to your customers. So somebody could be running a database today and there's certain expectations of performance and reliability and latency. In this new world, AOS 2.0, those expectations will be entirely different, right? So, looking at the adoption of it, right? So AOS 2.0 basically, everyone's going to run it at some point. Everybody is running it already, upgrading it is going to be, in Nutanix style, pretty easy. But I'm wondering, the other storage products, what's the adoption there? How many people are using it? What are they using it for? Sure, sure. And by the other storage products you're referring to file storage and things like that, right? So, files, of all of them, files is the most mature. We released this as G about two years ago and we have close to 1,000 customers using files today. So not just purchased, but using. So 1,000 users as customers, pretty decent, good adoption. And there's also been a bit of a journey here. We started with files being a SMB protocol product. So it had a bias toward Windows environments, user data. About a year ago we released NFS support. So now the game changes a little bit. You're talking machine data, machine generated data, right? So it's very different. And that's also forced us to rethink how we go about scalability, how we go about automation. You know, if there's a hotspot, the system should take care of itself. Does it go off and scale up? Does it go off and scale out? Does this happen automatically? So a lot of those things started to get beaved into the sort of the fold of the product. So that's files, I would say, the most mature outside of HCI. Objects is new. I mean, we just get it ready to go GA. We've done a bunch of early access with a few customers. Things are looking good, right? So looking forward to what we have there. Block storage. So we also offer just generic, ice-cazi-based block storage. That's also been in market for a while. And this has been use cases where somebody wants to run a bare metal database outside. Reasons of licensing or what have you. Maybe it's legacy databases. And I just want storage from the Nutanix cluster. So what we said is, you know what we'll do? We'll carve off a portion of the Nutanix cluster, logically speaking. Serve it out as volumes, right? Generic volumes. And you can use this for your databases. Performance and everything is very similar to what you would get if you were hyper-converged. So you're not giving up anything by doing so, other than the fact that obviously you're not in a true hyper-converged form factor. All right, since we're talking about storage, I wonder if we could drill a little deeper on some of the new stuff. So in 5.10, you're ready for memory class storage, things like NVMe. Where are we today? What is further down the road map? The storage industry, NVMe and NVMe over fabrics is pretty hot discussion. Everybody's getting ready for it. Is there anything that Nutanix is doing unique there? And give us what the customer expectation should be. Sure, sure, it makes sense. So I think at the more fundamental level, I think we all agree that if you're in a hyper-converged form factor, because you have storage right with compute, that gives you an inherent advantage to begin with, versus three-tier storage that goes over the network. So what we're trying to do is, hey, let's continue to milk that, so to speak. And in 5.10, we released, I would say, one portion of what we call AOS 2.0, and here what we did was we optimized heavily for metadata. So our metadata versus data model changes with AOS 2.0. And then what we'll do, we'll follow this up with major changes to the data model itself. So for example, now, when you're dealing with memory class storage, you've got to be able to address it slightly differently. You have to be using low-level APIs, things like SPDK, to circumvent the kernel, for example, and go directly into storage. So all those things are on the works. And the net result is going to be, well, I see higher performance, I see more consistent performance, I see lower latencies, right? And obviously more throughput as well. Now, you talked about NVMeo fabrics. Now, the idea there is, look, you've got the NVMeo protocol fabrics now. So what sort of fabric are you building? Because we deal essentially with Ethernet, ours will be an Ethernet fabric, right? So now we'll start to leverage RDMA more. We already do so in our systems today, but I would like to see end-to-end RDMA, where you start at the application and then right through the pipeline, through the data path, it's RDMA all the way down to storage. And even for your replicas, it's RDMA. And now you're talking a very different kind of latency, right? You're not talking, forget about a millisecond. You're talking about less than 100 microseconds of latency end-to-end. So that kind of sounds like the perfect use case for IoT, heavy data processing. What are some of the efforts you're undertaking to optimize for XI IoT? Right, so IoT, you know, there's obviously two pieces to IoT, right? There's what the computing I do on the edge, and then the computing I do after the fact, somewhere else, machine learning models that I can feed back to the edge, right? So this new technology would apply in both places. Now, when you're on the edge itself, there's certain situations where your real-time processing needs to be real-time. It better be quick, right? So faster my storage, the faster my decision-making. And then, so let's say you're able to make decisions faster, inferencing decisions faster in real-time. Now you go to the cloud, shall we say, where you're doing the long-time processing. And there too, it's a matter of, okay, I'm doing all this machine learning. I have a bunch of, say, AI or ML packages running here. There too, there's an angle of time, you know? If I do this in two weeks and feed it back versus two days, there's a big difference in business value that's being delivered, right? So I think the applicability of all of these changes is across any use case. All right, Tim, I want to give you the final word. You know, you've got the core products there, but what are you hoping that customers walk away from as they leave the show this week? So I mean, I would say, you know, your customers, we are ready for all use cases, all workloads. We are getting better and better. You will see us be on the leading edge, on the bleeding edge, when it comes to core technologies. I think we are a first mover. All the things we talked about, we have been investing in, this is not the first time. It's released for the first time, but it's been around. We've been developing it for multiple years. So you can think of Nutanix as someone who's on the forefront of all of these new technologies. And at the end of the day, it's all about your applications, being ready for all of those applications, traditional as well as new. And in your choice of form factor, you wanted to go hyper-converged, great. You want to go storage-only, it's up to you. All right, well, Tim, really appreciate the updates. Congrats on all the progress and look forward to watching where things go in the future. Thank you guys. All right, be sure to stay with us. Got a couple more interviews left here from Nutanix.NEXT 2018 in London, England. Thanks for watching theCUBE. Thank you.