 Live from Las Vegas, it's theCUBE. Covering VMworld 2017, brought to you by VMware and it's ecosystem partner. Hi, I'm Stu Miniman here with my co-host Justin Warren and we're at VMworld 2017. You're watching theCUBE worldwide leader in tech coverage. Happy to welcome to the program a first time guest, Juan Gaviria, who's with ADP and he's the senior director of technical systems engineering. Juan, thank you so much for joining us. Thank you for having me. It's a pleasure to be here. Yeah, so VMworld, it's my eighth year coming to the show. I've been part of the community for a long time but one of the things that people love at this show is about 20,000, maybe a little north of that. It's peers talking to peers, people that dig into the technology, find out what works, how to do things better and everything. Tell us a little bit about your role. I think most of us, no ADP we've ever gotten, checks with the logo on it or lots of various other services but what's your role inside the org? Yeah, sure, so really quick about ADP to your point, the logo's pretty well known. We actually pay one in six people in the United States so over 25 million employees we pay. We have over 650,000 clients in our mobile app which is really the way I would recommend that you look at your paystubs, 401k, benefits, et cetera has been downloaded over 12 million times. So the ADP brand is doing well. It's a healthy business. My role specifically is that I manage all computed ADP so think about servers, server operating systems and server virtualization, that's my role. Yeah, you brought up mobile so maybe start there, Pat Gelsinger this morning was talking about kind of digital transformation. We look at financial services as how do you reach those users, what does that kind of ripple through to all of the things that you manage? How long have you been there and what kind of changes have you been seeing? I've been there 15 years and I've seen a lot of changes. 15 years ago they probably weren't even virtualized. No, no, in fact I remember rolling out ESX 2.x and using the good old movie so we've come a long way and mobile has just been explosive. From a product perspective the goal now it's mobile first. So even now if you think about your benefits when you go and roll in your benefits every year the goal is to make that experience translate to mobile and that's a little harder than it seems but that's the goal for ADP, it's everything mobile. Right, to bring us in, what's kind of the scope of what you manage, you said ADP globally, what you handle, but what's kind of the team size, how many devices or VMs or however you manage, what are you listed in? Sure, so my team is responsible for computers I mentioned so think of everything from demand management through operations. We have globally about 50 associates that are responsible for that. We have over 3,000 ESX I-hosts deployed across seven global data centers with well over 40,000 VMs. So it's a pretty good size infrastructure. And bring us inside, VMWare, how long have you been using it? What pieces of VMWare in the ecosystem or have you been using? We've been using VMWare again since the early days of server virtualization. We're a VROPS customer, a VRA customer, in fact, VRA, we're leveraging it for infrastructure as a service to our dev community. We have for ADP thousands and thousands of developers. So just the amount of churn in our private cloud is tremendous. AirWatch, we're a big AirWatch customer as well. Yeah, expand a little bit on the developer piece. What do they look for? How does that impact? You know what you're doing? Yeah, sure. I don't know what they're looking for because it's always changing, to be honest with you. But we have somewhere around 6,000 developers and they're obviously developing ADP's next generation product. So they're just looking for us to get out of their way. They want VMs, they want them now. They want containers, they want them now. And every day I turn around, they want bigger VMs, bigger containers and it's getting harder and harder. So through VRA, we provide those pools of capacity and then they're able to spin up, tear down, rebuild VMs as needed. On a monthly basis, what I see through VRA just in the developer community lab is about 3,000 or so actions a month. So it's pretty high amount of change in that environment. Yeah, but based on what was announced in the keynote particularly around the partnership with AWS, do you think that's going to resonate with the developers? Yeah, absolutely. Most of our, not most, but a fair amount of our next generation products are being developed on AWS. So everyone wants to be on AWS. In fact, we're bringing in a lot of college hires and as soon as they come in, they say, I want to work on AWS, right? So for us it resonates because what AP does, security is key and we want to have a hybrid cloud. So we were actually part of the Lighthouse program. So we were early customer, got to see the logo during the keynote, which was nice. So yeah, we plan on leveraging that relationship to help us, for example, burst in that dev cloud. I'll unpack that for us. So one of the things we look at, when I hear hybrid cloud, I need you to explain that because every customer I talk to, it means different things to me, especially you mentioned things like bursting that little scary sometimes. So maybe explain what that actually means in your environment. Yeah, so in the dev environment specifically, what it means is as I mentioned, we get requests that come out of left field, right? I need a 300 gig memory VM and 10 terabytes of storage. You're just like, I don't have this, right? I don't have hundreds of those. So we can put that capacity out on AWS much faster and as those projects materialize, we can then bring that back in. So that's what I mean by hybrid cloud. Okay. For us. One, so you're using the VMware on AWS. You've been testing that out, you said. My understanding is you're also using VSan. Is that separate from that? Because VSan's part of the, you know, what's it? VMware cloud or cloud foundation suite piece of it. So what's your interest been in VSan and how does that fit into the entire picture? So it is different for us. The AWS relationship is going to be more of a managed service, obviously. We're actually going to come a consumer, right? So we're going to feel like our own customers. To answer your question on VSan, yeah, we've made a huge investment in VSan. So all of our VM storage, which again is, you know, 40,000 VMs worth, which is, you know, well over, you know, four plus petabytes of storage. We're moving that all to VSan. Wow. What's happened to all those arrays? They're going to be gone. Yeah? They're going to be gone. That's a really big move. Can you, you got to take us back, you know, how do you, is this a top down or, you know, bottoms up, you know, what would walk us through some of that life cycle? What dotted that? Like how did you come to the, even begin contemplating replacing all of your storage? So it's been both, to answer your first question, both top down and bottom up. I mean, we've been looking at the technologies for a while, right? And just kind of keeping close to them. At this point, they're mature enough that we feel they can run, you know, our business critical products. And it's been a journey, right? For the last year, we've spent looking at, you know, all the different market leading technologies and figuring out, you know, which ones make sense in an environment our size. You know, how do we operationalize this thing? So it's been a journey and this is the beginning for us, right? So we're actually, as I speak, we're starting to deploy our first VSan clusters in production. And, you know, we're deploying it in hundreds of servers at a time. So it's exciting and interesting times for the team and I. Yeah, one of the interesting things is some people look at VSan and they're like, oh, well, it's kind of small deployments. But we had some of the VMware people on earlier today. They're like, we're deploying internally, but it's lots of clusters. Because if you tell me hundreds of servers, I'm like, well, that's not a single cluster. That's lots of clusters. How do you carve that up? How do you manage that? How do you roll that out? You know, what does that look like? Yeah, that's the trickiest part, right? And by the way, as we looked at different solutions, the cluster size became one of the reasons why we chose VSan. A lot of the other solutions that are out there will limit you to about eight node clusters. And to your point, we have thousands of hosts. That's hundreds of clusters. So VSan gives us the ability to have slightly larger clusters. Today, we're going to look at about 16 node clusters. To start, that seems to be where VMware is going as well. So we'll follow their lead. We figure they know what they're doing. And we'll manage that using V-Rops as well. Yeah, I was curious as to what was actually driving the change to VSan. And what was it about VSan that said, yes, this is great. This is the one that we're going to pick. You've mentioned cluster size. Were there other things that made you sort of decide that it was clear that VSan was the right choice for you? Yeah, so to me, the way I look at VSan from a VSphere perspective, it's that they've made storage a feature. You know, and our V-Sphere administrators, they know how to run V-Sphere. Now they just have another feature. So that was one of the main reasons, just the operational efficiencies from a team perspective. There are a lot of other reasons as well. Security, some of the other competitors out there, for example, didn't have encryption when we were looking at it, which is everything we do revolves around security. So that was another key reason for VSan for us. And what drove us at first was really, with the traditional models, we found ourselves to not be very agile. Because our business is growing so fast, we're building about six months of capacity at a time. And if you can think about the cost of that much capacity at a shot, it's millions of dollars. It's kind of sitting idle, right? So with HCI technologies and VSan specifically, we think we're going to be much more modular in our approach and closer to just in time. So we expect significant capital benefits from that. So if I hear you right, it's kind of the pooled nature of what you're doing and that the building blocks are small enough that you're not getting to what people usually have, is like, oh yeah, I've got all this capacity and I'm three years in and I'm still not using a lot of what I run into. I over buy so much because of that. Exactly. And think about that first purchase. You got to sit with finance and say, hey, I've got to go buy an array and I've got to go buy a couple hundred servers. Now it's, I don't have to buy that much up front. So it's a huge benefit for us. And it sounds like it's going to be core deployments as well because a lot of the HCI deployments traditionally have been for remote office things or just particular workloads like VDI will be one thing that it runs on. But it sounds like this is going to underpin pretty much everything that you do. Pretty much everything, yeah. And in addition to VDI, we have a very large VDI deployment that supports all of our customer support reps. And it's going to underpin that in addition to underpinning all the business products that you use to view your pay statement. All right. So you talked about the finance people. What about the storage people? And I have to imagine you had storage admins. You know, you look at it and you say, okay, are they out of a job? Are they going to work on new challenges? Can you walk us through how you approach them? How they've looked at this whole migration? And yeah, what happens to them versus the VMware people? The virtualization admins, I should say, yeah. It's a funny question, right? Because I've become a little bit more popular now with the storage team. They've actually knocked on my door and said, hey, anything we can help you with, right? But no, it's a good partnership. My peer and I, who run storage, we actually built a team together that's going to help us roll out vSAN. So we know that there are skills in the storage team that we can leverage. And our vision of it is that we no longer are going to have vSphere administrators or storage administrators. We're going to have cloud engineers and they have to know compute network storage really because we view the skills as converging as well. It's not just the software and the hardware. Right, okay. How about the management of that, though? So are you going to just essentially be managing a team together rather than it being separate people managing different people? Correct, it's one team. One team. Yeah, this is really interesting, Juan. You know, I'm just curious in your kind of evaluation phase, what did you learn that kind of, if you had known it at the beginning, you might have either accelerated or you might have positioned things a little bit differently now that you're ready to kind of this massive roll out. You know, I think I would have had maybe stricter entrance criteria. If you think about a company our size and all the partners we have, we looked at a lot of different solutions. We spent a lot of time in the lab. We're in the end, we knew that, for example, an eight node cluster or not having encryption, we're show stoppers. But yet we spent the time in the lab to do that. So my recommendation or advice to my peers out there is come up with good criteria that you know you have to have. And then from there, do the paper exercise and bring in the ones that you know will actually be able to get to production. What was that entire kind of evaluation phase? How long did that take? More than six months. And can I ask what underlying deployment you're going to use for VSAN? From a hardware perspective? Sure, HPE servers, DL360s. Okay, and what led you to choose that versus, you know, the DL people are on all lined up to say, you know, come on, we own VMware. You know, you should do VxRails or? Yeah, well, VxRail to me is a little bit different than just VSAN. But yeah, absolutely, DL was pretty interested in that business as well. And the beauty of VSAN is it gives us the choice, right? We've been a long time happy HPE customer. So for this first phase, we'll continue to be with HPE. And for some reason, if something changes, we know with VSAN, we have that flexibility. Okay, you've been with VMware for quite a while. I'm sure you've been watching VSAN. What are you still asking them for? You know, they've had a very aggressive roadmap. I think they've got most of the basic check blocks done. They've heard a little bit about the roadmap, but what's on your to-do list for VSAN or any kind of the associated pieces? Okay, you mentioned VxRail as an example. And the automation that they've brought with rail is significant, it's very valuable. I think they need to bring some of that same automation to VSAN stand alone. So as I think about patching thousands of hosts with VSAN and the drivers and that entire matrix of things, they've got to help us there. I think they've got some work to do in terms of improving the performance management of that because environments this size, managing that manually is too much work. So I think we've got some work to do there. But they've been a great partner. They've been listening to us. So I'm pretty happy about where they're headed. Earlier you mentioned deploying VMs in containers. Is that like Docker or how do containers fit in? Yeah, so Docker has been sort of a religious debate internally, to be honest. Do you deploy it on bare metal? Do you deploy it on VMs? And I think right now we're settled on deploying Docker on VMs, but very large VMs. We're thinking 200 gigs and the goal will be, we're going to try to do that on VSAN. So we're still in early development there, but that seems to be where we're finally landing on. All right, interesting. And I'm assuming that's Linux on top of the VMs to allow that. Yes. All right, well, Juan Gaviria, really appreciate you sharing that. Really interesting use case. I wish you best of luck on the rollout. And thank you for being on theCUBE. Thank you, thanks for having me. All right, for Justin, I'm Stu and we'll be back with lots more coverage here from VMworld 2017, you're watching theCUBE.