 Live from Las Vegas, it's theCUBE. Covering Discover 2016 Las Vegas. Brought to you by Hewlett Packard Enterprise. Now, here are your hosts, John Furrier and Dave Vellante. Okay, welcome back everyone. We are here live in Las Vegas. This is Hewlett Packard Enterprise Discover 2016. This is Silicon Angles theCUBE, our flagship program. We go out to the events and expect the signal from the noise. I'm John Furrier, my co-host Dave Vellante. Next guests, Milan Shetty, CTO of HP Enterprise Storage and Eric Holberg, CEO of Opus Interactive, Cloud Service Provider. Welcome back to theCUBE, great to see you. Welcome back for the first time. Thank you for having us. Cube alumni now. No, he's been on. You've been on before? Yeah. Cube alumni, all right. 2014. You got it. There you go, great memories there, dude. Running a cloud service provider requires a lot of storage action, multi-tenancy, a lot of complexity. Tell us about how you're using HP Storage. Yeah, so we picked HP Storage before HP even acquired this particular type of technology. It was a software-defined storage, right? So Left Hand Networks was the original company that we went with back in like 0405. We've been using Hewlett Packard Enterprise Equipment since the early 2000s. And we really wanted a storage platform that was scalable and we didn't want to be controller-bound, so we made that selection early on, one of the very first customers. So when you say you don't want to be controller-bound, you mean big, expensive, disc controller-bound? Yeah, the big monolithic sands with, you know, yeah, they got redundant controllers, but we're still controller-bound. Eventually, after you add enough shelves, you're going to need to then do a lift and shift, move all your data. That's obviously time-consuming, it's risky to our clients since we obviously have to cater to a lot of different verticals. We wanted something that every time we added a shelf, we would add performance. And that's where what now is obviously store virtual VSA. You know, we're nearing a petabyte of that particular platform. Almost all store virtual for us at this point on our platform. What was the biggest problem that you had to solve that software-defined storage paradigm fit the bill for you guys? And then how did that shape going forward in today's market, what did that solve for you guys? Well, you know, it solved a couple different key things. Obviously, right, cost was one aspect. You want enterprise-class services with all that multi-tenancy redundancy for customers. So having each of those pieces sort of, you know, woven into the story behind it was sort of key for us. And then obviously just with the enterprise support from the services side of Hewlett-Packard and the partnership we've had, that maturity over a long period of time since we've been with them for so long just made a perfect marriage for us. So Milan, usually when you come on theCUBE, we talk about three par. We're not, let's save that maybe for storage day tomorrow for the party tonight. But this whole notion of software-defined is one of the vectors that you're going down. That's right. Maybe talk about that a little bit and how it's evolved into this notion of a storage fabric. Yeah, so in HPE, HPE has not been a pure storage player, right? We have an external storage business. We have also lots of servers with internal storage we ship. And in fact, with the latest analyst of IDC report, HPE is the number one storage provider period. You heard that today, the keynotes. Number one now. Yeah, number one. And both internal and external storage combined kind of number and very first time in a very long time. So we always saw two different design centers. There is a design center, which is an external storage, three parallel as you lamented. That's something which I've talked about in every CUBE session I've had with you. And this time, thanks to customers like Eric, I have an opportunity to talk about the software-defined storage and the common data fabric, as we're calling it. The data fabric to us, the second design center is about the software-defined storage and also anything which has got servers with internal storage, whether it's a rack mount or a bladed form factor. In Eric's case, Eric has the bladed form factor which is predominantly been using. So in the common data fabric, there were three major attributes which we have been looking at from a common data fabric standpoint. You need a stack which can deliver storage services and call them the three Ms. maturity, manageability at scale, and mobility. Those are the three main characteristics required for the common data fabric. And the interesting part about 2006, 2007 timeframe, due to its adoption by service providers such as Opus Interactive, we had acquired left-hand networks and that was the foundation of our common data fabric. We have over two million licenses out there today of the common data fabric already, right? And the maturity aspect which Eric mentioned, Eric's been in production with this in 2005, a decade of maturity. And from the storage industry, maturity, it takes a long time for the data stack to get mature. So the first time covered, manageability, the dimension about the one view and expanding that around the one view and the manageability at scale, and mobility. We know from our three-part experience as well how federation moving between the old and the new expandability and everything are the key characteristics. And that's the core principle of our common data fabric. So Eric, you're essentially running your business on this scale-out fabric, is that right? That's correct, yeah. Production workloads across multiple verticals. So when did you start the company? We started the company in 1994 and we split it off from this parent company in 2008. We've grown to five locations now. And so obviously like Malan was talking about manageability with one view is very important because we've got multiple locations to obviously manage all the different storage platforms. But we have at this point just shy of 1,000 customers and managing that many customers and their different workloads and requirements obviously is very important because we're dealing with that multi-tenancy aspect. And the vast majority are running on this fabric, is that right? Are they all running on this fabric? Yeah, they're all running on this fabric. Because most service providers that are, that have been around that long, if you poke around under the covers, there's one of these and one of those and three of those and just sort of depending on the storage platform does yours. So you basically swept the floor of everything about this fabric. Yeah, we're 100% Hewlett Packard Enterprise for all of our hardware and storage. Well, obviously we also do co-locations. We've got co-location customers that have their own stuff in their cabinets but that wouldn't be equipment that we purchased or managed. And everything is provisioned through an API, all software defined. Yep. Can you talk about that a little bit more in terms of how you got there and what that business impact has been? Yeah, I mean it's been a long journey, right? And actually Milan was asking me how many engineers we have and we're up to 11 engineers right now, right? So our ratio of the amount of servers we're managing engineers is quite high because of that automation and all of the software platforms that we have. In order to do that kind of scale, you're going to have to have, right, a very good software defined architecture and infrastructure to be able to manage that much servers. We had 30 minutes to go into overtime on this segment because we've got a little bit of cloud too here. So I got to ask you that kind of engineering, you have that 10X kind of developer mindset. Correct. You've got real good leverage. It's not like you have 400 engineers banging away at code, the old waterfall way. You guys are classic DevOps. Yeah, I mean really it's virtualization behind the scenes of cloud, right? I mean we were doing cloud. We launched our official cloud product in late 2005. Four is really being called cloud. We do a lot of hybrid between the COO and our cloud offering in addition to the public cloud bursting, right? So do you, I shouldn't say it this way but I'll try to fade it in a good, elegant way. A lot of people come into, I want to be a cloud developer or I want to be DevOps. I'm a DevOps guy. You can't really be a DevOps guy. You got to kind of earn it. And the early DevOps days or early cloud days, pure DevOps really was forging new ground. You guys are part of that wave. What have you learned over those years and how would you give people advice today that kind of want to go to school to be a DevOps engineer? Really I would say, from transition to also managing developers, being flexible in the platforms that are out there. Things are changing. Even like moving in from virtualization. Now we got containers and you got Docker, which is another announcement that was happening today, right? Having Docker being bundled on all HP servers. You know, that's huge for us too because we're getting ready to launch a container as a service offering. And but really from a DevOps standpoint, right? I think it's going to be coming down to that flexibility and being open to the different platforms as the change happens. And containers of service is one of those things where there's a lot of dogma out there right now relative to your view of it. Is there one size fits all container? What microservices are out there? What's your take on all that containers as a service? Is it best to be open 100% or? No, I think there's going to need to be a combination of this. So the platform we're looking to launch is going to be, you know, platform agnostic. We'll be able to do both windows as well as StarNix type servers. And we want to be not everything to everybody, but to provide that one missing piece for us, which is to really cover the developers and the people that are looking for microservices at scale, right? So here's some conversations we had earlier on our crowd chat. I want to bring up because the cloud guys can answer it. Keith Townsend said, I don't believe hardware is material. What value is containers of service, specifically over infrastructure service? I believe customers will roll their own container platforms and look towards providers for pass. Containers are just a footnote in the conversation. Pass will run on top of container plumbing. Thoughts on those two comments? Keith Townsend is a great, great CTO out there. Yeah, I mean, with any shift in technology, right? You're going to have to worry about those changes over time. But, you know, like we just heard on the keynote earlier, it's going to be about 50% of the workloads are going to shift to containers because it is like more nimble and you can move that across a lot of different geographies. And yeah, the ones that try to pull it in-house, like most things are going to eventually learn, like even the power. Like if you look way back in the beginning, not everyone wants to run their own power plant. Not everyone's going to be able to run all their own little individual servers. They want to focus on their application. The developers don't really care about the infrastructure or anything behind the scenes. They just want it to work and they want to be able to launch that fast. All the wrapper, whatever you want to call up, developers love this concept. Milan, try me on this. Absolutely. And I think one of the things which Barry and I were talking about this is that as a service provider, Opus Interactive and Eric and the team saw the value of virtualization and how virtualization can be delivered to the thousand customers which they have. And they chose on a common data fabric to make sure that at least that piece of the infrastructure is stable, mature, manageable at scale and provides the data mobility. And as they're looking to the container set of technologies, it's not going to be virtualization taking over containers and there's going to be both these technologies are going to be around for a while. There will be overlapping applications. There are different use cases and everything. And the fact that Eric and his team can make a choice of keeping the data fabric common across both their virtualized environment and also the containers in future is a real value proposition for them. And that's the power we see with the common data fabric. So at least on the data side and the data management side and the data layout side and the storage side, they don't have to worry about that. Container stuff, total home run, down to the storage. How much of that is automated? And how much is, because it's all about focus. We got some human involvement, but this kind of 10X developer can be applied to infrastructure if you look at it that way. So the goal is to free up the labor resource. Absolutely, absolutely. I think the provisioning aspect in addition to the application provisioning, the storage provisioning aspect is also pretty heavy, if not done right. Right, API's station, rest station and everything and trying to make sure that all the storage nodes in the software defined world, servers with internal storage, they may have different generation of servers in them. They may have different characteristics. They may have SSDs, some may have hard disk drives, some may have old generation hard disk drive and everything. And how do you automate and provisioning thousands and thousands of users and thousands and thousands of storage nodes very easily, right? And that's the work we are doing from an infrastructure provider to a very standpoint to make sure that in one click of a button, they can go. You do the heavy lifting. You do the enterprise as they lift it. HPE does the heavy lifting so that stack above is what Eric's team can focus on, which is that value. And that's the bottom line. I mean, we've been talking for a while now in theCUBE that there's a shift going on in the market from, we said about $200 billion from non-differentiated infrastructure lifting into vendor R&D, essentially. And that's really why you went in this direction, right? You get scale, automation. You know, we call it server sand. We've pegged by 2020 in the Wikibon forecast show more than a half of the spend is going to be on what we call server sand, what you guys are doing, fabric, software defined, pick whatever name you want, not the big external controller. I love that phrase server sand because customers who are in the traditional shops, right? The traditional shops, they always have been trained and beaten up by the other three-letter company which will soon be joined by a four-letter company that's sand, sand, sand, sand, sand. And when we talk to those customers, who are traditional customers, who have not gone to virtualizations and who have not gone to container and we say, this is server sand, they get it. Three plus four is seven. Three plus four is seven. I like that. Of course that could change the number one again. That's a horse race. That brings us to a good point. Server sand is a great word because it redefines really what the operating model is. But again, the constant theme for all of our conversations here in theCUBE, you nailed it, is the operating model of the customers changing and also the delivery of the value is changing. The 10X developer I talked about, I think Mark Andrew recent quoted that term around kind of the new age developer. You don't need 100, you can do 10. Same with infrastructure. You guys can enable that. That's game changer. That's right. And one of the advantages we have from an HPE infrastructure provider standpoint, the synergy platform, composable infrastructure, bladed, rack mount and everything, doesn't matter. But it is actually the same software data storage stack across all the different form factor. And that provides, even from an R&D standpoint, we do automation one across multiple platforms. So Eric can choose, he wants rack mount or Eric can choose if he wants blade or combination and it's good to go. Final question, since we got a break here, I want to get your thoughts because we were talking earlier, silo stove pipes as Dave calls them, I call them silos. We're about best of breed. Certainly when you talk about being a box mover back on the old days, now to a more solution oriented, how does best of breed change when ultimately you got a horizontally integrated architecture where that's industry standard hardware and or specialized boxes like service and whatnot? How does the definition of best of breed change? Because they're about converged and composable. That's not best of breed because there's now a lot of different components. Some might be best of breed for a workload. What does best of breed mean? Does it mean anything? What is the new term then? It doesn't mean anything. Thoughts? The new term is server sign. No, so ironically what's happening right is that there is a lot of changes which are happening in the blade and the rack mount servers as well. The non volatile memory come in, the 3D X point and everything. There is a lot of the fundamental shifts which are happening in the hardware itself. And the best of breed now is going to be first led with automation, whatever has to be done at scale. Automation is going to be a key characteristics and a key attribute. Second is going to be that whenever the non volatile memory and the prices of these are going to come down, non volatile memory technology is coming in the servers and everything, is that what is the software stack? What is the mature software stack which you can deploy on is going to be very, very critical? I think maturity is going to be a big, big dimension of the automation and maturity are going to be key characteristics of server sign and what was- Integration too is a big table stake item, right? You talk about dealing with customers, making sure you've got multi-tenancy, very complex, absolutely. Exactly, and I think the definition of best of breed now is probably going to morph very quickly to customers such as Eric and Eric's customers want to adapt and be flexible and move very fast. And how do they do that? I think best of breed is changing to winning, just being fast to the finish line. That's right. Versus speeds and feeds. Exactly right, exactly right. I'm trying to find and put my finger on the word. Well, I mean, best of breed made sense when you're talking about like this box is really good, it's got X MIPS or whatever speeds and feeds that those days are over, but now every outcome is different based on the workload. You can't really say based on this spec, that's better, it kind of depends. That's right, that's right. And you know, the funny thing is right, I think the, if in two years, we're still talking about block, file and object, we failed. We failed, it should be invisible. It should be invisible. And maybe the new phrase is the infrastructure is generally invisible, and that it just provides agility for applications and the service providers to deploy what they need to be. Maybe you're going to change your business model to a revenue share, free storage, what's the revenue share? Share the outcome. Guys, thanks so much. The customers would never do that. Innovation in the queue, we're just kind of connecting the dots every day. Real time, Bill and Eric, thanks so much for sharing the insights. Congratulations on your business. Great to hear about the success and being through the early days, pioneering the DevOps and being successful. Congratulations. Now this is theCUBE, connecting the dots here in real time, best of breed, server sand, all here in theCUBE. We'll be right back with more live coverage from Las Vegas after this break. You're watching theCUBE.