 Live from Gillette Stadium in Foxboro, Massachusetts, extracting the signal from the noise, it's theCUBE, covering VTUG's New England Winter Warmer 2016. Now your host, Stu Miniman. Hi, welcome back to theCUBE here at Gillette Stadium, actually on the EMC level of the VTUG event here at the home of the Patriots. Happy to have on the program two guys from EMC, Brad Malt, who's the Senior Director of Converged Infrastructure, Strategy and EMC's Corporate Office of the CTO. Brad, your second time on the program, first time in your role at EMC. Thank you for joining us. Thank you. And I've also got John Mark Walker, who's the Director of Open Source Programs with the EMC ASD Group. John, your first time on the program, I believe, right? This is my second time. I was on when I was the Gluster guy for Red Hat at the OpenSack Summit. Excellent. Two years ago. So both of you, your second time on the program, but first time in your roles at EMC. Maybe let's start with, what brought you to EMC? What are you currently working on? Brad, we'll start with you. Sure. So because I'm in the CTO office, working under John Rose's organization, I have the fun time of getting to look forward a lot and getting to look at all new technologies as they come out. How are they impacting EMC? How do we utilize those technologies? And really trying to help the VCE organization now the EMC Converged Platforms Division really drive more of their products forward. All right, so John Mark, same question for you. I came here from Red Hat like four months ago, specifically because they EMC released Copperhead as an open source proposition. I thought, well, huh, that's interesting. So I started talking to these guys and wouldn't you know it? There are lots of open source plans and I'm here to kind of direct open source traffic in the Advanced Software Division. Okay, I guess, yeah, let's jump off of there and talk a little bit. What does it mean, you know, what's happening in open source at EMC and how's the culture of EMC fit with open source? Brad, maybe to start with you? Sure, so we definitely have two different perspectives. So I'm gonna take the perspective I have, which is open source for EMC is actually a very vital for us to really embrace. And we've actually been doing a very good job of that. As John could talk about, I would say his organization and once tightly tied to John's are doing an exceptional job of getting different projects submitted through code, EMC code and other organizations. I am very much tied back from the CTO office to one of our projects, Rack HD, which we're gonna talk a little bit about. But for me, the open source initiative is to make sure that customers understand we have nothing to hide anymore. It's out there, look at what we're doing, work with us on it, adopt it in different ways and then watch as we introduce it into our EMC technologies. And I just think the open kimono approach that we're basically pulling on here is really a benefit from an EMC perspective. It's a cultural shift. Yeah, Brad, I'm curious, your perspective, you came from kind of the partner channel side. So was this a gap in EMC's offering before, in your opinion? So the funny thing is, specifically in the Rack HD project, which we'll talk more about, I wouldn't say that's so much a gap in EMC's portfolio as it was something that was being done multiple times over. So we were looking for a stream lineable, automatable thing that we could latch onto. It worked out really well, hardware management orchestration, let's automate that more, let's publish it out through open source and let's try and simplify EMC's go to markets. And that's really what Rack HD is doing through open source now. So John, you came from the open source background. The place known for open source. I've been involved in open source companies since 1999 when I was at VA Linux and we had our big IPO and then fell off a cliff with everyone else in the dot com boom. But it's interesting, back then, the question was, will open source survive? Will open source be important? And it's really interesting to come to the point where it's about, wow, look at all this open source stuff in the infrastructure side. And the big shift that's happened now is that customers are actually asking and demanding for open source. They want to reduce their risk. They don't want the risk of vendor lock in. But at the same time, there's a place for companies like EMC to come in and reduce the risk even further by saying, okay, we're going to take this open source code and we're going to productize it, make the experience better for you, we'll make it integrated, you can do a lot more with it, it'll be a lot more flexible, but it's still going to be a product that we sell. And that's kind of what I came on board to explore and define internal EMC. Okay, so you've mentioned Rack HD, Brad, can you just for audience explain the basics, what it is, what problem it solves and why it's needed? Sure, so the Rack HD project at its most simplest level is called Hardware Management Orchestration. What it basically is, it's a API driven hardware abstraction layer that allows you to manage different server slash compute platforms and network platforms from their lowest basic bits. It allows you to manage the BIOS and the firmware, to configure the BIOS and RAID controllers and NIC cards, to inventory and discover servers and switches upon initial boot up of that device. It allows you to lay down the operating system with a predetermined configuration from somebody northbound. It basically allows you to do this but across a heterogeneous set of vendors. It's not tied to EMC hardware. It allows you to do this with white box vendors through Dell and HP and Cisco and U-Namen. So because of that, when people northbound want to automate infrastructure, why call five different tools because you have five different components in the infrastructure? Why not be able to make a single call and have that abstraction layer disseminate that amongst the actual vendors underneath the covers? And that's really the purpose of where RAC-HD is going. Okay, so I guess one of the questions that leads to, and maybe John, you can help answer, is it's great that I think the idea of being able to manage my infrastructure from an IP standpoint, I hear a lot of customers asking for that. The question is, who gets involved in it? Will the community rally around this? Will this just be an EMC thing or how does the community embrace it? I think there's a strong demand in general for from customers, from partners, from all sorts of individuals who want this flexible, open source infrastructure. And if you look at the kind of overriding themes between our worlds, it's all about, it's all about the flexible infrastructure. It's all about integration with existing workflows. And when you think about the best way to do that, obviously open source gives the best pathway towards that kind of fully integrated environment. So if you look at something like Copperhead, it's all about how do we take these multiple storage systems underneath and find a way to attach them to these multiple orchestration frameworks sitting above everything. How do we connect them in a way that makes sense with a customer in a way that's expected and delivers you user experience they're looking for? When you look at how much open source enables from the converged infrastructure point of view, there really can be no other alternative except for the infrastructure to be open source. What allows things to connect without going through the traditional rigmarole of, well, I'm a product guy, let me talk to this other product guy at another company and see if we can do something. All the middleman is completely cut out with open source because now we got engineers talking directly to each other. And to that end, Intel is a huge partner of ours on Copperhead. They've devoted an architect and engineer and they're actually bringing four more engineers to the project. They're leading the integration with OpenStack, for example. So you talk about that northbound integration. And they also connected us with some grad students at Oregon State. And so they were working with us on the southbound side helping us connect more storage systems underneath. So it's all about the, you know, multiple parties coming together to solve this problem that everyone has. Yeah, so, you know, as an industry, we've been talking about automation for quite a while now. So I guess, you know, two parts. Number one is why now? And two, you know, we hear some organizations, especially small organizations that say, you know, automation, oh, it's going to be too hard. We don't have the time. You know, I guess, what's the sweet spot, you know, and how do we kind of get over the hump to automation? I think right now you're seeing a maturation of open source tooling that's allowing these things to be pieced together more seamlessly. You know, OpenStack didn't exist five years ago, or if it did, it was a very immature version of it. And now you're getting to the point where between OpenStack and the container world and all the pieces that make up what we're talking, you know, calling cloud native, you know, cloud foundry as an example. All the pieces are adding up now, and now with REC HD you're starting to see that go all the way down to the hardware layer. And you can get to the point now where you can build this fully integrated stack that's fully open source, and that's a really exciting time. You're just seeing more acceleration based on the maturation of all these individual components. Well, I think part of it also is the adoption of not just open source, but automation as you're kind of hitting it on. When you look at the industry, you can break out the service provider cloud types of vendors out there. They very much are heavy into automation. They're the ones that are pushing, driving the open stacks of the world and all those different frameworks. So at somebody at that level, they want something open sourced, able to automate, API driven, they want all that. The enterprise guys are kind of in a really funny space right now, right? We have this whole DevOps movement, cloud native apps, whatever you want to call it. They're trying to turn themselves into what the web scale guys are today, right? They realize for their business to succeed long term, they have to automate, but they don't know how. They don't know they're learning on the fly. So the more technologies that we can put out there that enable automation of different aspects, I think the easier overtime it will become for them to do it. Now the real trick is not to just give them a layer of technology and say, here, go use it. It's going to be, here's a layer of technology that integrates with this other layer of technology that has this really cool puppet chef ansible thing on top of it. And all you got to do is build a recipe and everything works together in a stack. And that was the promise of OpenStack at some level, which is, hey, here's this gigantic thing, but there's also other smaller variations out there, Mesosphere, Kubernetes, we can go through the list. Now the real trick I think in the industry is going to be in the next five to 10 years in the mid-market SMB space, which is, how does that keep going downstream? I don't see the SMB mid-market space going after API integration on their own. They just don't have the staff, the people, the IT expertise. But the learnings from enterprise and service provider are going to probably end up rolling up into probably simplified UIs. And this is where converged infrastructure, hyper-converged infrastructure is interesting. Because if we can take the learnings of the service provider, cloud guys, enterprise guys, roll that in through UI automation layers that use the same tools as the enterprise guys, but deliver them through different packaging, you see the future where converged and hyper-converged needs to go. It's so interesting to hear you talk about that because about 10 years ago, I worked at a company called Hyperic, and we were all about open source management. And one of our mantras was make your stuff manageable. And at the time, no one could figure out what on earth we were talking about. I guess we were ahead of our time. Eventually it was sold to VMware and it became part of that suite of products. But it's funny now because to your point, everything you're seeing now that's part of the cloud native or third platform is built to be managed and built to be integrated into this larger orchestration framework. And that's a real sea change, I think, because people already have in mind that their tool that they're creating is going to fit into this larger context and they have to make it able to do that. All right, so Brad, I'm wondering if you can help connect the dots for us between kind of where converged infrastructure is. So obviously- Didn't they just do it? Yeah, so let's go down a level just because you'd say, okay, a customer, I bought a V block or maybe I'm looking at the VX rack type stuff. And VCE has their vision software, they've got the VCE V scale, which is trying to help pull together some of really the EMC pieces. So how do I get from there to what we've been talking about with the rack age today? So I think what's going to happen is it's going to be a maturity of things over time, right? When you look at the original premise of a V block, the whole purpose of that was to deliver all of these technologies that you could buy yourself and DIY it, but to deliver it more of a purpose-built, best practice as engineer solution and allow you to consume it that way. The original intention of that was never to get into the DevOps space, was not in to get to this API programmatic space because we didn't know about that space honestly back when VCE started. Whereas when you look towards like a VX rack type of a delivery model like we're aiming at now with that, that is dead set in the middle of this whole cloud native apps platform three type of a world. So by its nature it is being developed with those primitives in mind. It's using rack HD through on rack underneath the covers, right? It's using vision as part of its governance and compliance engine and it's wrapping that together with UIs and workflows and stuff. Well what you're going to see is those types of learnings that we pull out of the VX rack world are going to go into V scale. They're going to go into V block and all of a sudden over time you're going to be able to see the EMC converged infrastructure and hyperconverged portfolio be able to be consumed through UIs and APIs that should scale across V scale, V blocks, VX racks and all the different aspects. All right, so at Wikibon we've got a new definition we put out recently called true private cloud. True private cloud. Well, to be honest what we say is most of the stuff we've been talking about for the last few years if you go to 10 customers and you ask them what a true private cloud is and what they have. The pseudo cloud. They all have different pieces and there's not a lot of similarity. So, you know, similar. We have some VMs sitting over here. The cloud, right. It was virtualization plus is kind of what it is. So the management orchestration is a big piece but also one of the things I'm wondering if you can help kind of square the circle on is what about the public cloud integration because customers, every customer's using public clouds and tying it in and from an application standpoint my data is going to be in a variety of places. We've got applications and reasons to have things at the edge or on-prem but there's resources that I'm going to leverage from the public cloud. So how do these things that we've been talking about tie into public? The trick with the public cloud is it's the definition of cloud and I'm not going there again. But in the public cloud space you actually have to get into what do you mean by integration of public cloud? So what I mean by that is you have the data aspect of integration, right? Well, is my data replicated to a public cloud thing or does it have to be translated when it moves to a public cloud, right? So there's the data integration aspects which obviously EMC has done very well at. We've made a ton of acquisitions recently and these are kind of what the inter-cloud linkings that you have to talk about. Things like cloud array which allows us to cache data as well as move data up into public cloud. You know, mad genetics and all these different things that we have that back up in the cloud and all that. You're seeing EMC push people towards we got to treat data in a hybrid, you know, private cloud stance almost the same way you do in the public cloud and we just had to extend. It's an extension game. M&O is different though. Management orchestration is a very different discussion because this is when you get into the discussion of I have a virtual machine running in my local cloud, my local site. How do I get that to run up into a virtual stream into a Amazon, into a Azure or whatever it's going to be? That's a very different game because you're not dealing with the data aspect as much as how do you manage them in the same way with the same SLAs? You know, how do you extend all of your localized data center primitives into and translated into the public cloud? That's where we look at things like virtual stream helping us out a lot in that space. That's where we look at a lot of VMware technologies and we say if we can extend virtual stream and some of the VMware aspects into the public cloud space which is what's happening very quickly, then we have a great story. Data side happens on the back end from an EMC core perspective. Virtual streamer ready is that hybrid public private depending on how you utilize them because they can be used in all stances. VMware has again public private cloud instances and if they all get packaged and delivered through VCE then you can see in the future of VXrack or VBlock being an extension of a public cloud and not vice versa. Because some people look at public cloud as an extension of private cloud but why shouldn't you look at your private cloud as an extension of a larger public cloud? And that's what another stance you can actually take on it through our tool set. I think customers definitely want to be able to manage all the things whether it's public, private or hybrid. I'm not sure that customers actually want to migrate data to and from but they definitely want to be able to control and that's kind of the biggest thing that they need to be able to do. Whoever can deliver that the first test and what the most test is going to be is the big one in the data center. I believe it's automate all the things. Correct? Automate everything, exactly. All the things. It's nirvana. Absolutely. So just, I know we're running short on time here but what about containers? How does that fit into everything we've been discussing? Is this ready for it? Containers. It's not, yeah. What's? Sorry, just joking. I was like containers, what's that? So container, I mean actually Rexray is probably one of the great discussions we have but Rexray is a project started by the RMCE code guys talking about basically storage abstraction for containers and it's kind of like a container's eye view of storage underneath whether it's local to the container or stored somewhere else in some other storage system. But containers in general, I mean it's what automates it. If you look at the vision of cloud and what it was intended to do, containers I believe allow us to move more quickly to that ultimate destination which is ultimate application delivery, ultimate application support regardless of platform. There's a word I'm looking for that I'm not finding but you know where I'm going with that. The ability to have a application work in an expected way no matter where you're launching it or where it's going or where it's pulling data from. I think for me containers, well first of all when you look at EMC, I'm going to put containers on the same level as virtual machines. Not that they do the same thing or they act the same way but when you're looking to deliver an application or a microservice it could be in a VM, wholly owned separate container that's totally secluded or sorry that type of container or the native Docker container which has a lot of different aspects to it. There's a lot of concerns and issues in the natural container space which is security, persistent storage management. We can go down the list things that in the virtual machine world have been solved but they both are delivery mechanisms for a application or a service. With that being said, as you move forward into a lot of these natural features moving up into the application, into Cassandra, into Redis and all those fun things, you don't need all those natural things to live down at the virtual machine level. It's okay living higher and getting less from a container world and that's how we view it. So if I was to look at containers and for me the converged infrastructure world, they're the same class citizen as a virtual machine. There's no reason in that infrastructure layer I should be able to handle both of them and do similar things maybe for different use cases depending on the applications running on them. I guess one of the things I just want to make sure is if I look at what Rack HD does, if containers become more important, it hasn't, Rack HD is ready for that. Rack HD, well the cool thing is we don't care. We will lay down whatever operating system you ask of us, VMware, or you can go into any Red Hat or Linux with KVM enabled or without it. And so in the end of the day, if you just need containers, we'll lay down CoreOS or whatever you want for that. If you want virtual machines, ESXi, or whatever you want there, we can handle all of that on the same hardware. So for us, we don't care what you're going to put on top of it. Right. Okay, so please. I would say there's still the perception that containers are faster. Whether that's true or not, there's still that perception that definitely persists out there in the ether. Okay, so if people want to find out more about these projects, it could contribute to them. Where do they go find out more? We can go to copperhead.org and start downloading today. We have vagrant scripts, OVF files, or just source code that you can play around with. And for RackHD, you can go to github.com slash rackHD. The other best location for all of this is actually the EMC code website for EMC, where you're going to find links to all of our open source projects, including Rex Ray and many others. Okay, Brad and John, thanks for coming on the program again, and we'll be back with more coverage here from the VTUG, the EMC level of Gillette Stadium. Thanks for watching.