 Live from Austin, Texas, it's theCUBE. Covering OpenStack Summit 2016, brought to you by the OpenStack Foundation and headline sponsors Red Hat and Cisco. Now here are your hosts, Stu Miniman and Brian Gracely. 2016 in Austin, Texas, I'm Stu Miniman here with Brian Gracely. It's our fourth year doing theCUBE at the show, the sixth year of the event and 7,500 people have gathered here. Happy to welcome back to the program. Randy Bias is currently the VP of Technology at EMC's Emerging Technology Division. Randy, welcome back to the program. Thanks, always great to be here. All right, so we've got EMC world coming up next week and I'm sure there's stuff leading up to that. But first of all, we're doing a lot of reflection about OpenStack. You've been involved since the early days. Day one. Day one, were you at the first event? One of the 75? I was not one of the 75, but I was part of the original launch. Yeah, you were definitely there. One of the OGs as it were of the event. So give us your thought as to where OpenStack sits. I mean, I think back in the early days, there was only one true cloud and it was the public cloud. And where does OpenStack fit in the environment? What do you think about where it's come? Does it exceed expectations? Is it where you thought it would be? It's not where I had hoped it would be, but it's where I predicted it would be, which was basically dominant in private cloud. I mean, I've been saying this for, I mean, pretty much the beginning. I basically said, OpenStack's never going to be relevant in public cloud and I distinguished between public cloud and hosted private cloud. OpenStack's got a great future in hosted private clouds, I think. But in terms of public cloud, big multi-tenant systems like Amazon and Google, I just don't think anybody's going to take OpenStack and make it relevant there. And they haven't so far. And you look at the speed and velocity of Amazon and Google and Azure and add that all up and OpenStack's not even in the same zip code. But on private cloud, it's a completely different story. I mean, you see this thing happen with Volkswagen and you just see this almost huge shift over to people really getting cloud native applications, getting that they've got to modernize their data centers and then believing that OpenStack is sort of one of the foundational components to making that happen. And so that's great. The only thing that I wish is that OpenStack had more velocity and more deployments and more customers. And I think they're the community and the vendors to a certain degree have shot themselves in the foot by being too inclusive and having too much diversity and every OpenStack deployment still is largely a snowflake, which I said what, four years ago? One of the things we've been talking about is where OpenStack fits in that kind of multi-cloud world. As I just said, OpenStack's really private cloud solution hosted whatever flavor you want in there. But we're starting to see discussions of how it interconnects with the public clouds. We had Microsoft on this week, we had Google on this week, talking about how if I use containers, Kubernetes, things like that, and have OpenStack as an application, that it can have interactions and live in that both public cloud and on-prem solution. What's your thoughts on those pieces? Well, I mean, if you look at the conversation around OpenStack being connected to the major public clouds, the conversations being driven from the major public clouds are iconic class and renegades like me. The majority of the OpenStack community still holds, as far as I can tell, it's sort of this sort of weird idea that OpenStack will become this sort of like, you know, ubiquitous like platform that's totally interoperable, somewhat like Linux across all data centers everywhere, including the public cloud, and that, you know, that'll magically come together. Except it hasn't magically come together yet. It's still not magically coming together. And even though we make incremental steps like deaf poor towards it, we just really don't get there. And so I'm a little frustrated, as you can tell, that we're not making those steps. But I'm really gratified to see that many of the public cloud providers see OpenStack's importance in the private cloud, arena, and they're reaching out the other direction. And actually, you know, some of the things that we're looking at doing is interconnecting Google compute platform to Keystone and potentially doing some interesting things there. And my team is going to continue to run down this road with whatever resources I have to basically make OpenStack interoperable public cloud. Regardless of what the rest of the community does. So, obviously, what's the, the big theme this week is around telco and NFV. We're seeing some large carriers, you know, talk about deploying very, very large OpenStack thing. Help us rationalize that. If OpenStack doesn't make sense in the public cloud, which is a scale game, it's an operations game, how does it then make sense for these humongous carriers supporting, you know, potentially billions of mobile phones or IoT devices? Help us rationalize that a little bit. Yeah, so there's actually several different things you're asking there, and I have to break them apart. So the first thing is, does OpenStack scale? And I think OpenStack can scale just fine, depending on which components you're using. I mean, our POC deployment for Walmart was 25 racks. You know, we didn't have any problems. We ran Black Friday on it twice. I mean, you know, it scales, right? Is it gonna be Amazon scale? I don't think so. Do carriers need to be Amazon scale? I don't think so. Kind of the next question you're asking is, you know, what's going on with OpenStack and NFV? I think that's a little bit overblown. I apologize to the foundation in advance, but I'm not on the board, so I don't have to pull my punches anymore. But the reality is, is that when I go out and talk to carriers, the whole challenge with NFV is that you have to try to take these network functions and turn them into cloud-native applications. And the people who provide them these, the vendors, they're nowhere near doing that across the entire stack. So NFV is a great use case, is clearly important. The carriers care about it a lot. They're going to drive it to conclusion in a way that we didn't get to conclusion with SDN. But there's going to be a lot of drag because they ought to go out and convince every vendor to refactor their apps. And then third is, you know, how does OpenStack do with the carriers? And the reality is that they've tried public cloud with OpenStack and largely failed. And that's part of why they've gone over to OpenStack as private underpinning NFV because they can see they can get a tremendous amount of value of refactoring and modernizing their data centers and using OpenStack as an underpinning for that. Yeah. So you talked a little bit, you've always been a little bit of the rebel here at OpenStack, you know, kind of pushing the boundary, trying to push them in directions maybe that the mainstream's not going, you know, you used to give a presentation that was, you know, VCE and the VBlock were sort of the old way of doing things. Here's the new way. Yep. You're now at EMC, you know, as part of the cloud scaling got acquired probably about a year and a half ago, two years ago. That's right. What's going on at EMC around OpenSource, around OpenStack? Like how are you infusing that DNA into that culture which, you know, three years ago you would have said, you know, it doesn't make sense here at all. Yeah, it's a big question to answer. So I'll try to be succinct, it's going to be hard. So first, when we were doing the transaction with cloud scaling, EMC came in with like two weeks to spare out of nowhere. And I was like, we're going to use these guys as a dark horse, you know, against the other bidder. And what actually went to happening is I came around very quickly to realizing that at the executive level of the EMC, there was a shared vision around what cloud scaling was trying to do too, which was to drive an appliance-like experience into the customers so that they could deploy scale technologies like in Amazon does without having to have all the Amazon skills and labors and capabilities. And that was really at the heart of what we were trying to do. We're really trying to have almost like a software-like abstraction of the Vblock. And so I saw that as a shared vision. So I was actually happy to go over to EMC. And then what's happened is the cloud scaling team amongst others has been on this journey to deliver sort of a new capability which we reveal next week at EMC World, which is, you know, sort of something I can't talk about in detail right now, but you can think about it as sort of being what would a Vblock look like if maybe Amazon created it and delivered it to a typical enterprise that can't consume Amazon's normal internal software. They need to be kind of like, you know, really kind of regimented and enterprise-grade kind of product. And so that comes to fruition finally. And so even within EMC, you see sort of a clear bifurcation between the technologies and products that service sort of the second platform or legacy applications and those that service kind of this next generation. So Randy, you wrote a post leading up to the event and there's a lot of pieces more than we'll be able to get into, but you talked about Amazon, the difference between where you build function into the hardware versus the software. I know when I left EMC, I got kind of a retraining because I lived in the enterprise world, everything was HA, you harden it, you make sure that the hardware, hardware, hardware works as much as possible and as the line we always use is hardware will eventually fail and software will eventually work. So maybe you can share with us a little bit about your thoughts and how's EMC's views changing on kind of the role of the hardware versus software? Yeah, so EMC at the executive level understands that it's a business that's really transforming from being hardware-centric to software-centric and that's like a big, many year change, requires a lot of change to the culture and DNA. But if you look at some of the teams that are now the team that builds our object storage product which is software-only, stripped object store, those guys all came out of Microsoft Azure. So there's a lot more of that DNA in EMC these days, scale out folks, people who worked at Apple and built the infrastructure under Siri, people like myself who worked in OpenSource, there's a lot more of that kind of DNA being injected into EMC and you're seeing it in the software-only products. And if you actually look at where all the heat is, where our revenue growth is, it's all in the software-only products, ECS, Scale IO and so on, 200%, 300% growth year over year. And so that transformation's happening, the battleship is turning, I'm always wondering are we going to make the turn, are we going to make the turn, the battleship's going to break in half, like I don't know. And then just for giggle-sake and throwing a wrench in the works, we have the deal acquisition happening and then there's another thing, like I don't know, how's that going to affect the battleship turning? But it is turning. Yeah, you talked about in that blog post running Oracle Rack on top of OpenStack which years ago would have seemed like you'd have gotten shouted out of the room by the- Crazy talk. Yeah, the cloud-native sort of folks. Are you hearing that from customers? Is that what they want to do? I mean, do they want to move the old to the new or where's that going? On the average enterprise's journey, where they wind up is really understanding that right now it's best to keep the, not cross the streams, right? Keep the second platform or third platform stuff separate. But in the course of that journey, typically in the middle of it, you see some wishful thinking, you see some desires for unicorns and ferries and things like that. I'm aware of one deal in Russia that is going down that we're not really a part of, where the customer wants OpenStack to be overloaded to support containers on risk-based platforms and literally they're going to add so much junk to OpenStack to support everything that exists in their current data centers that I just look at that and I'm like, that's going to be a fun project. I'm really glad I'm not part of it because I don't think it's going to end well. Yeah, I mean, it's always interesting. Customers, they get that blue sky vision of where things are going. Yeah, you have to be a little bit more pragmatic as to where things are. So, where are we from the maturity? I think most people agree, kind of base functionalities in pretty good shape. Do you agree? What do you think OpenStack needs to, what needs to happen moving forward and what are the boundaries? How far will OpenStack grow? The base functionality is strong. It depends on which components you're using as it always has. I think that it's, I still think it's best to consume for a vendor. The folks I see do DIY, just really invest incredible amount of engineering resources into effectively building their own distribution, which doesn't make a lot of sense. The folks I've seen be very successful do less experimentation. I hate to say it, but doing SDN and SDS and putting, doing experiments in your entire stack, you're just stacking up risk after risk after risk. So I encourage people, if they're going to go do an initial POC with OpenStack, use networking you already know, or use storage you already know and have on hand. Don't do one experiment in the stack. Get comfortable, learn more before you go crazy. So I think when you keep that all in mind, that it's relatively mature for just the core set of services. One of the things that I think is interesting is that some of the technologies like ScaleIA are getting to the point where they can create the ability to take on some of these P2 production workloads. I don't think everything's there, but I think some pieces are there. And that thing that the unicorn that people are asking for, I think we'll never get to the unicorn, but I think there's a possibility to get some elements to the unicorn. One of the examples, or things that I look at as a guidepost is Amazon Web Services. You know, when they started off, there was no VPC, there was no load balancing. There was a bunch of enterprise-y services, particularly VPC that just didn't exist, right? It was all layer three networking. You couldn't have a VLAN or network segmentation. And you know, I really feel like EBS and VPC are a direct response of Amazon Web Services to sort of what the enterprise needs, particularly for P2 workloads. You can't put all P2 workloads there, but you can put some. And so I think that's really important. So Randy, we haven't had much discussion of VMware this week. How do they fit into this whole discussion? I feel corned, Stu. I'm not sure what to say. VMware's part of the federation. They love me. Every time I come visit them, they roll out the red carpet. Randy, come on in. Or maybe the opposite of that. But I don't know what to say there. Without getting myself in trouble, I will say that VMware is charting its own course for better or worse. So one of the big themes, last year the big theme was open stacks and integration engine. This year, Jonathan, the keynotes, were a lot of talking about, hey, here's how we're going to collaborate and partner with PaaS layers, container layers. Where do you see open stack in a couple of years? I mean, they've kind of gone down the path of they wanted to be more than I as. A lot of those projects have kind of been soft. You've had other foundation startup. Where does all this stuff kind of mash together? If you're a customer, do I have to learn Kubernetes by itself and open stack by itself? I mean, is that a good thing? Is that, do you see these things sort of mashing together? Where's all this going? I mean, I think it's an excellent question, Brian, and I'm not sure I have a great answer. What I'd like to see is I'd like to see, gonna get on my hobby horse here, I'd like to see what I've been promoting at the board for the last couple of years, which is that open stack stops of viewing itself as a single monolithic thing and it gets broken down into groups or subgroups of projects. And that is structured a bit more like the Apache Software Foundation. Each of the projects can actually operate somewhat independently. And there are sort of half steps to doing that, like Big 10, but there's still all these attempts to put together sort of a full release that's been tested together. And I think that we should just admit that that should be what the vendors should do. Like I would rather see the Kubernetes guys go and say, hey, let's take these two or three components of open stack and roll them up underneath our default Kubernetes deployment and package that up. And then there'd be an open stack certification and interoperability test for open stack past platform. And I would like to see a bunch of different flavors of open stack that were designed for NFV, that were designed for enterprise private cloud, for public cloud compatibility, and for us to have different certifications around that because right now you still cannot draw a tight circle around open stack and say, this is it. And that's why Def Core is stalled and has been unable to sort of get, people agree about what's required for interoperability. So rather than trying to say there's one thing because we can't agree on one thing, let's admit that there are several different flavors that are sort of designed for different workloads or verticals. And then basically assume that each of those will be a subset of components of open stack. And then the vendors will take those to market and have a way to test for interoperability. I think that would be a much better model. But so far that's fallen on deaf ears. Yeah. And we've heard different opinions on that. I know Red Hat told us, hey, look, we don't want to have variation between the NFV version, the enterprise version, the cloud version, obviously for their testing reasons. But yeah, it makes sense. Otherwise you're talking about saying, get away from monolithic applications and we're going to deliver you sort of a monolithic delivered package type of thing. So I can see both sides of it. Absolutely. If one size fits all, public cloud wins. Period. If one size does not fit all, then private cloud has a place to play and private clouds need to be flavored for their particular workloads. I just don't see how you can see it any other way. Interesting. So Randy, outside of OpenStack, what's got you excited these days? Love containers. I think security around containers is deeply misunderstood and that people don't understand that containers have an opportunity to basically destroy the type one hypervisor. I'm really, really excited about some of the new security startups that I see doing stuff around containers. The other thing that I'm excited about is I feel that we're about to hit a pressure point that will be a real sort of change in the security community generally, even beyond containers, which is that security is still like the big moat and castles, like build it and try to surround everything, but the environments are getting more dynamic, more agile, more nimble, and the demand for the business is for that. And so I think that the pressure is going to increase in such a way that the security industry is going to get real disruption because it's been the stalwart resistant, resistor to disruption around the cloud. And I'd like to see that happen because I think that whole industry needs to be really shook up. And so I think that's a real opportunity. And then the last thing is that the culture change within the enterprise is starting to happen. I mean, I don't know about you guys, but what I've seen is that as people struggle to adopt open stack or containers or these things, they've started to realize they need to get back to first principles, which is how do we deal with the problem between the line of business and centralized IT? And then they start to latch on to DevOps, which starts to get them into the process of culture change and thinking about how to make those teams work together and align the time to market with the reducing risk at the same time. And so I'm really excited that, you know, that seems like a huge wake-up call that I just didn't see coming. Like I didn't believe the enterprise would see that that need to happen and they are seeing it. And so I think that that's winds up being the thing, the seed that we can all grow within the enterprise to basically get them all transformed. Definitely, I totally agree. There's so much change going on there. I'm curious, you know, now your viewpoint, you know, you're in, you know, biggest storage company out there. Yeah, you know, what does that mean for some of the individual silos? Will we finally get beyond some of the silos? And, you know, what's that going to mean to kind of the application roles? The infrastructure silos, the team silos of the application silos. Sure. All of the above. Take your pick. Yeah, I mean, the silos have to go away in order to get operational scale and efficiency in order to have speed and time to market. One of the things that I've seen that's really funny is though, as I see everybody trying to get to the unicorn, they're like, we want one silo for everything. And so what I've been trying to tell customers is like, look, you know, like they'll latch on to stuff and they'll like file, block an object all in one system. It's going to be perfect and great. And then they get to the reality, which is that, you know, you just can't build one system that does all things. And, but the thing is that there's a fatigue over having hundreds or thousands of silos. And that's real. And that has to change because, you know, that's what the web scale guys have taught us. However, the web scale guys don't have one silo. They just have a very small number of silos. So if you're going from hundreds or thousands to three silos, you got one block, one file, one object system, you'll have more in that. But as a, for example, that's a huge amount of operational efficiency that you're going to get just massive, 70, 80, 90%. And so that we need to get the focus away from having the magic pixie dust we put in the data center that solves all problems and just realize that there's some very pragmatic things that we can do to reduce the operational problems by having a very smaller number of silos, right? It's not bad to have silos. It's bad to have thousands of silos. So, you know, sort of last word on this. When you first got started, it was pets and cattle. That sort of, you know, got your name out there, became the thing. One of the things. One of the things, obviously. What's the next, what's your next sort of phrase that you're using to get people to go here, think about this new thing. Do you have a thing that's going to catch on yet? I can't predict that stuff. I had no idea that pets versus cattle was going to take off. That was that, you know, it floored me. Like I just wasn't expecting it, right? I was looking for a way to synthesize things down. And so that's what I'm always trying to do is like, how do I get this to be bite-sized so that like somebody can consume it? So, I can't really predict what it'll be. Rancher's dilemma sound pretty good though. Kind of rings, doesn't it? It's got a little flow, but I don't, you know, we'll see what happens. Great, all right. Well, Randy Bias, always great to catch up with you. Your unfiltered view of everything going on here in the industry. We'll be right back. Getting close to the end of day two of three of theCUBE's coverage, OpenStack 2016. Thanks for watching.