 The Cube at OpenStack Summit Atlanta 2014 is brought to you by Brocade. Say goodbye to the status quo and hello to Brocade. And Red Hat. Here are your hosts, John Furrier and Stu Miniman. Hey, welcome back everyone here at Live in Atlanta for OpenStack Summit. This is The Cube, our flagship program. We go out to the events, and instruct the students from the noise. I'm John Furrier, the founder of SiliconANGLE. I'm joined by my co-host, Stu Miniman, analysts at Wikibon.org, and our next guest is Cube alum and frequent guest. We've done a couple OpenStack events at San Francisco. Randy Byas, the CEO of Cloud Scaling. Great to see you again, welcome back. Thanks for having me, John. So OpenStack's happening. It's the real deal. Things are coming together. Big board meeting, technical committee last night. You can't really talk about it. We'll try to poke at you. There's always been kind of the API debate. We had the thing on Neutron in San Francisco. Real active community. So I really want to get your perspective on a couple things. One is the size is ridiculous. It's 4,200 people, almost double in size from Portland. Clearly the developer community is galvanized around OpenStack. It is the future of Cloud here, and the enterprise and service providers. But they're still being built up. The foundation's being set with hearing about the trough of this losement, ramping up into reality now. And so I got to get your take on what's happening. What do you see as the key momentum shifts up into the right here for OpenStack? Yeah, so I think when we are kind of talking to our customers about it, we look at OpenStack as having evolved from kind of a science project into a mature open source project that has a number of different vendors that are building products around it, and that the enterprise is going to really be successful at consuming by going to those vendors, trusted vendors, and basically working with them to get an OpenStack deployment up and running. So I've been here since the very beginning, even before OpenStack, working on Cloud, and we've seen just a lot of failures, right? And unless you're prepared to really invest into doing it yourself, it doesn't really make sense. The way I like to think about it is, OpenStack isn't like downloading MySQL and turning it on, turning on a database with a couple of commands. It's more like downloading the Linux kernel and building your own Linux distribution. That doesn't really make sense for most businesses to be in that particular business. You're really referring to a product versus platform. Product versus platform, that's right. And challenges there, people have learned and avoid now with the kind of that position. As a platform, what should folks be thinking about in terms of embracing OpenStack, whether it's an enterprise CIO or a service provider? Well, the big thing that we see is that if you comment this and you're trying to get sort of a cheaper VMware, you're going to run into a lot of problems or run into a lot of roadblocks, because the reality is that something like VMware is doing a pretty good job of what it does inside the enterprise today, which is supporting a number of legacy applications that were designed in kind of the pre DevOps era. If you look at where OpenStack really shines, it's for the same kinds of workloads that exist on Amazon Web Services and Google Cloud Platform today, which is applications that have been designed in that DevOps model and are actually architected to route around failures. So they notice when there's failures in the underlying infrastructure, and that's why somebody like Netflix can have a four, five, nine app running on a two and a half nine infrastructure. And we think that enterprises, if they really want to see success, are going to actually take something like OpenStack and use it to support these next generation applications, which in the next five years there will be as many next gen apps as there are legacy applications today. So Randy, if you think about the enterprise, they're used to buying solutions. Most of them aren't good at taking the pieces and rolling it up together because it takes too much time and takes too much expertise. And most IT staffs have kind of flat to negative budgets and they're already overworked. We spend, they spend 70, 80% of their budgets on keeping the lights on. If we look at OpenStack, there's dozens of distributions out there. I think you called it in one of your articles that it's the open snowflake that we have out there. So if we're going to have the solution be repeatable and delivered broadly, what do you think we really need to see? Well, first of all, I just want to make sure but it's clear that like OpenSnowflake applies to things besides OpenStack. You know, I remember talking to the Stradius guys before they were acquired by Dell and they were having the same problems supporting multiple different cloud stack installations because different architecture choices had been chosen when it was deployed. The problem isn't really about the software. The problem is about the architecture. When you go and you take a piece of open source software that's this complicated, it distributes a software system that's designed to build a private cloud at scale and you have to make a certain number of design decisions as you're going through the deployment process because it is a platform, you can use it for a number of different kinds of things. And so what the networking choices you make, the storage choices you make, the hypervisor choices you make, all of the choices that you make in aggregate create an architecture and if you made significantly different choices from the guy next door to you, it doesn't matter that your software is the same. Conversely, if you have completely different software and you make very similar choices, you wind up with systems that are almost identical. You look at Amazon Web Services and Google Cloud Platform today and they're semantically, architecturally and behaviorally almost identical. It's like looking at the same system but they have a different API and they're written with completely different software. It's just that they largely made all the same design choices. And so open stack as a platform, it has no reference architecture. If you just go take it and start to do your own thing, you're gonna back into making a reference architecture. That's why somebody like Cloud Scaling made a decision that the reference architecture is Amazon and Google and just to use that. And what we'll see is we'll see a number of different flavors like we have in the Linux distribution world that are designed to solve different problems for the enterprise, for HPC, and so on. And then enterprises will adopt one of those flavors and within the flavor there will be high levels of interoperability. All right, so Randy, you said that if people are just thinking that they're, say, gonna do something like VMware but it's cheaper. That's really not what open stack is for. Where do you really see that the big opportunity is the customer's gonna save money? Is it the agility? Where is the big opportunity? It's all about agility. Saving money is a side effect, right? I mean, if you're coming into this and you're trying to get a cheaper VMware, then you're still stuck on the server consolidation bandwagon, right? You're still trying to reduce the cost of your data center instead of generating new opportunity. I mean, we've moved into a world now. I think it was Forrester, their analyst, Craig Leclerc, who said that basically over the past 10 years, 70% of the Fortune 1000 have turned over because they were unable to adapt to change. And that is a sign of the times, the era where we have these big, gigantic companies that are incumbents that can sort of sit on their structural advantage like as an oil company or steel company or something like that is gone. Almost all companies today are starting to realize no matter what industry they're in, that they're a technology business first and they're in their vertical second. And they have to use technology to get strategic competitive advantage and that if they don't do that, then their competitors will and they'll be taken out to the woodshed. And so I think what this is about is understanding that OpenStack's the foundation for that next generation modernized software-defined data center that will allow them to drive new value faster into the business and stay relevant. Do you see OpenStack as an extension of kind of the software-defined meme that's going around then? No, I actually hate to use the software-defined meme. It's just a fancy way to talk about cloud when you're one of the enterprise vendors, three-letter ticker symbol who feels threatened by cloud, right? So it's not software-defined, whatever, it's the same thing as cloud. Really, containers are all the rage now with Docker. How do you see that impacting OpenStack or at the past layer, certainly Cloud Foundry versus Pivotal, you got to see Red Hat promoting it heavily. HP didn't really talk about containers in their announcement. Do you see that as a big part of this debate or conversation? Yeah, so in full disclosure, I'm an advisor to Docker, so I want to make sure I've got some skin in the game on this, but one of the things that we saw really happen over the past 10 years is just this big shift away from virtualization, especially in the past three to four years. And the reason is that if you look at next generation applications, cloud-native applications, they're actually designed in sort of a scale-out, multi-core way. A lot of virtualization was built around the era where the applications, say like SharePoint, was single-core, single-threaded application that could only use four gigs of RAM, sticking on a gigantic box, and suddenly there's all kinds of unused space. And that's sort of what drove the server consolidation meme. But we're now entering into an era where you've got workloads like Hadoop, scale-out, Java applications, Python, Ruby, all of these things that can actually use up 16 cores, 128 gigs of RAM on a single box without breaking a sweat. So virtualization and that kind of situation actually adds a lot of overhead that doesn't make sense. And I don't know if people saw it, but there was actually a blog post that went out in the OpenStack ecosystem a couple of days ago, I think Friday. I can't recall the name. And the gentleman was doing performance comparison between Docker and KBM. And in some cases it's incredibly dramatic, especially for like database workloads. So I think there's a lot of good reasons to go to containers or to go to bare metal on demand and see kind of those as sort of alternatives to virtualization. Still always going to have virtualization, but you're also going to have containers and bare metal as alternatives and you're going to mix and match depending on the workload. So you've got the big guys coming in, obviously IBM and HP, big commitments in terms of dollars, in terms of ecosystem support, in terms of code, red hat out there. What's your take on the big guys in terms of the balance between organic community growth and the big leaders coming in into the sandbox? I mean, it's a tricky question. You know, on one hand the big guys are great because they validate the space. They really help the enterprises understand that OpenStack matters and it's really going to be their future. It's the new Linux of their sort of, you know, cloud operating system they're going to put in their new modernized data center. And so that really helps. On the other hand, you know, what we've seen today is that the vast majority of large enterprises aren't actually innovators. I mean, the whole reason that startups exist and are very successful is that startups tend to be the place where the innovation happens and then they're acquired by large players who then roll up their offerings into it. So if you look at something, you know, just to call somebody out, you look at something like vCloud Director, which VMware built internally, you know, it slipped two times. It was nine months late each time when it came out, nobody was happy with it, even VMware. Now it's deprecated and they're basically not pushing it anymore. And so, you know, that's a very common kind of thing to happen. And so I think that the big guys are doing a lot to validate the markets based and improve the code quality and to, you know, make a platform for OpenStack better. But, you know, I'm not seeing a lot of innovation from them around the product. So for example, you know, if you look at OpenStack today, a key shortcoming is the networking stack, which, you know, in my opinion, is unacceptable for any enterprise to deploy into the default networking options. Yet, you know, none of the big guys who've been in there contributing, you know, large amounts of code over the last 18 months has done anything to fundamentally fix it. Which means that I can go in and I can basically kick those guys out of, you know, big deals fairly easily because I've got a networking stack that works. Randy, can you help us unpack the state of OpenStack in the hybrid cloud environment? If I look at some of the players here, you know, IBM's got SoftLayer, HP made a big announcement last week and they've got their HP public cloud. You know, Rackspace can either host your environment or they can help you with the private. So, you know, most of the big guys have both, you know, an on-premise solution that they're helping build as well as kind of some service provider or public offering. You know, what's your take on that and you know, where OpenStack fits in that kind of hybrid world? Yeah, so, we see hybrid as being the future. When we look at pretty much everybody that we talk to in the enterprise, even financial services, everything one of those folks is really thinking about hybrid cloud, even if it's in the future and they want a hybrid cloud solution. And the problem is that, you know, when people say they want hybrid cloud, what they really mean is they want application portability. If you boil it all the way down, you want to be able to define and deploy and manage application on multiple environments and have it be and act and look the same. Now, this sort of gets to the heart of the problem which is that most of the people who are pushing hybrid cloud today haven't come at it with the same kind of nuanced kind of thinking that we have. They're just seeing hybrid cloud and they're just mashing any two clouds together. And if you don't have matching performance, if you don't have matching SLAs, if you don't have a matching architecture, you get into real problems. So for example, you know, Rackspace really pushes their hybrid cloud capabilities, but their public cloud has virtual machines that default to having two network interfaces on them and their private cloud operating system has a virtual machine that defaults to having one nick on it. One private cloud uses Zen or the public cloud uses Zen, the private cloud uses KBM and so on. And there's just a ton of fundamental differences. The reference architectures for their two clouds are hybridizing aren't the same. And when we've gone in and talked to customers and pointed that out, they said, yeah, that creates a problem. I can't get application portability. And I think that's really sort of what is happening here is that a lot of the big guys haven't figured out what does hybrid cloud mean if you really boil it down to the technical realities? Yeah, do you think we'll get there, Randy? Wikibon did a survey a year ago and we asked customers, do you want that same environment in your private environment as your public? And they said, yes. And then when you poked, of course, what do you have? Say what's your public environment? AWS, and what's your private environment? It's VMware. So they talk about that's the nirvana state that we want to get to, but the reality on the ground is people are moving in a lot of different directions. We're hopefully breaking down some of the old silos on specific bespoke architectures for specific applications, but there's a lot of work to get IT to take their application portfolio and create some kind of standardization. Yeah, well, that comes down to the fundamental disconnect and misunderstanding about what cloud is and where the value is, right? If you have a VMware environment, that's, I mean, I think it's a stretch to call that a private cloud, no offense to VMware. I think that's a virtualization platform, which could be an element of a private cloud, but it's a single, small individual element. So cloud is not about virtual machines. I mean, we've established this years and years and years ago. Cloud is about bringing the new IT paradigm that the web scale folks like Google and Amazon and Twitter have created and bringing it down into the enterprise that are using the same techniques like something like Hadoop is clearly a cloud technology and it's just a recreation of Google's MapReduce technology with its same reference architecture. That is core cloud technology. And so, if you were trying to take your virtualization platform and hybridize it with Amazon, you got a whole ton of problems. First of all, the kinds of applications that go in those two systems are fundamentally different, right? One's designed around DevOps style and routes around failure and the other's a legacy application that assumes that the infrastructure is always there, highly available and is going to prop it up. And so, you just can't come at it from that point of view. So I think we're at today is that people are starting to understand that they need a net new private cloud system that's designed and looks like Amazon web services and certainly that's where we're seeing success and then that system can more easily be hybridized with an Amazon or Google cloud platform. Yeah, so that makes sense. So I got to ask you, what are some of the workloads that are popular running on the clouds right now in this net new cloud? Because obviously Amazon's pretty clear in the enterprise in those guys. They know what they run on there. It's the shadow IT market. They're trying to be more enterprise. But what are some of the most popular workloads running on the clouds now? And what's emerging? So any web application basically is a great fit. The way I like to think about it is that if you got to share nothing architecture that was designed kind of anytime after 1999, that that's almost always a good fit and maybe with a little bit of work on the cloud. Basically, if I had to boil it down the key to finding characteristic of all of these systems is that the application is taking responsibility for itself. It deploys itself onto the cloud. It scales itself up and down. It monitors its own progress. It looks for failures and routes around them. The other sort of emerging places are like social and mobile. A lot of people are doing social analytics and things like this. Sometimes you're using big data platforms for that. And the other is mobile, right? There's sort of this API in the sky kind of effect happening now. You want to go and you want to engage with a customer based on Facebook or iOS or Android, then you've got to develop an application. You've got to push it onto those platforms. And then almost always that application has sort of a back end that's in the cloud that it needs to talk to in order to push data backwards and forwards and do correlation and things like this. How are customers choosing cloud providers? I mean, is there a groove swing at this point? Is there key things that you see in terms of it? Is it simply I want to consult with it? What are you seeing in terms of how they're making decisions? Do you mean a public cloud provider or both? Or it could be the, like you said, the net new. I want to tinker with it. I want a little bit of Amazon. I want a hybrid. I mean, the hybrids are the action. So there's no doubt about that. I agree with you on that, but I'm an enterprise and maybe there's some shadow IT. And let's just take this use case. How do I decide who to go with? I have legacy stuff. I got a data center. I see the economics. I see the tool and opportunities. I want to go open stack or I want to go build some Linux on there. What do I do? How do I decide? So it's a little bit of a broad question, but basically I've had to boil it down. I sort of see, we sort of see three entry points typically with open stack and our product open cloud system. And the three entry points are people who are focused on, you know, driving agility into their businesses. They're adopting DevOps. They're sort of early adopters. An example of this would be somebody like Hyatt, for example, who's got a whole big initiative in there. Around building new cognitive applications and moving to a DevOps model. Ancestry.com is another one as well. There's a whole bunch of these enterprises that are emerging. The second bucket is people who want to build hybrid clouds. And what they're doing there a lot of the time is it's like Zynga kind of use case. They're trying to cost optimization. They've got steady state workloads on the Amazon. They've run the math and they know that they can actually run them on their own system for, you know, much less expensive price. And then of course they want to hybridize with the existing public cloud because they're going to do all the new product launches there and so on. So what do you tell the CIO out there in your own words, OpenStack today? What is, what's new about it? What's different about it? Obviously there's some investment, either tire kicking and or participation. Or just I'm jumping in now, what do you advise those guys now when you have the elevator chat or the meeting, the first meetings? What do you advise them around? What's happening today around OpenStack? Well, first thing is I just try to make sure they understand the difference between investing into a virtualization platform that won't get them to that modernized data center versus going with a true elastic cloud platform that looks like Amazon that will get them there that's focused on business agility. And once I've gotten them to that point, then it's really a discussion about, look, you need a trusted partner who's actually got experience building private clouds, understands what these new application workloads look like on top of them and can work with you in a consultative manner to make sure that your teams are successful in not only bringing the cloud up but adopting it and using it to drive new value into the business. All right, so Randy, you've been coming to the show, Ice House is out now. Is OpenStack ready to take the next step? When you come back next year or even let's look out six months to Paris, what are some of the things we should be looking for to see if OpenStack is really starting to break into the mainstream? So, the OpenStack platform's been ready for production, most of the components that matter since probably full sometime frame, maybe grizzly depending on who you talk to. So, we're already there. The problem is you just have to consume it via a vendor who's got a baked product, right? I mean, you can't consume it directly. And then I think what we're looking for as we are moving forward into the future is, and this is the place I think where I maybe diverged from some of the others is that I wanna actually see more of a, I hate to say fragmentation, but at least the emergence of OpenStack for different purposes. I think right now what happens is there's this sort of misguided belief by both the users and the operators and developers and even the members of the board that OpenStack is a monolithic thing, right? That you take OpenStack and you use it for in the same way everywhere. And I actually don't think that's accurate. If it's like the Linux kernel, then we know that the Linux kernel actually has a number of reference architectures. Look at Android. Android is the handset operating system that Google creates. That's based on a Linux kernel. It has none of the normal Linux user land. At the same time, we've got the Linux kernel that runs on a Cray supercomputer and those are two very different kinds of Linux. So we need to see the emergence of OpenStack for enterprise, OpenStack for HPC, OpenStack for web scale businesses. And then as we see those flavors of OpenStack emerge, we'll see kind of the early formation of kind of like the Red Hat versus Ubuntu versus you know, Monovista type kind of flavors inside of Linux. So on the future of OpenStack, obviously I know you can't talk about some of the minute stuff, but how do you feel the governance is going? Obviously, there's people talking about that's a key part of it. It feels good. I'm not really seeing anything negative at this point. Still a lot of rah-rah, some good signals, a lot of good love coming out towards some tweets last night, although nothing specific. I know you can't share it, but give folks a taste of the feeling around the governance right now. Yeah, so, you know, governing something that has grown as fast as OpenStack and as big as OpenStack is very, very challenging. We're into, you know, really a lot of uncharted territory. And, you know, it's taking a while to start to, you know, get some of this stuff together. And, you know, there's more than two governing bodies. I think if you bake it all down, there's something like five governing bodies. You've got the board, you've got the technical community, you've got the foundation itself, you've got the user committee, you've got the operator folks who are doing kind of their own summits and talking a lot. And so there's a lot of people who are deeply involved with trying to figure out the direction forward. And I think that the key challenge, as it is with any kind of large development project, it comes down to communications. And so we're starting to see some breakthroughs. I know it took a while, but, you know, yesterday was the first time that we had the first kind of joint board and technical committee meeting. And I think there's a lot of reasons and a lot of back history why it took us so long to get there, but, you know, I think the net of it was that people were able to air a lot of concerns and issues and thoughts and feelings about the way that we could communicate better together. And even that made it, through that process, it became clear that we weren't actually very far apart, both the technical and product folks. And we made some very key decisions about how we're going to do specific governance around the product roadmap and the product management going into the future. And I think that that's going to really smooth things out a lot. Who's leading right now in your mind? In the OpenStack community, who can you point to in saying that horse is running well, those guys are doing great? In what regard? In both technology, community support, community participation, really pumping on, the ideal citizen, if you will, of the OpenStack community. It's hard for me to call anybody out like that. I'm not sure that there is sort of a shining light there. Okay, well, we're excited to have you on theCUBE. We'll give you the final word. Obviously, we wouldn't be here if it wasn't for Red Hat and Brocade. Want to give a shout out to those guys for supporting theCUBE this year. Thanks to those guys. Certainly, winners in our book right this week. Final word, share the folks out there in your own words. Why this year is so important? What is going on this year around OpenStack that really is going to tip the scales for the kind of traction that people want to see? Oh, well, we've seen the adoption really start to shift this year. I think I've got some stats in my upcoming slides on Wednesday, but we're a small startup. We're still in the process of getting traction, but if you gauge anything by us, then I think we're doing pretty well. We're supporting over 10 clouds. I'm not going to say exactly how many, but we're supporting over 10 clouds that are in production, that are running production workloads, and those 10 clouds constitute 8,000 cores, almost 600 servers, and eight and a half petabytes of storage. So that's a pretty significant amount of stuff to have deployed as a small startup. I'm sure some of the other folks, they're bigger than us, they're doing a little bit more. But the point is that if you see that with just us as a little startup, then you add up the entire OpenStack ecosystem and assume there's similar unparity with us, then the overall size of adoption and production deployments is actually looking very, very good. Very nearby, it's always a pleasure to see you in theCUBE here live in Atlanta for the OpenStack Summit. Great stuff happening, always good to see these startups and the emerging players. There's a great balance right now between the organic growth and then the big guys coming in and really putting their feet in, getting their feet wet, playing nice in the sandbox, certainly bringing some cash to the table. As they say, there's a lot of dough coming in from HP and IBM, putting a billion dollars in it. Like I said, the joke in the consumer web business is 10 million users is now the one, is the new one million. So 10 billion is the new one million. So IBM HP, come on, more than a billion, come on, put up 10 billion then it will be. That's just table stakes. That's just table stakes, a billion dollars is nothing in these days. Anyway, this is theCUBE, right back with more coverage after this short break.