 Live from the Computer History Museum in the heart of Silicon Valley. It's theCUBE, covering OpenStack Silicon Valley 2016. Brought to you by Morantis. Now, here are your hosts, John Furrier and Lisa Martin. Hey, welcome back everyone. We are live in Silicon Valley for the OpenStack Silicon Valley or OpenStack SV as it's called, the hashtag OpenStack SV or hashtag O-S-S-V-16. This is theCUBE Silicon Angles flagship program. We go out to the events and extract the signal from the noise. I'm John Furrier, my co-host Lisa Martin. Next is these past two days live action. Our next guest, Cube Luminary and living legend, Lou Tucker, VP, CTO of Cisco. Been in the industry a long time. And it actually is the only person here at the OpenStack Silicon Valley Conference at the Computer History Museum that actually has a product in the museum itself. Lou, welcome back to theCUBE. Thank you, I'm afraid that dates me quite a bit. You look great, going on 35, something like that. 25, okay, thank you. We were talking yesterday, we got a great photo. I just tweeted on my timeline at Furrier. Lou and I, really on your product, that's actually in the museum. It's actually a cube shape, so cube, very fitting. It was a super computer. Exactly right. We have a super computer of media pumping out as much signal as we possibly can. But it was a super computer and it was all built from scratch. Exactly right. That's the way we did it back in those days. I mean, there was very much of a, coming out of a lot of MIT and Cambridge, Massachusetts, you know, thinking that you had to do it all and you had to know it all. So yes, we had our own processors. We had a different model though, but we were also trying to disrupt the industry much as we see it today. We were saying, one direction in super computing was make the processor faster and faster and faster. And Cray Research did that. We said, no, take thousands of computers and put them together. So take thousands of processors and we can outrun the one getting faster and faster and massively power super computing. Now we see that with the internet. That's exactly the same paradigm that people are using now to build these massively scalable applications. They're running on hundreds or thousands of servers and that's the way they serve their public. They don't build one computer to go super fast. This is a great transition because back then thinking machines, the one we were talking about, you built the hardware, you built the processor, you built the OS, you built the compiler. Everything was built from scratch, basically in a silo. But today the super computer is the cloud. So essentially people are building things from scratch in micro elements, distributed all over the place with a network, get distributed computing and you got unlimited access to software. So maybe it's all about entropy. The universe is getting more and more distributed. Things are really, really branching out. And the advantage of that is that you can move even faster because now other parts of the industry are putting in the pieces. And many things that we're doing today are assemblies. We're assembling microservices, we call them in many corners, but also taking other services and bringing them together to make a new product. That's the new paradigm. Don't build it all yourself. So let's talk about that because the digital transformation in the data center, certainly Cisco, the business that you work for now it's got a big stake in that game. But the bigger picture is the architectural mindset of doing that assembly. Runtime assembly, writing software to assemble component parts, microservices, Ryzen, Kubernetes, Docker, success, all this and now with standardization and a little open stack, open compute, all this stuff's going on, what is the mindset now? And based on your experience, you've seen a lot of waves. This wave seems to be faster. What's going on here? How do companies deal with this? I think, well, we are seeing, Moore's Law is still driving most of this even though we're beginning to see the end of the processor and speed improvements and things like that. So we're assembling that larger systems as a way to keep that innovation going. But it's almost happening, I think, at two layers. One is in the transformation, I know that Martin spoke yesterday here and I've always been a big admirer of his viewpoints, is this software transformation of the data center. We want things can move much faster when they're in software and so we're seeing that change take place and that is true for private cloud, public cloud and that's in the infrastructure layer where open stack plays. Now developers are approaching it completely differently. They're approaching, that's why we see Docker and we see containers. That is serving developers. It makes it easier and faster for you to develop applications that sit on top of now the software infrastructure. So now we're seeing those two forces sort of coming together. I think it's very interesting because there is a debate, is Docker going to overtake open stack or what's the co-existence? And I think all of us in the industry are seeing this is exactly the kind of layering we want to see taking place. So they complement each other very, very well. Yeah, and one of the things, I know she's got a question, I want to just continue that thought because Martin also pointed out that the developer way is completely developer led not supplier led. So it's a different paradigm. So you can have a gardener, all these people buy and purchase and all that stuff to how they do it. So I've got to ask you, so it used to be in the old days, you would wait for a vendor to do something. You'd set up some infrastructure and you were constrained by the infrastructure and the capability of what you had and the developers would have to deal with that constraint whether it's how much RAM is in the box or what network speeds and feeds add. Now it's reversed. The developers are dictating policy to the infrastructure. You see that? Absolutely, I think that's almost the biggest impact of cloud computing. Once we made it so that somebody can go into a Starbucks with their laptop and they can be productive, they can develop an application, they can deploy it on a public cloud, that means all of a sudden the developer is in control and the innovation can happen in that way and so you can see Airbnb starting up very quickly developing an app that allows them to serve a huge, huge market and they never want any infrastructure. So I think that the power has shifted to the developer and now they're in those. That's going to disrupt industries. Absolutely, in a very positive way. So with that power shift, and we've talked about that on theCUBE before and we were talking about it yesterday, what's your advice for businesses who are, there's so much choice, there's so many options, we could do it ourselves, we could outsource. What's your advice for customers who need to navigate the complexities with data center transformation and also the power shift? Culture is something that's quite interesting that we talk about here. What's your advice if the developers are now really empowered, how does the CEO become empowered and align with that to enable businesses to truly transform their data centers? It's a great question and it probably varied by industry. If you look at the transformation in manufacturing to a much more robotic driven manufacturing, instead of like how do you manage large fleets of people working factories, now it's fleets of robots and how do you make the capital investment for that? In cloud computing, I think the same thing is going to be true. What is interesting is now, again, power shifting towards developers and the power shifting towards delivering of applications over the internet. That's a very different model that any CIO has to now approach because what they're trying to do is really serve now instead of the systems of record or whatever, it's the systems of engagement. I think as Jeffrey Moore has talked about, that systems of engagement is really where businesses now can leapfrog each other. If they're more engaged with their customers through the internet, through an app, they're going to have an advantage. So they have to do two things. They do have to understand this. They will need developers, they will need the smart technologists and their companies knowing about what's going on in this transformation. And then they have to be aware of a larger ecosystem, ecosystem composing of open source software and of other services that are available that they can start to use that are on the cloud side. Well, the proof point too that one step further is systems of intelligence now tease out the boom of AI that we've been seeing. Then we're way back because that's what a connection machine was built for. The reason why we built that machine that's downstairs was to solve AI problems. And that's because computers weren't fast enough. We modeled it after a lot of neural net kind of technologies using many, many large numbers of very weak, small processors. So AI's now come full circle and being open. You know, we were at, Jeffrey and I were talking about our CUBE alumni, which you're part of, are so prominent. We have our own neural network. And we want to tap into that neural network. But let's talk about your product down there, thinking machines. It was supposed to solve the AI problem. So I got to ask you kind of the historical question and kind of bring it to today. Why has AI failed up to this point? Is it because of linguistic ontology not scaling? Is it because of the horsepower? Is it because of the software? Is it because of the academics, rigidity, dogma? I mean, I mean. I think almost all of those. I think there was many, many reasons. I think we were largely underpowered. So I do think even at that, even when we put 65,000 processors together, we did not have the computing power of the ganglion and a sea slug. I mean, it's really, really too primitive. Now we're beginning to see the emergence of enough power so that we can have autonomously driving vehicles. We can have Siri answering questions for us. That's a lot of computing power that we never had before. And it's still beginning to see some of the changes in how it gets applied like in healthcare and everything else. So there's been several shifts in the history of AI. We had expert systems for a while where we thought we could construct logical diagrams to answer questions. Now we're, and at that time there was also neural nets, but that weren't performing quite as well. Now we're seeing neural net technology is coming on very, very strong. And so you're seeing AlphaGo, for example, which got much better when you had one AlphaGo system playing against another AlphaGo system. They could learn from each other. And now real-time, after the data in memory, the advances of in-memory technology. Yes, exactly. But we still could not come back down to the network as the bottleneck. So back to networking, right? So how do we solve the network bottleneck? Well, we have to make sure. Martin tried that with SDN. That's true, that's true. Well, I think there's been some success there as well. I think that we're seeing that we needed to take networking from being a hardware problem that we were solving to now a software problem. And so I think what we've been seeing in software-defined networking and everything else has been that transformation. We're just at the beginning of that. And now I think that we are seeing things such as virtual network functions and NFE, network function virtualization, where we're seeing, again, software coming in which can be deployed more quickly, can be updated faster, and can be run at a lower cost for a lot of the service providers. But we're just at the beginning of that transformation. So I'd asked about open source. I had to talk with Lauren Cooney, who's working with you, and Cisco. And I want to bring a Cisco question, but I also want to bring a Lutakar question with that too, is open source is now a tier one citizen and trying major innovation at many levels. Too many to even talk about right now, but we'll have another save for that. But the old way of locking in the spec or having competitive advantage that Cisco took advantage of was a nestedness and architecture of proprietary hardware and software to have a great backbone. And it was really hard to swap out a Cisco gear, because it would take down everything. It was a great competitive strategy, product strategy. Now as the world moves to open source, how do you guys view, how should people look at competitive advantage, and how is Cisco changing their game to be more open source? Yeah, so one thing about Cisco to understand the world, Cisco's always been built around standards. And then, so we would get together with others in the industry. It's a slower moving process in open source, but the standardization process, that's how we got TCPIP and everything else. Then we would compete on implementation. And then we would say, okay, we're going to put this in the silicon, we're going to do ASICs, we're going to make the most competitive product adhering to that standard. In open source, I think it's very similar, but different motion is at play here. That's where we are seeing that. So it's tactically a different work flow. It's tactically, but it still serves the same purpose, which the customers do not want to see completely proprietary systems. What they want is something that they can avoid vendor lock-in. OpenStack has all been about avoiding lock-in of vendor, but now they're asking the vendor, show me that you have the best implementation. Show me you can deliver the service the best. Show me that you can move with the greatest speed. Or scale. Or scale. So those become the features, not the APIs and everything else. We're going to agree upon the APIs, we're going to agree on the platform, because that's what customers want. So Cisco's involved in like 35 or 40 different open source projects today. And so we're seeing it across the board within Cisco's business. So along that front, I wanted to get your take as the vice chairman of the board of OpenStack Foundation. We've talked a lot yesterday, we've seen a lot of success cases. In fact, Jonathan Bryce talked about wanting to be able to share that success. We've seen a lot, heard a lot about AT&T, SAP, some of the super users of OpenStack. With what Cisco was doing with OpenStack, when are we going to start seeing more diversity in terms of use cases in retail, in manufacturing? Are we on the precipice of that? I think we are. I mean, I think even in our shows, the OpenStack summits everything. We try to showcase different use cases that people have had from manufacturing. We're seeing a lot in terms of automobile manufacturing and IoT and those other areas in gaming. So we're starting to see that spread out. And that reflects something that the board is trying to drive very directly, which is that we want to start addressing users of OpenStack, not just the developers of OpenStack. Of course, OpenStack is an open source project and it's about the developers who are developing all of the different services in OpenStack, but we want to start focusing on how does this apply to the users and get much more input from the users so that they can inform the different project technical leads about what they need OpenStack to do for them. And that's the timing issue too, when you have incubation, you're building a foundation of OpenStack as a developer centric, obviously. Now you start to see this customer success has come in and you do see some diversity. SaaS players, you mentioned some gaming, you do. I think we haven't seen it, I think that Craig McGlocky was showing in Google and Kubernetes that they wake up every day and they discover there are new players that are using Kubernetes and in the application they never even had envisioned before. That's the fun part about building a platform, is that you all of a sudden are enabling other people to do all these wonderful things. Yeah, and the Docker is a great example. That comes, no, I won't say it came out of the woodwork, but I mean, it evolved very fast. Now Kubernetes is one of the fastest growing trends. What is the impact of Docker and Kubernetes to this whole system here? I think it's, again, it's a reflection of the fact that the developers now needed a better way for them to develop and compose applications. We've heard about microservices, something which is actually a design pattern that's not that new, it really is a design pattern around making a composite of a set of services that are each have to do a job very well and you bring them together to create your application. That is ideally suited, that's where Docker and Kubernetes don't allow that kind of thinking to take place so that you can construct smaller components, put them in containers, deploy them very, very rapidly, deploy them on your laptop, deploy them up in the cloud, deploy them on bare metal, deploy them on a virtual machine. It allows you to have the flexibility that you want, whereas in a container, you're putting everything you need into a container, that's why it's the shipping container, you know, analogy. It's an envelope. And it's a packaging system, but it's ideally suited then for developers to be able to share code and... Interoperability's been a key thing there. Let's talk about the future, let's talk about the vision. Okay, you've been around, you've seen the mini cycles, we've seen the computer industry grow from scratch. I mean, you've seen that, and everyone's talking about the main, the mainframes, the client server, all that great stuff, but let's go forward. What's the possibilities to do? When you think about IoT, you're really kind of thinking about a whole new data model, you're thinking about a whole new transit of packets, maybe different protocols. I think there's a direction. There's an arrow that we are following here, I thought, reflect way back into the mainframe era and into personal computers and client server computing, now into the internet, which is that computing is becoming distributed, it's becoming in multiple locations, that we're seeing an application has to, the more distributed you are, the more resilient you become to any particular failure. We've seen failures most recently in the airline, reservation systems, and that's why we're not fully distributed enough today, so that we can have cascading failures that can take out an entire sort of segment for a reservation system. Those need to be much, much more distributed, so we're going to rely much more on the internet and our computing, our ultimate computing platform is the internet, or the next generation after the internet to make it really a functional computer at a global scale. So if you and I started a consultant scene, we called it Tucker McKinsey Consulting, we have to advise in benchmark companies, how do you look at success of a company that's competing on a global scale? What does their company look like from an architecture from a philosophy standpoint? What is the benchmark of a good company and what is the benchmark of a marked for death company? I mean, you can almost like see certain patterns. Yes, yes, I think it's pretty clear. The companies, I mean, if you look at Amazon or Google, even Microsoft shifting in that direction, they are becoming software-based companies. They are fully distributed in what they're serving, they're serving global markets. So they're moving very, very quickly, much more quickly than almost any of their competitors around in that space. I think that if you look at, you know, Uber, the Airbnb, the other disruptors, they are using the same model. They're using the internet to connect to their customers. They're building virtual businesses that might not have any physical, you know, infrastructure they're holding with it. They are becoming fully digital companies. And that I think is those are the companies that work. That's the asset test for you. Are they fully digital? Absolutely. And the resiliency thing around distributed means just more fault tolerance. It's got a network theory kind of concepts. While the internet was built, right? Yeah, exactly. Is there a DNS for these companies? Yeah, yeah, yeah. Look, great to have you on theCUBE. Thanks for your insight. And congratulations on the great product you have in the museum theCUBE. I'm old and remember our thinking machines in Cambridge. A lot of my friends went to go work there right across the river from Northeastern. Great to see you again. Thanks for taking the time to share the perspective and it's always an honor. Thanks for having me on theCUBE. Thank you so much. You're watching theCUBE here. I'm John Furrier, Lisa Martin, live in Silicon Valley with Lou Tucker, living legend, CTO and VP at Cisco. We'll be right back with more great coverage after this short break.