 Live from the Julia Morgan Ballroom in San Francisco, extracting the signal from the noise, it's theCUBE, covering Structure 2015. Now your host, George Gilbert. This is George Gilbert. We're at the Iconic Structure 2015 conference. This is theCUBE's exclusive coverage. We are, we have a special guest with us, Jim Donaldson and Jonathan Donaldson. Jonathan Donaldson? I knew I was gonna mess that up. Hey, can we? Yeah, yeah. All right, so. Busy venue. Yeah, a little, all right. We'll have a little humor mixed in. All right, so Jonathan is a guru on all things software defined in Intel infrastructure. Why don't you tell us, Johnson, about the sequence of compute infrastructure or data center infrastructure that went software defined and what that meant. Yeah, so obviously, if you go back and you look at kind of the evolution of this, right? You started with the virtualization of the compute itself. How do I get virtual machines on a server? How do I get the one server to look like many? And that brought along with it certain challenges. Okay, so then how do I then go programmatically orchestrate the networking for that, right? So if I'm going to move these virtual machines around or I'm going to really try and derive my operational cost down significantly, I have to start to do that programmatic automation of each of those components over time. And so compute was the first one, right? We saw things like DRS from VMware and vSphere come out and then so. DRS meaning disaster recovery. No, dynamic resource scheduling. So it's their automated scheduler for how do I move around those virtual machines in my cluster environment to have the best performance, right? To take things out of, you know, put them into maintenance mode so I can go replace things, that sort of thing. And then we started to see that same sort of trend in the networking side and that became software defined networking. So how can I take, you know, my switches and my routers and when I move a workload programmatically move, you know, the IP addresses and all the security features and load balancing things along with it. And then we saw, now we're seeing the trend for that also on the storage side, right? So if I have, you know, sand based storage attached to some compute nodes and that's where my data is predominantly, it's really hard to then to go migrate those workloads elsewhere in the cluster. And so we're not just programmatically but because you have bandwidth limitations. Right, so bandwidth limitations, programmatic limitations, you know, how do I open up more of the pipe on a programmatic basis when I know that I have a migration event that I've got to go do. And so it's really kind of stitching together all of those components that really is driving the value. Does one vendor have to provide all those programmatic interfaces or if multiple vendors try, will they sort of step on each other? So I think that, you know, what we're seeing in the industry is, you know, things like OpenStack are trying to build that kind of, you know, predominant control plan for these types of things to happen. You can kind of slot many different things underneath. You know, VMware does it with VCD. I think that you're seeing kind of lots of emergence of, you know, different ways of doing it. I think having, you know, the well-published standardized, well-known APIs, either through standards bodies or through usage, right? Those are the, that's the big thing to make sure that people don't step on each other. It's not a problem that's been solved yet. It's one that continues to evolve. Even in a single vendor environment. Even in a single vendor environment, because you don't have a single vendor environment anywhere, right? So if you're talking about, you know, taking a single software ISV vendor, right? Whether it's an OpenStack vendor or, you know, VMware or whomever, right? Until they start to actually layer on their own flavors of each of those software-defined places, they still have to interact with, you know, EMC sands or NetApp filers or, you know, SolidFire arrays. They still have to interact with, you know, Rista switches or Cisco switches, you know, and routers. So the concept of being a single vendor environment doesn't exist. So would this be sort of like systems management where that's a nirvana we'll never reach because we're all stuck in the labrea tar pits of, you know, someone trying to manage, someone trying to manage someone that. Yeah, I think prior, you know, in recent history that the overall, you know, unified theory of systems management was a, it was a nice to have. I think we're actually entering the time where it's a must have. And so I think a lot of those problems will be solved because if a vendor or an ISV or even a customer wants to stay in the game, as far as being able to deliver the services, they're going to have to go do that work. Okay, so VMware that you cited is primarily on-prem. They have their own cloud infrastructures also? Yeah, but, you know, it's not sort of widely adopted. OpenStack may be different, although, you know, of the big three, you know, Amazon Azure and Google, they're not really, you know, in any one of those. Because each of those, they're rolled around, right? So they're secret sauce to their business and how they deliver their services is based upon the software that they've built in-house. Their own proprietary middleware, which has open APIs, but, so have customers in their own data centers been able to achieve the operational efficiencies that come from programmatic control of their infrastructure? Many have, right? So it's a journey and we see, you know, when we go and talk to customers, we see many that have started the journey, many of them that are well along the journey, that typically the further along they are, the more proficient at they get and the more benefit they see from it. And it's traditionally, you know, what pushes somebody along that path faster than others is how strategic is their IT to their business, right? So if having, you know, agile development environment, if having, you know, easy to spin up and spin down resources is actually strategic to my business, you know, it deals with my communication to my customer or my order entry or my manufacturing or something along those lines that I'm actually, you know, deriving revenue from, they tend to be further down the path. If it's more of just a, you know, IT is something that I just consume because it helps lubricate, you know, the communications of my business or helps, you know, kind of just drive the business itself, but it's not really critical path. They tend to be more on the, you know, the early stages. Is this something then perhaps that is so strategic to Intel and getting the end customers, not the big clouds, but, you know, your traditional IT shops to be able to achieve the operational efficiencies, get to close to the operational efficiencies that the big cloud vendors do, is that something then you would invest in whether it's OpenStack or, you know, VMware to get customers solutions that are closer, you know. Absolutely, that's actually, that is the goal of our cloud for all initiative that we've announced, right? It's to get the hybrid and the private cloud instantiations, you know, tens of thousands of those, right? Getting, you know, all of the global 2000, the Fortune 500, getting all of those customers delivering in that methodology because we know that when we looked at what happened in the large cloud service providers, when you start to make, you know, the per unit cost incredibly low of a resource and also at the same time, create the barrier to entry as very low or the ease of access of those very easy to get to. Jevons Paradox kicks in and then you actually see a massive growth in those because the innovation cycle kicks in. So if we were two gentlemen in a bar and we didn't wanna go out in the rain to catch a cab, or we didn't have cash in our pocket and we said, there's gotta be a better way to do this. And we, you know, if you napkin'd out what Uber is today, 10 years ago, you wouldn't have been able to do it, right? You would have had to go purchase servers, you would have to go find a COLA facility and then you're not gonna be able to scale against that, right, because, you know, if I, the first 100 customers that come in have a great experience, but when somebody goes and puts us on social media, this is the best thing ever, right? Then the next thousand that come in have a terrible experience and then they never come back. So being able to do that programmatic scaling and growth of those, that is one of those things that has driven, you know, the likes of AWS and Google and Azure forward. If we can go do that same thing in the enterprise space, we're very, very confident that there's a whole set of innovation locked up inside enterprises that will be unleashed the same way. And how much of that is a technical challenge and how much of that is changing work processes, like, you know, the San administrator knows a San, you know, in a hyper-converge box, like, you know. You know, so that is absolutely the biggest problem. It's the people in process, right, that's always the biggest challenge, you know, technology tends to lead against that and this is probably the third or fourth or twelfth wave, right, depending on what you count of this type of technology trend. You know, as part of the voice over IP transition that when I was at Cisco and we would walk into customers and, you know, the PBX administrators and the networking administrators hated each other and there's just no way that you'd ever get them to go. The bell heads and the net heads. Exactly, but over time, right, if they, as they transition their processes and their value from, you know, being able to just do the PBX to being able to do the voice on the network and the network staff were able to understand the value of the voice sitting there and you're able to get those processes to merge, right, that's when the big, you know, uptake of that technology and the benefit actually started to be seen. But in this case are, well, you would be in a better position to know than probably anyone in the world how fast workloads that would have been done on premise are migrating to public clouds and how much time you have to get the people and the processes on the private clouds fixed. Right, so I think one of the things that we're seeing is we're not, we're not predominantly seeing that, mass migration of workloads, right? I shouldn't say, you're capturing an important point. It's not the migration, so much as the new apps. Yeah, the new apps, so absolutely. So apps that are created in that cloud native style, right, is not just happening in, you know, the AWS as a new user is, although that is predominantly the first place that they tend to spin up, right? We're also seeing that kind of delivery model with this, you know, kind of the new DevOps style of microservices and container, highly automated, orchestrated types of situations, Kubernetes and Tectonic and MesaSphere and all of those types of, you know, words that we hear, right? They're all kind of wandering around, right? So we're seeing that happen even on the private cloud and in the smaller service providers too, where somebody will say, okay, the tools that I have at hand, I'm going to spin up 10 or 15 VMs and then I'll overlay those technologies in there and I'll treat it just as if I was in a large public cloud provider. So the methodology for creating those applications, absolutely, is taking off the likes of Docker and Rocket and CoreOS and, you know, all of those vendors, you know, Red Hat's Atomic and just, you know, there's this mass explosion of ways of doing application development that are new. The big challenge is giving the best landing platform for that and we're seeing it happen both in the public cloud and on the private cloud side. To what extent does the buying power of the public cloud vendors get to a point where there's a meaningful cost advantage or is the labor cost, I guess, what are the economics beyond the, you know, we're going to do new apps in the public cloud or some of them or, you know, more of them and some in existing? What are the economics that are driving, you know, the choices? It's mostly operational expense, right? So when you look at like an enterprise IT budget, right? So predominantly it's, you know, it's a very small slice that actually goes to the capital expenditures and a very large portion that goes to the operational side of things. And so, you know, reducing the operational expense actually helps balance out, you know, kind of any advantage that, you know, public or private has. You know, we've always said from the beginning of time, right, that hybrid wins, right? That is the best model for, you know, cloud type of applications, right? So it'll be some that you want on your premise because either they have access to in-house systems or security concerns or data gravity concerns. There'll be some that you float out in the public cloud either for, you know, burst capabilities or there's some feature or function that you have out there or it gets to get closer to the customer base, right? The, you know, placing your workloads for the right reasons for the application itself is where we think that this actually ends up having the most value. Jonathan, I won't call you Jim. Donaldson, very much enjoyed having you. Absolutely. This is George Gilbert. We are at the Julia Morgan Ballroom in downtown San Francisco. And this is Structure 2015. We will be back in a few minutes. Thanks.