 Live from Las Vegas, it's theCUBE, covering AWS re-invent 2018. Brought to you by Amazon Web Services, Intel, and their ecosystem partners. Well, welcome back here as we continue our coverage at AB2S re-invent, along with Justin Warren. I'm John Walls, we are live in Las Vegas in the sand. Stay one of our coverage here, three days with you all week. We're with John Masters now as the chief architect at Red Hat and John, good to see you this afternoon. Thank you, nice to be here. All right, first off, give me your impression of what you've seen so far on the show floor. What's the feeling you've got as you come in this week? Well, it's been absolutely fabulous for me. It's my first time at re-invent, so I've not had chance to witness firsthand the growth over the past few years, but I've heard stories that were up to 75,000 people, some very high number this year, and the growth is absolutely amazing. Very, very passionate people. It's very clear that the story of containerization and microservices is foremost this year, and yeah, it's just a fabulous experience to be here. Great, now yesterday there was that announcement from AWS about A1 instance. Tell me a little bit about how that comes to play in a Red Hat and is just your take on the release. Yeah, so Amazon did announce yesterday the new A1 instance type, and it's based on the ARM architecture. I think the interesting thing for me is that it's based on a processor that they themselves built called the Graviton, and this is really the culmination of what we've seen in the industry the past few years. As the cloud vendors get bigger and have greater resources and greater capabilities, what they can do is they can take that self-determination aspect and they can say, you know what, we're now big enough and we now understand and we're sophisticated enough that we can say we would like to deliver this to our customers and we don't have to wait for someone to build it for us, we can just go and do it. And so what they did is they licensed an ARM design from ARM Holdings, the actual core inside the processor, and then they built the chip themselves and contracted it out to a foundry, manufactured and deployed these and then they can snap their fingers and deploy these and surprise, now we have ARM-based instances, so it's been very interesting. So I'm curious, because we keep getting told that software is eating the world, and yet here we are building hardware and customized hardware. So what is it about the ARM architecture in particular, but also the fact that you can build custom silicon? What is it that Amazon, or indeed any other cloud vendor, what benefit do they get from manufacturing their own silicon here? That's a very good question. Well, I think there's multiple aspects to it. At the end of the day, people tell me that the future is serverless and I remind them that they're still servers somewhere, right? So we still need to have computers. Of course, we're going to have a smaller number of very big vendors on which we rely. I mean, we're seeing that with the adoption of public cloud. And as these vendors get bigger, they have that scale that they can invest. What for them is a modest amount of money. For anybody else, it'd be a fortune, but a modest amount and they can go and build a design. And now with the traditional microprocessor design, you take a team of four people and you would spend many hundreds of millions of dollars, maybe $300 million over four years to build a high-performance processor. What you can do with ARM is work with ARM Holdings, which is now part of SoftBank, to license kind of cookie cutter pre-made pieces. So you can license a processor core and you can stamp it out and say, well, I'll have 16 of those in my chip. And so you don't have to do the heavy lifting to design many of the building blocks, but you can integrate them together. So you get a lot of cost efficiency there. You don't have to go and do all that design, but you can integrate building blocks. And the key piece there, I think, is the ability to choose how you want to integrate that and what you want to build. And then what we're seeing in the industry is that computers becoming boring. I mean, everyone needs compute, but what are we talking about? We're talking about machine learning and GPUs and tensors and all kinds of other accelerators, right? So the interesting thing for me is once you've made the compute kind of so commodity that you can just license it from somebody and stamp out your own design, what opportunity does that bring later to maybe integrate various accelerators and other hardware goodies? I don't know what Amazon planned to do, but if I had a crystal ball, I would say, this is probably not the end. This is kind of the beginning of a journey. And now they will have the ability to integrate some very interesting and novel hardware advances of their own as well. Okay, because that does sort of lead into what my next question was going to be, which is for a customer of Amazon, as like, well, I don't know anything about the internals of chip design. Why would I want to choose the A1 instance type over one of the other existing instance types? What's in it for me? Yeah, a very good question. I think when Amazon announced it last night, the top line that the media picked up on first was the price benefit there, which was advertised as being 40% lower for certain workloads. Now, the design that they've chosen today is not about having that top shelf performance, that top line performance. If you want that level of performance, clearly you're going to use one of the existing instance types. But if you want to have something that is more cost effective for at scale deployments, maybe where you're not using all the compute resources that you need, you're more memory bound or you're doing web app serving, this kind of thing. In that case, you don't really need that level of compute. You still need the instances. And so this brings your cost down when you're doing that at scale kind of deployment. And that seems to be where they're targeting and in addition, they're targeting I think developers and those that want to invest in the ARM ecosystem because clearly this is the beginning of a journey. I don't know exactly where they'll go next, but one could imagine that it will continue from here. Okay, now, you are an ARM fan. I am. But you don't actually work for ARM, you work for Red Hat. That's right. So what's the Red Hat angle here? Well, so I'll tell you a story. Okay. I like stories. Me too. So back at the end of- I like stories too, John. Go ahead. Well, I'll spare you the long form. At the end of 2010, I was in one of my execs offices and I've been with Red Hat since 2006. And I had done a couple of things before that that kind of were very useful for the company, but kind of dull. So they said, all right, you choose something exciting to work on next, right? So I held up a Beagle board, which is a bit like a Raspberry Pi. And I told one of my execs, this will be a server one day. And I walked through Moore's Law and the pace of innovation and fast forwarded and say, if these things were to happen, this technology would be in a server. And now why is that relevant to Red Hat? Well, if you look at it from Red Hat's point of view, we don't pick winners and losers. What we do is we work with customers and what they want to adopt. But we also need to be able to respond to our customer's needs. So kind of the concern was, this arm thing looks like it could be interesting in a few years time. What if it is? And if it is interesting, and it's kind of a zoo, as I used to call it, a free-for-all, it's kind of an embedded mess, that works fine, well, fine in quotes, if you're building cell phone widgets and so on, because it's kind of a different ecosystem there. But if you want to have a mainstream server play, we had to have a few of us in the industry come in and say, all right, this looks interesting, but let's make sure that the level of standardization is there so that if this does take off, standard operating systems and standard software can run on it. That's why we cared was just in case it takes off. And then fans like me, of course, want to kind of promote it as well. But I think that's why Red Hat cared. Okay, and this is kind of off topic, but I'm just curious, because you talked about the acceleration of change. You've talked about innovation. You've talked about new wrinkles. Is it, and Moore's law, I mean, is it possible? Or do you see that the acceleration of change is so rapid that we're almost outpacing ourselves in a way in that change is happening so dramatically and so quickly that to make a decision on a particular solution or service is difficult because you're afraid of missing the next flavor in eight months or nine months instead of three or five years. That's right. And I think there's another piece there where the cloud makes even more sense, doesn't it? Because if you are a customer or an end user and you're deploying an app, you could say, well, you know, this arm thing could be interesting, I don't know. I don't want to go and build out physical infrastructure and go and pay that tax to go and figure this out. What I want to do is I just want to try it out right now. And the fabulous thing that Amazon did yesterday that no one had done, you know, there've been some efforts out there to provide arm to the mainstream, right? But Amazon put a giant rubber stamp on it and said, this is good enough for us and it works. Now, anyone who's used to a workflow in EC2, they can just use exactly the same flow to spin up one of these instances and try it out. It's a, you know, 30-second thing. Just try it out, see what you think. If you like it, great. If you don't, then don't use it. And because you are able to just consume it according to whatever you want, you don't have that commitment either, yeah. So a test drive? You can test drive it. If it works well, you can adopt it. There's no obligation. And that's, I think, key to exploring new technologies as well. Yeah, it does require you to have that software layer on top of it that runs. We were talking before that Red Hat has invested a lot to actually get the Red Hat software suite to run on ARM. That's right. So I'm sure that with this announcement, there's going to be a whole lot of other people suddenly discovering how to compile to the ARM architecture, which will be, that'll be fun. That's right. We've invested for the last eight years in this. And what we have now is a strategy we call our multi-architecture strategy. So again, we don't pick winners and losers. We have all these different architectures that we support. Obviously, X86, also Power and Mainframe and now ARM. And all of these architectures are treated equally going forward. So in REL8, which we just announced the beta of REL8, you'll see all these architectures treated just the same. And so the rule for our developers is, whenever they make a change, it has to run on all the architectures equally. Democratize it and then make it so that it is standard across the board. That's right. Makes sense. John, thanks for the time. Good to see you here at ReInvent and we wish you all the success down the road. Thank you very much. You bet. John Masters joining us from Red Hat. Back with more. We are here at AWS ReInvent. We're live in Las Vegas. And Justin and I'll be back in just a moment.