 Live from New York, it's theCUBE. Covering Micron Summit 2017, brought to you by Micron. To New York City, everybody, and welcome to Micron Summit, hashtag Micron Summit. This is theCUBE, the leader in live tech coverage. My name is Dave Vellante, and I'm here with my co-host, David Floyer. We're talking about NVME and Flash and NVME over fabric. But right now we have a session. We're going to talk to the practitioners, the CIOs, the folks that are taking technology and turning it into business capability. Don Duet is here, he's the former head of technology at Goldman Sachs and Trevor Schultz as the CIO of Micron. Gentlemen, welcome to theCUBE. Thank you for coming on. So, Don, let me start with you. As a CIO, when you're thinking about technology, there's all kinds of bits and bytes. Technology's always coming at you a million miles an hour. What's the starting point, though? I'm presuming it's not the product and it's not the bits and the bytes. What's the starting point for you? As a CIO for, I think, pretty much any company, but certainly a company in the financial services industry, the starting point is how do you really use technology to help leverage and enable your company to grow, to change, to fulfill its mission, help service customers better? And so you start from a point of view of transformation. You start from a perspective of what's the solutions and the architectures that are gonna boost, enable you to move and in the world we live in with the amount of changes that are happening, both in business and in technology that you really need to be moving very fast and very rapid. And not to, I mean, it's kind of a cliche in the financial services business, but you've got to think about it as a portfolio. You got a portion of your investment that's running the business and another portion that's growing the business and you got transformation. Were you able to, and do CIOs actually look at it that way, are you able to precisely or is it all sort of mung together? It's easy to use those as kind of your vectors of how you think about your investment profiles, but ultimately at the end of the day, as you get to, I mean, it becomes one and the same. You unfortunately run businesses that are growth businesses that may have legacy investments in technology, may have parts of your business that are greenfield, which are maybe not at the forward start at this point in time and how you're thinking about the next opportunities for revenue. So we really think about it very comprehensively, which is that you look at what's your agenda, what's your structure, ultimately, what are the key paths towards creating innovation, towards managing to the best you can the ability to keep the cost of innovation as low as possible. So keeping the ability to run a digital business as cost effective as possible so you can channel more of your dollars and your budget and your people into really helping kind of create the next generation and drive innovation forward. You guys had an interesting discussion about cost versus value, and I want to come back to that, but let me bring Trevor into the conversation. I always like to talk to CIOs within a technology company. I call it dog fooding segments. Guys like Oliver Boosman, former CIO SAP says, can we call it drinking our own champagne? Okay, fine, whatever term you like or phraseology, but we heard some statistics early on today that Flash was only 10% of all terabytes shipped in 2016, and that's beginning to change quite rapidly. You maybe were on the leading edge of that because you had to bring in and you work for a micron, but so what are some of the things that you're doing at micron and maybe talk specifically about some of the things you've done with NVMe and NVMe over fabric? Yeah, so if you look at the strategic objectives of the company, there's always that go-to-market flavor. There's always that how does micron on micron work? Where would we employ that? How do we kind of figure out how to get business value and validate our own technologies that we're going to market with? And if you look at micron as just a, just in very basic light, it is a manufacturing company. It's a highly capitalized market and it's always about how do you get the highest capital asset utilization, right? I mean, it's the more I can produce out of the fabs, the more profitable and the better it is. So when you look at the strategic objectives of the company, it was, well, time to market is important, but time to yield is important, cycle times, turns. And so a lot of the technologies that we looked at the infrastructure, which were value add, which would actually move the needle for the business was in the operation side. So looking at how we run wafers left to right was important, that was interesting, but it was how we got the yield up. How do we make sure that we can change the recipes? And so we were starting to look at the analytics around that, the platform of data analytics. We looked at that as a real differentiation. And so when we, you can look at your transactional systems and find value, you can look at your process control capabilities and find value because you're working within the operation standards, but really where we started to see huge business values and this large, big data of information that was in this warehouse. And getting our process engineers and data scientists to get to this information they never could get to before, correlating pieces of information in order to get the capital asset utilization up was great and Flash actually drove, and the new architectures drove the ability for them to get insights faster or to get to capabilities that they couldn't before. So it became very clear over time that the new architectures were extremely important for our business and it would make us competitive because our competition's doing this too. I'd like to talk about the business impact and we can talk about some of the technology as well, but when you, we have one of our cube guests, Alan Nance was the former CIO of Phillips of Infrastructure and he made the statement about cloud and I wonder if we can apply it to Flash. He said, if you don't change your operating model and you just lift and shift and go to cloud, you might get a little benefit, but it's really tiny. You want to make a big business impact and affect your competitive posture. You got to change the operating model. First of all, do you buy that premise and how has the technology like Flash allowed you to change the operating model, whether it's speed, whether it's data sharing, whether it's changing the development mindset? I wonder if you could talk about that a little bit. Sure, I would say that first of all, unless I 100% agree with his thinking, you can take something that's problematic and you can move it somewhere else, but that doesn't fix the issues. And I think that as we thought about cloud and cloud architecture, much of the things that we really needed to learn and do was to really change the way we thought about building, deploying and delivering technology and moving from what we would have termed a very artisanal model where everybody kind of owned their own middle part of the world to a model that was much more structured, much more built to be durable, much more multi-tenant in nature. And that was very important both because it gave us the ability now to run at scale and get the benefits of that, but also gave us the advantage of saying, as you looked at things like Flash or enabling technology, that that benefit could be returned to everyone. So everybody would benefit by having those becoming part of your infrastructure stack because you're now sharing that and you're leveraging it across all of your customers not in consumers internally, not just the ones that chose to be kind of at that early part of the spectrum. And for making large, you know, kind of both cultural, organizational, and then technological change, you need those kind of benefits. You need the ability to go in and say, I'm not gonna go on a paint by numbers basis and convince you I'll have a Picasso at the end. I'm gonna go in and tell you right away you get to value fast. And you know, in our business, data is critical temporality or on data. So not just getting the data, but getting it soon when it's still fresh, enormously important. And again, so it was a big part of our cloud strategy was really positioning it so we could get those benefits fast and we could get them more universally. And that's been a huge part of our story at Goldman, you know, about 80, probably in the 90s percentage now of the environment is now being run in a multi-tenant internal cloud architecture. And there's a lot of work going to begin to now federate that out to public cloud providers. So a question I have is when you're introducing something like NVME over F, which is a new architecture and it's gonna change so much and it's changed the way that applications are being built, for example. The ability to have 10, 100 times more data in the same time that you could before just does, just blows your mind about how you design systems, the change of it. So a question I have is, as a CIO, how are you setting up the organization to take advantage of that? Because there must be some enormous constraints on the people who are doing it a certain way. But when I've been talking, for example, to early adopters of Flash, they've been saying, well, the first stage is I solved a problem, but then they were able to use it to do things they couldn't do before, reduce the workflow, the data flow time around the organization. Whatever the particular problems were, how do you manage that change? That's a question for both of those. Yeah, I think there's a retooling going on with a lot of different roles. I think, just very pragmatically, when you address the question about cloud versus on-prem, the buy versus build is there. And as part of that, you also think about the people. Do you have the people who actually can deploy and take advantage of this technology? Do you have the business that's ready for that as well? So when we are going through infrastructure decisions, we do talk about, do we have the right people who know how to take advantage of these new architectures? And so we're looking at this new capability and saying, wow, do we take legacy apps and extend them in a scale up way? And is this the architecture to do that? And the answer has been uniformly yes, absolutely. And then when we're looking at new capabilities, we do have the scale out conversation. And again, that's a different set of people or new skill sets that you have to go towards. So people get really wrapped up on the technology, but as a CIO, we also spend a lot of time talking about, do we have the people and the skills to take advantage of this new architecture? And the answer is absolutely, but there's work there. Yeah, I would say that I think we look at innovation as being push-pull. And in many cases, fortunately, it starts with the pull side. So Flash was a good example of where there were immediately parts of our business that said, hey, just shaving off microseconds in how we conduct our business can make enormous difference to us. So we're happy to be the early adopters, the guinea pigs, the pioneers, whatever you want to call it. And so they led the way into ultimately, where we got to a place where a number of actors, both the maturity of the product, the ability to get it supplied at a price point that was now kind of more competitive with spinning media. Now you're in the push side, then you're really going. It's like, no, I'm gonna bake this into my cloud architecture. I'm gonna bake it in so I can start getting mass innovation out of this versus the early adopters. I think for NVMe, I agree with you. I mean, I can't think of many things coming out of the finance space, very OLTP, very transactional architecture that has more potential to change the design pattern of how we build software and build acid-type solutions than NVMe. So I'm very curious, I'll be very excited to see and watch as that really evolves. And I think it will come with that pull side. I'm sure in our electronic markets trading groups, the concept of being able to say, can I now begin to transact on exchanges at wire speed because I can leverage the fact that I'm getting memory checkpointing and don't have to go and write or persist to databases or log files. I'm sure that they will be in that early stage of the pull of innovation, wanting to both adopt this and then really coming up with the software design patterns for how do you make that work and how do you make that efficient? But I think that, in my opinion, will be very, very interesting to see that unfold and see how kind of people begin to innovate on top of the technological capabilities that are being provided through the architecture. So this is a good place to close. I know we got planes to catch. So I want you to break out your telescope or maybe the binoculars. Maybe that's not go too far. And Don, you mentioned acid compliance and changes in architecture. You talked about low latency trading in the panel today, which was kind of about bringing speed of light, bringing compute to the data. We talked about sharing data. This whole notion of no schema on right. Things are changing in the way we design applications, design architectures, each of you. What do you see as the future of application development and IT generally? Yeah. So I think the total cost of ownership of these new models are a foregone conclusion. And so with that benefit, suddenly you see the business benefits when people can actually transact things faster or pick any use case. If I kind of look at the future, I think this fabric notion is unlocking a piece of the architecture that has been long needed. So we've always been talking about multi-core compute. We talked about IO bottlenecks and then memory is coming up now. So you've got this nice balance, I think that's coming. And if I look at what my customers are looking for is that we're swimming in data now. I mean, we used to talk terabytes, now we talk petabytes and what I do when I run my business. And they want to be able to get the same transactional and latency feel as in yesteryear. So they're demanding this scale of performance on datasets that we've never been able to attack before. So we can't keep talking about tiered sand solutions anymore. This new paradigm shift actually continues the scale up and scale out that my business applications are asking for. Yeah, I think in many ways, forever people talked about Moore's law and how you're continuously getting faster and cheaper computing. I think that what we're seeing now is that an extension of that, maybe it's Steve's law, I don't know. But ultimately, we're seeing more and more of the business applications as people are moving from being more facilitating workflow into truly digital where you're having system to system that the ability to get extremely fast, extremely dense, extremely scalable is becoming absolutely necessary. And I think that the products that were announced today by Micron are great examples of that now becoming common. You don't have to be a three-letter US government agency and have to build those solutions yourself. You can now buy them in standard form factors and begin to benefit from them. So I think it'll be, again, an amazing accelerant to more businesses as businesses become more digital and their use of technology will only grow from the result of these architectures being available. All right, we'll leave it there, gentlemen. Thanks very much for coming on. I appreciate it for having us. You're welcome. All right, keep it right there, buddy. We'll be back with our next guest. It's theCUBE, we're live from One World Trade Center in New York City. Right back.