 Live from Las Vegas, it's theCUBE. Covering Dell Technologies World 2019. Brought to you by Dell Technologies and its ecosystem partners. Welcome back to Dell Technologies World 2019, here in Las Vegas. I'm Stu Miniman with my co-host, Dave Vellante. Talking multi-cloud, talking about Dell Technologies and all the pieces of the environment and we're going to drill in some to some of the storage environment. Happy to welcome back to the program. Sudhir Srinivasan, who's the Senior Vice President and CTO of the Storage Division of Dell EMC. Sudhir, thanks so much for joining us. Thanks for having me, Stu. All right, so as I said, day one, a lot of the vision, digital transformation, multi-cloud with Satya Nadell up on stage, got a little bit about today, got back into the products. Everything, you know, such a broad portfolio, everything from the latitude, you know, business devices through, of course, many updates in the storage world, been digging in with a number of your team, gives a little flavor as to, you know, what you've been working on, you know, I know as a CTO, you can't have a favorite family in the family, but some of the things you and the team were really proud of to unveil. Absolutely, thanks. It's been a big day as well, and I would say a big year for us, right? So we've shown incredible growth in our business in the last four quarters, taking share every one of those four quarters. So just a phenomenal year. A lot of that has to do with just the strength of the portfolio. We've been investing a lot in innovation in the portfolio. So I think the biggest one today that I'm really proud of is the Unity launch. I think it's a long time coming. We've been working on it for quite a while. The amount of performance that it's going to deliver, while also delivering incredible storage efficiency, data reduction, that's a huge, huge boost. But what we haven't spent a whole lot of time talking about from a technology point of view as a geek, what's cool about Unity XT that you may not have heard a lot about is that it actually is using machine learning inside. So last year we launched the PowerMax that had machine learning inside for making all these real-time decisions. We're taking that across the family. And Unity XT uses machine learning in order to actually deliver that data reduction that we just talked about, the five to one data reduction. And why that's cool is because we've had products that do data reduction with brute force, where they use a lot of memory. You can't do that in a mid-range product because that kicks you out of the cost profile. So we use machine learning to take advantage of a little amount of memory, but to still not compromise on the data reduction. Actually, I had this conversation today talking about PowerMax, we made a big deal about what was happening internally as well as what does that mean for the customers and the decisions that they don't have to make. In our industry, we've talked about intelligence and automation and storage for decades. So yeah, in the mid-range, what does that mean? What will be different from customers as they roll out the XT product line? So I think it's simplicity, it's just ease of use. We talk about ZeroTouch, in this case, there's fewer knobs and dials, right? You actually don't have to do a lot of tuning at all. Out of the box, it will serve the majority of the use cases and the requirements. You still have the option if you want to go in, if you're sort of the black belt type and you want to do, customize it to your own needs, you can do that, but that's sort of the journey we're on is we call this the autonomous or self-driving storage. So a lot of people are talking about it, we're actually doing it across the portfolio. And it's actually coupled with two parts or a couple with another part. There's intelligence in Unity XT and PowerMax, but there's also intelligence in Cloud IQ, which is our global brain in the cloud. And we saw that on stage today as well, where it's doing long-term analytics, deeper learning across longer-time horizons to help you manage the system without really much effort. So a couple of follow-ups, if I may, on the data reduction front. Sounds like that's a new innovation. You guys develop kind of from scratch. Are you bringing it across the portfolio or is it sort of obviously Unity XT today? Does the technology apply to other products potentially? Absolutely does, and in fact, that's something we're doing across the board from last year to this year, is you've seen we've become one storage team. And there's a lot of technology views going on now inside the portfolio. So things that we're doing in unstructured, for example, we're looking at applying it into other parts of the portfolio. Data reduction is obviously one of the key ones. It's the first example that people think of, so we are definitely looking at that. And what I'd also say is from a technology point of view, we're changing the way software is built. We're not building it as monolithic within microcode anymore. It's containerized assets that we can embed in different products. And then in terms of the autonomous storage piece, go roll back five, 10 years ago, tiering, and you had a lot of knobs to turn, and that was always featured as an advantage because people wanted to play with it. Yeah. What you're talking about today is an environment that's much more complex. And talk about more what autonomous storage is. Is it hands-off and? So that's a great question. So we have this internal, almost a joke. We call it, we're talking about self-driving cars. Surely we can build a self-driving storage system way now, right? It's kind of a shame that we're not doing that. But I would say it's four steps. Just like you have four levels of autonomy in self-driving cars. If you followed that level five, I think is the ultimate. Fully autonomous. Fully autonomous. And maybe we'll never get there. But similarly in storage, I break it up into four parts. One is, it's got to be application aware. So you're not dealing with LUNs and file systems and RAID groups anymore. You're dealing with, this is my application. That's how the user interacts with it. That's easy, relatively easy. It only took 50 years. Okay, go ahead. Too shame. The second element is sort of self-awareness. Actually, before that is policy-based. And so if you're driving a car, you're not telling the car which route you want to take. You want to say, I want to take the fastest route or I want to take the scenic route. That's it. And the car needs to figure out what that is. So that's policy-based. I want to optimize for latency, performance, blah, blah, blah. Third element is self-awareness, which is the storage system needs to know where it's operating in its comfort zone. Is it close to the edge? Is it going to drive off the cliff? Is it going to exit the lane to use the car analogy, right? It needs to know how far away it is from the car ahead. That's also, that's the stuff that we're now releasing with PowerMax and what we're doing in Unity. That's where we're using learning to figure out how close to the operating edge the system itself is. And once you have that, then you can start optimizing and self-healing. So that's- And that's level four. That's level four. And that's self-awareness. So you've got decades of data. Are you able to leverage that data or is that not as much use? So you have to- Absolutely the case. Yeah, okay. That's the key differentiator actually. Thanks for bringing it up because there's a lot of AI washing going on, right? As everybody says, they've got AI. AI is one thing you can't just deliver, develop overnight. No, no, no, no, exactly. So we have used all of the decades of dial home data. We've been working on machine learning technologies for the last five years, I would say, at least. Those models are being trained with the dial home data. And Cloud IQ is doing that on a daily basis now. Why now in 2019, Sudhir, is it are we at the point where this has become reality? Is it compute power? Is it the amount of data? Just better algorithms? It's two things. You nailed it, those two things. It's first and foremost compute power, but also I think the algorithms, they're much more sophisticated now. And we're well understood on what algorithms to use for what types of problems. I think there was initial, 30 years ago, there was like Uber intelligence. That was a very ambitious goal. I would say even today, that's not a reality. Why we're succeeding is we're applying it to very focused problems, right? Just like in the rest of the industry. We're applying it to focused problems that we can solve and then broadening our aperture. And to be clear, this is metadata. It's not customer data that you're utilizing obviously across the portfolio. That's right. We're looking at things like how much CPU it's using, how much memory it's using, how's the latency varying over time, how far it is away from its service level, things like that. Just another advantage of being old. Yeah, so we talked about that's metadata, but one of the things we talk about is when you talk about digital transformation, its customers become data driven. So we've covered this show, this is the 10th year we've been at this show. In the early days, it was storage. And oh my gosh, my growth of data and I can't take care of it. Big data was the bit flip of turn that from a challenge to I should be able to turn that into an opportunity and the next wave of AI is I should be able to monetize that and run my business and the data's one of the most valuable things we have. Bring us inside how that shift in thinking and data is impacting storage architectures and how you work with customers. That's awesome, great question. So data capital is the big thing around. You've heard that today as well. We are definitely sort of going beyond thinking of ourselves as a storage division to a data division and unlocking the data capital. I'd say there's several elements. One is building the best storage for data applications, especially AI and ML. So I think our unstructured products clearly are leading the charge in this. We've got the machine learning solution with Icelon. It's a perfect fit for that kind of application. That's here and now already, right? Using of GPU technologies in conjunction with our scale out architectures, critical. But going beyond, we're looking at does it make sense for some of these data crunching applications to be closer to the storage layer? Thinking similar to what hyperconverged has done for general compute. Is that a thing that would really unlock the data capital? We think that's a lot of potentials. And I'm glad you brought that up because the storage geeks talk about NVMe, NVMe over fabric and storage class memory. Explain how that fits into what you were just talking about and not just the next major wave of a tool inside the infrastructure. So storage NVMe and NVMe over fabric was part one of a two-part story as you know. That allowed us to get that super low latency, high-speed connection from the application to the storage, to the data. But the data devices themselves were still, I mean, flash is great compared to HDD, but you're talking single microsecond type of, or sub-microsecond applications that meet that kind of latency. And that's where storage class memory comes in, right? So we're finally getting to that point where the storage devices are operating in that 10 microsecond range, which will start to really get us to, but if we can get those things co-located, close by, it unlocks a lot of things. And the beauty of NVMe over fabric is that it can give you the sense of being close by without actually physically being close by. So you can still be disaggregated and that opens up a whole lot of architectural options. Can I ask a follow-up question on storage class memory? The skeptics would say it's just way too expensive and you're not going to get the volume of flash that you get with these. What do you think? That's what they said about flash, too, didn't they? No. Flash went into all the consumer devices and that got the scale to bring the price down. Yeah. I mean, maybe before iPhones, they said that. But iPhone was the catalyst for, so is there a consumer analog for storage class memory? I don't think it would be a consumer analog. I think that's fair. But I think there will be volume to drive it down. However, I'll say it's a fair point. I think that the actual magic lies in combining super fast, perhaps expensive storage class memory with cheaper flash storage. And so you almost have a hybrid solution again. So the old hybrid becomes, you know, hybrids back in fashion, even with solid state. The storage pyramid lives. Exactly. We think that's going to be the killer combination. All right. So, Sidir, can't let you go without, give us a little bit of a look for it. We talked about where we are. You talked about some of the journeys that were there. So it's our 10th year here at the show. Wow. And come back for year 11. You know, how do you foresee the industry maturing and moving forward? I think for year 11, the big things we're going to see is cloud. Two things. One is cloud, and the other is software defined. I think those are the two that are going to be big news next year. And we're seeing some sneak reviews of that this year with the cloud announcements we made. You'll see a lot more of that next year from the storage side, both in, as being part of the Delta Cloud Technologies cloud platform, but also cloud enabling our storage arrays across all the public clouds. And then the second part is software defined. I think that's really the next wave, right? So as I said, we are a long journey internally. We've already been on it, where we're transforming our internal storage assets to be more software-centric. And you'll start to see some of that. All right, well, Sudhir, really appreciate you helping us geek out and dig into a lot of the pieces here at Delta Technologies World 2019. Thank you. All right, for Dave Vellante, I'm Stu Miniman. This is the end of two days of wall-to-wall coverage. We're coming back for one more. And as always, check out thecube.net for all the videos, siliconangle.com for all the articles and tools. Wikibon.com for all of the in-depth analysis. Hit up Dave, myself, John Furrier and the whole team. We're available on social media channels. And as always, thank you for watching thecube.