 Hello fellow hardware nerds, and welcome back to Dallas, Texas, where we're reporting live from Supercomputing 2022. My name is Savannah Peterson, joined with the John Furrier on my left. Looking good today. Thank you, John. So are you. It's been a great show so far. We've had more hosts, more guests come in there before. I know. Amazing, super. We've got a whole thing going on. It's been a super computing performance. Wow, and we'll see how many times we can say super on this segment. Speaking of super things, I am in a very unique position right now. I am flanked on both sides by people who have been doing content on theCUBE for 12 years. Yes, you heard me right. Our next guest was on theCUBE 12 years ago. The third event, was that right, John? Third event, first ever VMworld. Yeah, the first ever VMworld third event theCUBE ever did. We are about to have a lot of fun. Please join me in welcoming Justin Emerson of Pure Storage. Justin, welcome back. It's a pleasure to be here. It's been too long. You never call, you don't write. Great to see you, likewise. How fun is this, has the set evolved? Is everything looking good? I mean, I can barely remember what happened last week, so. Well, remember, a lot's changed. That VMworld, you know, Paul Moritz was the CEO of your member at that time. His actual vision actually happened, but not the way for VMware, but the industry, the cloud. He called the software mainframe. We were kind of riffing. It was quite the decade. Unbelievable where we are now, how we got here, but not where we're going to be. And you're with Pure Storage now, which we've been, as you know, covering as well. Where's the connection into the supercomputing? Obviously, storage, performance, big part of this show. Right, right. What's the take? Well, I think, first of all, it's great to be back at events in person. We were talking before we went on, and it's been so great to be back at live events now. It's been such a drought over the last several years, but yeah, yeah. So I'm very glad that we're doing in-person events again. For Pure, this is an incredibly important show. The product that I work with FlashBlade is, one of our key areas is specifically in this high-performance computing, AI machine learning kind of space. And so we're really glad to be here. We've met a lot of customers, met a lot of other folks, had a lot of really great conversations. So it's been a really great show for me. And also just seeing all the really amazing stuff that's around here. If you want to see what all the most cutting-edge data center stuff that's going to be coming down the pike, this is the place to do it. So one of the big themes of the show for us, and probably, well, big theme of your life, is balancing power, efficiency. You have a product in this category, Direct Flash. Can you tell us a little bit more about that? Yeah, so Pure, as a storage company, right, what do we do differently from everybody else? And if I had to pick one thing, right, I would talk about it's, you know, as the name implies, we're an all, we're purely Flash, we're an all Flash company. We've always been, don't plan to be anything else. And part of that innovation with Direct Flash is the idea of rather than treating a solid state disk as like a hard drive, right? Treat it as it actually is. Treat it like who it really is. And that's a very different kind of thing. And so Direct Flash is all about bringing native Flash interfaces to our product portfolio. And what's really exciting for me as a FlashBlade person is now that's also part of our FlashBlade S portfolio, which just launched in June. And so the benefits of that are myriad, but, you know, talking about efficiency, the biggest difference is that, you know, we can use like 90% less DRAM in our drives, which, you know, everything uses, everything that you put in a drive uses power. It adds cost and all those things. And so that really gives us an efficiency edge over everybody else. And at a show like this where, I mean, you walk the aisles and there's people doing liquid cooling and some immersion stuff. And the reason they're doing that is because power is just increasing everywhere, right? So if you can figure out how do we use less power in some areas, it means you can shift that budget to other places. So if you can talk to a customer and say, well, if I could shrink your power budget for storage by two thirds or even, you know, save you two thirds of power, how many more accelerators, how many more CPUs, how much more work could you actually get done? So really exciting. I mean, less power, consumption, more power and compute. Right. And the power is at its center. So how about the AI implications? Where the use cases are? What are you seeing here? A lot of simulations, a lot of students. Again, dorm room to the boardroom we've been saying here on theCUBE. This is a great broad area. Where's the action in the ML and the AI for you guys? So I think not necessarily storage related, but I think that right now there's this enormous explosion of custom silicon around AI and machine learning, which as you said, welcome hardware nerds at the beginning, I was like, ah, my people. We're all here. We're all here in Dallas. So wonderful. You know, as a hardware nerd, you know, we're talking about conferences, right? I attended hot chips and there's so much really amazing engineering work going on in the silicon space. It's probably the most exciting time for CPU and accelerator, just innovation in since the days before X86 was the de facto standard, right? And you could go out and buy a different work station with 16 different ISAs. That's really the most exciting thing. You know, I've walked past so many different places where our booth is right next to Habana Labs with their GAUTI accelerator and they're doing this cute thing with one of the AI image generators in their booth, which is really cute. We're going to have to go check that out. Yeah, but that to me is one of the more exciting things around innovation, especially at a show like this where it's all about how do we move forward the state of the art. What's different now than just a few years ago in terms of what's opening up the creativity for people to look at things that they could do with some of the scale? That's different now. Yeah, well, I mean, every time the state of the art moves forward, what it means is that the entry level gets better, right? So if the high end is going faster, that means that the mid-range is going faster and that means the entry level is going faster. So every time it pushes the boundary forward, it's a rising tide that floats all boats. And so now the kind of stuff that's possible to do, if you're a student in a dorm room or if you're an enterprise, the world of possible just keeps expanding dramatically and expanding almost geometrically like the amount of data that we have, as a storage guy I was going back to data, but the amount of data that we have and the amount of compute that we have and it's not just about the raw compute, but also the advances and all sorts of other things in terms of algorithms and transfer learning and all these other things. There's so much amazing work going on in this area and it's just kind of this Cambrian explosion of innovation in the area. I love that you touched on the user experience for the community. No matter the level that you're at. It's been something that's come up a lot here. Everyone wants to do more faster always, but it's not just that. It's about making the experience and the point of entry into this industry more approachable and digestible. For folks who may not be familiar, I mean we have every end of the ecosystem here on the show floor, where does pure storage sit in the whole game? Right, so as a storage company, what AI is all about deriving insights from data, right? And so everyone remembers that magazine cover, data's the new oil, right? And it's kind of like, okay, so what do you do with it? Well, how do you derive value from all of that data? And AI and machine learning and all of the super computing stuff is about how do we take all this data, how do we innovate with it? And so if you want data to innovate with, you need storage. And so our philosophy is that how do we make the best storage platforms that we can using the best technology for our customers that enable them to do really amazing things with AI and machine learning? And we've got different products, but at the show here, what we're specifically showing off is our new FlashBlade S product, which I know we've had pure folks on the cube before talking about FlashBlade, but for viewers out there, FlashBlade is our scale out unstructured data platform. And AI and machine learning and super computing is all about unstructured data. It's about sensor data, it's about imaging, it's about photogrammetry, all this other kinds of amazing stuff, but you got to land all that somewhere, you got to process that all somewhere. And so really high performance, high throughput, highly scalable storage solutions are really essential. It's an enabler for all of the amazing other kinds of engineering work that goes on at a place like super computing. It's interesting you mentioned data as oil. Remember in 2010, that year, our first year of the cube, Hadoop World, Hadoop just started to come on the scene, which became, you know, kind of went away. But now you've got Spark and Databricks and Snowflake. Yeah, it didn't go away, it just changed, right? It just got refactored and right size, I think for what the people wanted it to be, easy to use, but there's more data coming. How is data driving innovation? As you bring, as people see, clearly the more data is coming, how is data driving innovation? As you guys look at your products, your roadmap and your customer base, how is data driving innovation for customers? Well, I think every customer who has been, you know, collecting all of this data, right, is trying to figure out now, what do I do with it? And a lot of times people will collect data and then it will end up on, you know, lower, slower tiers and then suddenly they want to do something with it. And it's like, well, now what do I do, right? And so there's all these people that are reevaluating, you know, we, when we developed FlashBlade, we sort of made this bet that unstructured data was going to become the new tier one data. It used to be that we thought unstructured data, you know, it was emails and home directories and all that stuff, the kind of stuff that you didn't really need a really good DR plan on. It's like, ah, we could look, now of course, as soon as email goes down, you realize how important email is. But, you know, the perspectives that people had on. Yeah, exactly, sorry, yuck. The perspectives that people had on unstructured data and its value to the business was very different. And so now- Good bet, by the way. Yeah, thank you. So now unstructured data is considered, you know, where companies are going to derive their value from. So it's whether they use the data that they have to build better products, whether it's they use the data they have to develop, you know, improvements and processes, all those kinds of things are data-driven. And so all of the new big advancements in industry and in business are all about how do I derive insights from data? And so machine learning and AI has something to do with that, but also, you know, it all comes back to having data that's available. And so, you know, we're working very hard on building platforms that customers can use to enable all of this really- Yeah, it's interesting, Savannah. You know, the top three areas we're covering for reinventing all the hyperscale events is data, how does it drive innovation? And then specialized solutions to make customers' lives easier. It's become a big category. How do you compose stuff? And then obviously compute more and more compute and services to make the performance go. So those seem to be the three hot areas. So, okay, data is the new oil, refineries, you've got good solutions. What specialized solutions do you see coming out? Because once people have all this data, they might have either large scale, maybe some edge use cases. Do you see specialized solutions emerging? I mean, obviously you've got DPUs emerging, which is great, but like, do you see anything else coming out that people are- Like from a hardware standpoint? From a customer standpoint. They're making the customers' lives easier. So, I got a lot of data flowing in. It's never stopping. It keeps powering in. Are there things coming out that makes their life easier? Have you seen anything coming out? Yeah, I think where we are as an industry right now with all of this new technology is we're really in this phase of the standards aren't quite there yet. Everybody is sort of like figuring out what works and what doesn't. You know, there was this big revolution in sort of software development, right? Where moving towards agile development and all that kind of stuff, right? The way people build software change fundamentally. This is kind of like another wave like that. I like to tell people that AI and machine learning is just a different way of writing software. What is the output of a training scenario, right? It's a model and a model is just code. And so, I think that as all of these different, parts of the business figure out how do we leverage these technologies? What it is is it's a different way of writing software. And it's not necessarily going to replace traditional software development, but it's going to augment it. It's going to let you do other interesting things. And so, where are things going? I think we're going to continue to start coalescing around what are the right ways to do things? Right now we talk about ML ops and how development and the frameworks and all of this innovation. There's so much innovation, which means that the industry is moving so quickly that it's hard to settle on things like standards and or at least best practices at the very least. And if the best practices are changing every three months, are they really best practices, right? So, I think that as we progress and coalesce around kind of what are the right ways to do things, that's really going to make customers' lives easier because today, if you're a software developer, we build a lot of software at Pure Storage, right? And if you have people and developers who are familiar with how the process, how the factory functions, then their skills become portable and it becomes easier to onboard people. And AI is still nothing like that right now. It's just so fast-moving and it's so, Wild West kind of. It's not standardized. It's not industrialized, right? And so, the next big frontier in all of this amazing stuff is, how do we industrialize this and really make it easy to implement for organizations? Oil, refineries, industrial revolution? I mean, it's on that same trajectory. Yeah, absolutely. Fourth industrial revolution. We've talked a lot about the chaos and sort of we are very much at this early stage. Stepping way back, and this could be your personal not Pure Storage opinion if you want. What in HPC or AIML, well I guess it all falls under the same umbrella, has you most excited? Because I feel like you're someone who sees a lot of different things. You've got a lot of customers. You're out talking to people. I think that there is a lot of advancement in the area of natural language processing. And I think that we're starting to take things just like natural language processing and then turning them into vision processing and all these other, I think the most exciting thing for me about AI is that there are a lot of people who are looking to use these kinds of technologies to make technology more inclusive. And so the ability for us to do things like automate captioning or the ability to automate descriptive, audio descriptions of video streams or things like that, I think that those are really, I think they're really great in terms of bringing the benefits of technology to more people in an automated way because the challenge has always been bandwidth of how much a human can do and because they were so difficult to automate and what AI is really allowing us to do is build systems, whether that's text-to-speech or whether that's translation or whether that's captioning or all these other things. I think the way that AI interfaces with humans is really the most interesting part and I think the benefits that it can bring there because there's a lot of talk about all the things that it does that people don't like or that people are concerned about. But I think it's important to think about all the really great things that maybe don't necessarily personally impact you but to the person who's not cited or to the person who is hearing impaired, that's an enormously valuable thing and the fact that those are becoming easier to do, they're becoming better, the quality is getting better, I think those are really important for everybody. I love that you brought that up. I think it's a really important note to close on and there's always the kind of terminator dark side that we obsess over but that's actually not the truth. I mean, when we think about even just captioning, it's a tool we use on theCUBE, it's, you know, we see it on our Instagram stories and everything else, that opens the door for so many more people to be able to learn. And the more we all learn, like you said, the water level rises together and everything is magical. Justin, it has been a pleasure to have you on board. Last question, any more bourbon tasting today? Not that I'm aware of, but if you want to come by, I'm sure we can find something somewhere. That's the spirit, that is the spirit of an innovator right there. There you go. Justin, thank you so much for joining us at Pure Storage. John Furrier, always a pleasure to interview with you. I'm glad I can contribute. Hey, hey, that's the understatement of the century. It's good to be back. Hopefully I'll see you guys in 2034. No, you've got the Pure Accelerate Conference. We'll be there. That's right. Yeah, we have our Pure Accelerate Conference next year and yeah. I love that. I mean, feel free to, you know, have that. That's awesome. Great company, great runs. Stay true to the mission from day one, all flash, continue to innovate. Congratulations. Yep, thank you so much. It's a pleasure being here. Have fun ride. You are a joy to talk to and it's clear you're just as excited as we are about hardware. So thanks a lot, Justin. My pleasure. And thank all of you for tuning in to this wonderfully nerdy hardware edition of theCUBE live from Dallas, Texas where we're at supercomputing. My name's Savannah Peterson and I hope you have a wonderful night.