Transcript
Greetings, and welcome to The Pure Report. I'm your host, Rob Ludeman, and it is time to bring the orange. Coming up, a special two-parter where I'm bringing back two of my favorite technologists, the ones that got us going with the 15 Architectural Decisions and the Unplugged series from Audio, and we're going to be digging into all the keynote announcements and doing it at a layer that's going to help you learn why we made the decisions to bring out those technologies and capabilities. Two of my favorite folks in the company from left, Mr. Andrew Miller. Hello, hello. On my right, J.D. Wallace. Always a pleasure. Gentlemen, thanks for making the time to come in. We are right at the one-year mark, if you can believe it, when we brought out 15 Architectural Decisions. So much work to put out a six-episode series, but that was July last year, so it seemed a great reason to bring you back. I had forgotten that it had been that long, but it's great to be back in the beautiful studio you have built here. Yes, yes. Six episodes. That seems like it must have been a lot of work. Somehow, I think I blacked it out. Maybe we just talk about it enough that it's like it happened, but that was a lot of fun. It happened, and honestly, for people that keep discovering it, they're finding a lot of value, whether they work at the company and maybe are new, or when I bump into customers and they say, yeah, I didn't really understand this feature, this capability, Evergreen or why Flash, and that really helped double-click into their understanding, and that's what we're going to do today, because back in June at Accelerate, we brought out a lot of really exciting new technologies, capabilities, products, and we ran some episodes on that, but we've now had a couple months, right, for you all to get out and talk to customers, to talk to prospects, and really get some perspectives and bring those to everybody on what those mean. So folks listening may have, if you have had actually a session with your AE or your SE, you might have actually, they may have actually used something that came out of Accelerate in a session that JD and I did, a kind of a customer meeting Accelerate deck, because I actually used to do this a lot as a partner. It was like, hey, what's new from whatever event, whatever conference, et cetera, and it is sometimes a little bit of an inside-out, but it's the assumption, like there's cool stuff going on, so as long as you translate it into terms and outcomes I care about, it's all right to talk about the stuff that we do. A lot of times we start strictly from, hey, what do you care about in your business, and we should do that, but sometimes it's just fun to not quite geek out a little bit, but like we've got cool stuff to talk about, and we can then relate it to things that matter in your environment as a customer. For sure. So we're going to be diving in, and we're going to be hitting the Unified Data Plane announcements, Intelligent Control Plane within this episode, which is great, there's a lot of great products and technology. Episode two for everybody, we're going to be going to the Universal Data Intelligence and EverPure Data Intelligence, and then also kind of wrapping that on a bow a little bit with the EDC Blueprint, because Enterprise Data Cloud continues to be the thing that we talk about that incorporates all the technologies that we are bringing to market. And I feel like J.D. just got very sad here, because we'll also have an episode two, the Intelligent Control Plane, which is not solely fusion, but fusions within it. You certainly can't have the Intelligent Control Plane without fusion. Bingo. Yes, J.D.'s favorite thing. What's the why now for these things? If you look back from June and the session that you did, what is significant? Obviously, we did the rebranding in February with the company name and where we're focusing, and certainly there's more kind of upstack conversations, but what stood out to you back in June about why this is significant for us now? Just this continuing evolution of we've talked about, and we even started this out a little bit in the session, Accelerate, and so we'll hopefully not allude back to that too much, but J.D. and I did some of what we're doing now in like 45 minutes, like we were moving, kind of thing, so appreciate the opportunity to kind of let it breathe a little bit more. But one of the things that was there was we started by saying, you know, hey, we're not actually here to talk or introduce Enterprise Data Cloud very much, but we're actually here to walk through all the things that we're announcing that fit into that construct. So like a year ago, we were announcing Enterprise Data Cloud as both an architecture and a strategy, and it's a marketing message too. It's all of these, and that's all true. But what was cool about this session and what we're going to do now is it's actually starting to put a lot of the flesh onto the bones, bring closer to Halloween, whatever analogy you want to put, kind of thing. I think that works. So we're actually showing that it's not just a high-level strategy, and even so we talk about a marketing level, but it actually maps to how we're creating features and products and capabilities to flesh that out. And it's still around two years after we started talking about it, which is also awesome, so we can keep developing it out. Yeah. And what I would add is, you know, think about, just take a step back and think about Everpure, Pure Storage, as the evolution of the company that we are, been in business, we've been doing this for over a decade now. And I think naturally, as you would expect, early on, a lot of the conversations were around the hardware that we're building, around the products, right? And even in our architectural decisions, so many of those architectural decisions were built around changes that went into or things that went into the product itself. Very direct linkages. And so we're at a point of maturity, just as a business and as a company, where absolutely we're continuing to refine and bring out new products. But what is more important as we grow is how we put those products together, how they work together as an ecosystem, and the solutions and the services that we stack on top of them that start to deliver more and more value. Dare I say, a platform? Yeah, by another name. But doing it the right way. Coming at it the right way, not like, we have a platform and we'll pretend it. There's actually all the pieces that are coming together to make a platform. And the timing, I think, is really good. Because one of the narratives that we hit on keynote a lot at Accelerate and has pervaded the messaging going forward is this notion of where we are from changing from sort of an app-centric view of the world to more of a data-centric view of the world. And certainly really relevant as you look at what AI needs to do and how it operates. But that was one of the really interesting pivots that I saw, showing how historically apps have been siloed and connected to data. And now that model really breaks down if you want to be successful. So as we're going into this, and you were alluding to the first section, Rob. So we're thinking about how to do this for folks that are listening, because we did this with slides. And so this is a video format, so you get to see our faces instead of slides. Better, worse? You be the judge. I have no glasses. And yes. So I wanted to narrate a little. We're going to try and do this balance of narrating some of what you would get in the slides for free if they were visual, but also then adding some additional color that we didn't have time to. So in this first segment, I wanted to focus in on kind of the keynote pieces that people can go back and, keep me honest, Rob, I think the keynotes are still online. They're still online. Or maybe they're carved up into digestible pieces kind of thing. So section one in this episode is kind of what was said from a keynote standpoint. There's data primacy. We'll go into more of that. And then focusing in on the payload overall. And that's all we think about that is, because that's a little bit of an inside baseball, because that's maybe even a marketing term a little. What's the payload? You know, kind of thing. Not bombs dropping out of a B-52 or whatever, but like a good payload that you want to get. And then focusing in on the universal data plane. Make sure I've got my name. My name's right. Yeah. Unified data plane. Shame on me. So, you know, we all get this right. And then part two, we'll come back and focus on the intelligent control plane and on EEC success blueprint and a little bit of other pieces. So we'll try and signpost along the way. Hopefully not so much that it's annoying for the audience, but have them a sense of where we're going along the way. So with that, if it's all right, I'll keep going into a little bit of the opening from kind of what was talked about at a keynote level. And this is now what you were hitting on, Rob. So there were a couple of big ideas at the keynote level. We are now, I don't know, five minutes in. And I'm going to say AI. So I get paid today, right? I said it a couple of minutes ago. It's all right. Man, you beat me. That's acceptable. Apparently, I wasn't listening. So there was obviously starting out with some of, is your infrastructure, your data infrastructure ready for AI? And this is just like practical stuff, because even if this is sooner or later, the future is true unevenly, you know, that quote, if you will. So thinking about, you know, what is the new reality? Do you have visibility on your data? What data is relevant? And then you become my infrastructure scale. So Charlie and his keynote especially walked through that. And then he even started to walk through a good bit of kind of this idea of data primacy. And I think that was what you were alluding to, Rob. We were app centric and they were becoming data centric. I think, J.D., you had some thoughts. I think it's important to find that though, right? Because it's somewhat a new term that, and by the way, autocorrect, autocorrect is a pain I learned last week, because primacy goes to primary really quickly over and over. So watch if you're writing that, but it's a term that we're now starting to use. And it's super interesting relative to, you know, elevating data in the mix. Well, if you think about it, it's something we've actually been talking about for a while now using different language, different words. You know, think back to the initial introduction of FlashBlade. And we had the concept of the data hub, right? Really starting to recognize... That's a throwback? Yeah, really starting to recognize that, you know, applications as they were, as we were talking about this ecosystem of applications, they all understood their own view of the data. And so we were in this business of constantly moving data from platform to platform. ETL. From software to software. And especially as the creation of that data exploded, as we started moving from human generated data to machine generated data, we started growing and growing how much of that had to move around. And then as AI, Andrew, to your point, started to evolve, that data started to have more and more value because we're using it for training. We're starting to mine that in new ways. And so we got to a point where we realized we couldn't keep just moving the data around. We needed to move into a data primacy mode. Data lives in the middle. Data lives in the middle. Data moves around versus the other way. Absolutely. And so, again, we've been talking about this for a while using different language, but I think where I really appreciated Charlie taking the keynote at Accelerate was really focusing on that, or maybe refocusing on that, using some new terminology that I think fits today in some of the conversations we're having in data primacy. While it's a new term we've got to learn, I think really keys in nicely on what it is we're trying to convey and accomplish. It's even a follow up. We've talked about the idea of data gravity for a long time as an industry, as people, there's the data as gravity because it takes, by definition, time to move or do things with. So that only gets worse. And as soon as we start to talk about, I still like talking about order of magnitude. It makes me feel smart. But like anything, anytime something gets 10x or 100x, it's a different thing as far as how you have to handle it. Because that actually probably links back into AI as well because there's that much more data being consumed. You know, and there's two things that I think most people will be obvious to them. As that data grows, certainly your infrastructure gets fragmented, all those different applications we talk about. Certainly it becomes difficult to manage at scale. But the thing maybe you don't think about is something that we've kind of called semantic data sprawl, right? Ooh, is that a new term? That's how every application is really defining in its own way for its own purposes. Maybe the metadata or maybe the information about that data that's relevant to that application. But it owns that context, right? So as we move through a pipeline or a workflow, we start to lose that context. And so one of the things that we're focusing on or we're thinking about is how do we control some of that semantic data sprawl? How do we make that information available to the entire workflow, right? So we're joking back and forth. And this is stuff that we are learning. So hopefully maybe the audience is learning with this too. If you're like, I already know this, you're like, ha ha, you're hit. But semantic is a new term. Ontology is a new term. What are the relationships and the connections? And thinking back to the excellent episode that you had with Ashish, so we're not going to do an imitation of him, but we'll hit on some of those same ideas. But getting comfortable those terms, if you've been a practitioner in this space or you're new in this space, it's worth digging into those because it's not as easy to feel like, OK, these are just new buzzwords. But there's actually a lot of depth underneath these that's worth unpacking. And I'm starting to get to the point where I feel like I'm not saying it as a buzzword. It actually has some meaning inside my head when I use the words. It's an adjustment, though. It's something we're all learning, although this has been the strategy going forward. Because I feel like I had seen leadership strategy slides from three, four years ago where this was in our heads about where we were going. And at the same time, the different layers that we're talking about, not one of them can live independently and drive value. They are all interrelated. You must have, where we're going to dive into soon here, you must have a strong platform that has the ability as a foundation to manage data regardless of what type it is, regardless of whether it lives, and do it at the right levels of performance and efficiency and availability and security to then be able to go do things around automation, to then be able to do semantic things, context things. So they're all symbiotic. They all rely upon one another. But isn't it great that for years and years, we have spent all this time building up the platform, building up that foundation that so many people rely upon? And hopefully the way that we've grouped all the announcements really flow nicely into that visualization you just talked about. You know, the unified data plane, building the products and solutions that are foundational, the intelligent control plane, how do we connect all those things together? And then the universal data intelligence, now that we've built that infrastructure, what are the new insights that we can glean out of that, right? Bucketizing these things into these three categories. Roughly layers, but also sometimes not two when you look at the individual pieces. So I think that actually, we're wrapping up section one, which is the, okay, what was the general messaging? You want a longer version of this, anyone listening? Go listen to the keynotes, because Charlie and Kaz and Sean Rosemarine, anybody, Chad Kenney, how can I leave Chad out? You know, they did a great job. This isn't condensed, but this is the flavor of the overarching keynote messaging. And then JD, what you're alluding to is what we're going to do for the next time. Sometimes if you have this actually as a presentation or discussion with your AE and SE, you may even see two versions of the payload slide, like all the categories. And so the three items that you just mentioned, JD, were on the data plane, the control plane, data intelligence. So sometimes, and this is now a little bit of inside baseball. The reason we had two versions of this slide is that actually one is for analysts, because they're just focused on what is the brand new stuff that hasn't been talked about at all before. Now, just because you talk about something, we announce it, doesn't mean it's released yet necessarily. It may take time. So there's a version of this that actually just talks about Excel 190 and our partnership with Microsoft and OverDrive and some of the new things in Fusion and the acquisition of OneTouch being rebranded data intelligence. And that's kind of if you want the shortest, highest. highest level version of the announcements. Then there's also the version that actually is almost like hard to read because there's like 30 bullets on the slide, but they're all good bullets. And any one of those bullets may matter to you because it's the feature you were waiting for and was talked about six months ago, but now it's been released. Or it's been released with some new capabilities that weren't talked about before, kind of thing. So I'll try not to wave my hands in the air too much as we do it, but just as you're going through this, you may have a sense of there's a lot here. Yes, there's a lot here. We'll try and signpost along the way. Anything else from a opening kind of keynote and payload perspective? Because I'm not reading off all the items in each category. We'll just do that in the category. I mean, I know we're kind of recapping for the folks who joined us in Las Vegas, but we'll have follow-on events. And the thing that I just want to kind of remind us all about is the actual event itself that spawned this conversation. It's something we do every year. It's a great opportunity for our customers and our developers and our sales teams and so many different parts of our organization, all to come together in one place. We have such, I always come away from that invigorated, having great conversations, getting, honestly, some new insights and ideas from our customers. And if you haven't taken advantage of that before, certainly join us when you have the chance. It's like there's enough time to plan to be there next year. Is that what you're saying? Absolutely. There is. Cool. So, shall we? Let us dive in. Overview, payload-ish, and then the big section here is going to be Unified Data Plane, and it should be. And if folks are thinking like, hey, where is all the stuff that I know Pure for historically live, that largely lives here. Not entirely, but let's be real, we're not running away. And this is actually a great comment from Kaz even around the rebranding. There was an interview about bringing from Pure to EverPure. Hopefully by the end of this, you'll have a sense of, candidly, why there was that shift. But we're not running away from the stuff that got us here. And if anything, we're continuing to dig in and invest there. And I'll just say, we'll illustrate that. Actually, what is on the table, I don't believe there was actually anything in this payload about direct flash modules. We've got an older one here that has a nice bezel. It's one of the first times Kaz didn't kind of show up on stage and wave one of them, too, which he's done historically. But you could even put that a little bit, too. Am I allowed to say Nandpocalypse? I don't know. Or AI, all the stuff that's going on, right? So we are continuing to push there. But actually, we pushed as far against some of those boundaries, as makes sense, but we're continuing more with FlashBrain, FlashBrain. Okay, so items that weren't here. There were three major dimensions you can kind of group these into, things that gave higher performance, things that increased efficiency, and things that help with cloud. And there's even some other themes that we may bring up here. Like there's been this whole thing with this fragmentation, diversification of the VMware landscape. Now, the virtualization landscape, because it was really just the VMware landscape. So some of these is hopefully not a laundry list, but you can kind of think of all these been performance, efficiency, cloud. And maybe with the asterisk of a bunch of these are how we're helping customers with all the different things. Like a workload thing, you know, handling the workloads. So the first one, and this is definitely performance, is we talked about six months ago, eight months ago, FlashArray XL190. And this was actually a release, an announcement, and now it's out of the newest and largest FlashArray out there. And this is actually, we did a session on this, actually. You pulled me into this, Robin. I did. With Paul Joyce, and with Nathan, and with Brendan. And this was not necessarily about actually having the resiliency that you usually need at the top end. It's about giving more performance in certain ways. The two ways that did this is it has a very large read cache. It's actually a successor to what we used Optane in the past with direct memory modules. There's even a little bit of an interesting thing there. We made an evergreen commitment, like a legal commitment with evergreen, that you buy into this and we'll have a pathway for you. So we satisfied, even though the Optane thing didn't work out, that's not an us thing, that's an industry thing, Intel. Like we gave people a pathway to continue having that capability. As well as then we spent, and this is where we spent a lot of the time in the session around this, around reimagining how you could take advantage of the second controller in a modern approach. Not with, candidly, the drawbacks you may have had in the past with dual controllers of having to watch each controller and keep it balanced and stuff bursting and having performance impact when there's failover. But how can you do that for kind of optional workloads where we let you intentionally unlock some of the performance of the second controller, but being very thoughtful about how you do that in a way that doesn't backfire on you. In ways that, candidly, we've talked about a lot in the past, even 15 architectural decisions, like the dual controller architecture. We're not walking back any of that. We're thinking about what are the building blocks we have today, and especially in today's environment where people, you can't get gear very quickly. And things that were commodity before are now scarce and hard to get. How can we help you get the most out of what you have, especially at the top end? So that's a little bit vague, but I'm not trying to recap the whole session there. No, but it was a design decision, right? And for me, it is a demonstration of hearing our customers and seeing these corner case workloads and asking us, can we do something unique in the product while still retaining the hallmarks of Evergreen and other commitments that we have, and just help them optimize around a specific corner case. So, Andrew, I'm really glad that you took the dual controller conversation head-on, though, because, you know, you said it was a year ago we were here recording our architectural decisions. And absolutely, we absolutely made some bold statements around why we designed our controller architecture the way we did. And at first glance, it may seem like we're going back on that a little bit, but we're really not. What we're doing is we're honoring that architectural standard that we wrote for ourselves, and we're finding ways to leverage all of the breadth of the technology, because as you point out, these things aren't necessarily as commoditized as they had been. Memory resources are valuable. Controller resources are valuable. How do we make sure we're squeezing the most that we can out of those precious resources while still staying true to those architectural foundations that we planned for ourselves over a decade ago? The last piece here is I can see the slide I'm ahead of you, and it's like, we actually, when we announced this, we started to talk about it in different terms. Usually when you do a new product hardware announcement, you're talking in terms of throughput or IOPS or whatever else kind of thing. And so the metrics here, which resonate so much more today than they did six to eight months ago, was IOPS per rack unit. It was over 900% per, these are like big numbers. They're fancy numbers, otherwise we wouldn't talk about them. But just even the way we measured it, talking about IOPS per rack unit, IOPS per watt, and terabytes per RU, terabytes per rack unit, as a very different way to measure than the classic ways that we measure. But that's even more impactful today. Yeah, well, three-dimensional, but also taking into account some of the things that matter around environmentals as well. Space, power, cooling, also very important. Of course, we're continuing to work on new hardware generations, as you would expect, but we're not here to talk about that today, later. Or we'll chat back for that, maybe. Because I know he was even, we were actually at EBC on the road in Dallas a couple of weeks back, and he was talking about, he was just authorizing some of the stuff around this, because it always keeps happening. Okay, next one, then, is, so, FlashArray XL190, and that's in a performance kind of category, if you will. But then also on the FlashBlade side, if people are listening and they're thinking like, hey, that's not new, true. We actually, we've been talking about it, and we announced it, and we have customers who are using it. But we actually went with FlashBlade and announced some performance benchmarks. For those who are familiar, we don't often do performance benchmarks, but in this case, it made sense. So there's a spec benchmark where, I mean, I can read the numbers, 616,000 megabytes, that's the big B per second sustained throughput kind of thing, 7200 concurrent AI jobs. So these are actually noticeably higher. I'm not gonna read the press release at you, but this is where it made sense for us to go and do this. And the reason, anyone listening, like, why does Everpure, yeah, I did it one time, we'll see how many times, Everpure usually do benchmarks. Well, because we very much believe that we wanna do real-world workloads, and real-world workloads do not resemble benchmarks, except occasionally, or sometimes they do. So that was the logic behind this, because this was a space where the benchmark was applicable enough to dive into. I think, J.D., I even remember you were saying, like, your tagline was like, we don't do benchmarks, but when we do- We play to win. Yeah, I don't always do benchmarks. Oh, no. But when I do, but a great proof point, right? And Exa being something where we can go off and achieve these really great metrics, which is a good reason to go do big honkin' benchmarks, which provides, then, a great halo effect around the FlashBlade product in general, right? If we can do it here with this interesting architecture that we've built up, well, FlashBlade as well, right, which is gonna be a little bit more applicable to everything that people are doing, that's something that folks should check out. It takes work to do, for us, where it's not just, stand up a box, and then you run it, and it's like, okay, this is, like, the work of an hour. I'm almost, I don't have the precise number of hours, but it's a good amount of effort for that- It's people hours and money, right? I mean, I remember TPPC benchmarkers from years and years and years ago, and finally, somebody telling me the cost of the average TPCC benchmark to run being, like, a seven-digit number, right? Like, really big expenditure, but if you got something great to demonstrate as a proof point, why not? So, the next category, there's three or four here. Everyone's wondering, like, hey, Andrew's talking, like, yeah, JD and I kind of bounce back and forth in Intelligent Control Plane. You're gonna, it's gonna be more JD. It's the JD show, there. Yeah, so we'll bounce back and forth. He's here to look good for now, and then I'm here to look good later, so something like that, yeah. Okay, so, next theme is, these are, like, three or four together, but the continuing, I'm trying to be careful about the language that I use here, because the diversification, the fragmentation of the virtualization landscape, because VMware is still a very strong partner in the engineering level. We still have very deep, the deepest integration of any virtualization platform, and arguably deeper than any of our competitors, because that is even, like, that was one of the first use cases when we started the company kind of thing. But at the same time, we are getting a lot of requests in different areas, so we are investing a lot of engineering resources in there. So, in, well, no particular order, but roughly, we went GA with something we had announced earlier, which was, and I'm reading this to make sure I say it right, because it's a mouthful, Everpure Cloud Azure Native Virtual Machines. So this is the Azure service for VMware, and the key here is that while we've had, now, you know, don't get dizzy, Cloud Block Store to Pure Storage Cloud to Everpure Cloud, because once we had changed the name of the company, we needed to change it from Pure Storage Cloud. Okay, so the first time made sense. That's historically been something that you stand up in your own cloud environment with your own resources, but what if you say, I just want this to feel like a first-party service, where I go into my Azure VMware service, and I just want to consume the cool advanced option, the Everpure-backed one. That's what this is. So it's still the same underlying technology, but delivered and packaged in a much simpler way to consume, and that is a very tight partnership with Microsoft. It's not happened without a very defined process, and that's why we're talking about it here, because while you've probably heard about this for three, six, 12, 18 months, give or take, we are finally GA, because that takes some time, and it should, because of the richness and the integration that it has from an Azure Native standpoint. And it really plays into the idea that we're talking to different personas in the customer environment, right? We very classically have talked to data center owners who understand storage products and hardware, and when we originally moved into Cloud Block Store, now Everpure Cloud, there was very much this idea of building infrastructure in your cloud environment, but as we know, that's not typically how clouds operate. Clouds deliver services, not large data center products, so that conversation made a lot of sense for the original personas that in earlier days we talked to, but as we start to talk to more folks who grew up in the cloud, they want to consume these things as a native service, and I think this is a really powerful way to kind of see that evolution from storage product to data service, and that mirrors very nicely the journey that Everpure as a whole is going on. Well, there's also a lot of shops that are fully bought into the Microsoft ecosystem, full stop, right? It's not like they're just a little pocket. It is- I've got my Azure. We even understand my Azure and all the things that live within that ecosystem are things that I'm going to be using, and if you're not connected to that with a native service or something integrated, you're just not able to solve the problem, right? And, I mean, can dovetail to Azure Local up next, right? As part of that same ecosystem where we're continuing to invest. So I actually can spot on. There's a little bit of file. I'm going to come back because I actually want to embrace the virtualization theme. So flash drive support for Azure Local. Now, full disclosure, at the time that we're recording this, I don't know if we ever said it, Azure has been making some adjustments from a licensing and approach standpoint. I have heard that, yes. But the technical integration, the capabilities we're investing in, and we're assuming, candidly, there's a good partnership with Microsoft, so if you're listening to this and you're like, man, are there some recent, excuse me, are there some recent changes? Are there twists and turns? Yes, but we're continuing to invest in making the underlying substrate work. So FlashArray support for Azure Local, announcing that. Then, also, FlashArray with Nutanix. So that's not brand new, but that is in the category of we were the first from a block storage standpoint to support Nutanix. There was a little bit of interesting SDS stuff out there that wasn't really mainstream, if you will. They did it initially and they needed to, kind of thing. But as you would expect, as Everpure, we partner with more companies than Nutanix. As Nutanix, they partner with more companies than us. So let's be real. They have just announced integration with some of our competitors, but we fully intend to not just have been the first, but continue to be the best with the most robust and deepest integration. So this was about actually bringing in some new things, pushing that further along, so announcing that you can move VMware VVOL workloads directly to Nutanix very easily, if you think about what VVOLs are, it's the file system natively inside the volume about the VMFS layer in between. It probably makes sense how you can do that, but it'll still take some work to pull it off. Change block tracking, or the acronym is CBT. If you look at the backup or the data protection or resiliency, however we put this side in, or resiliency. if you only are transferring the changed blocks, that can be an order or two or three of magnitude difference. Like that can actually change, can I protect this workload versus can I not kind of thing. So doing that, and then also in having more integration of managing FlashArray directly from Prism. And I really like that just because it goes back to one of my favorite stories from even the first year I was here, Ryan Oller, who's in our training organization kind of thing in the customer experience group. He told a story that stuck with me of a customer called into pure support and forgotten their FlashArray password because they'd managed it FlashArray out of the VMware plugin for so long. We literally didn't even remember the password to get in. So I don't want to say that we're necessarily there, but like it's, it's changed that that's the philosophy. And even so, JD, like you were saying, we want to bring it to how people want to consume it, whether from a cloud standpoint or from a virtualization standpoint, you're in Prism a lot all day. You can do great stuff in the FlashArray UI, but you shouldn't need to be there all the time. So we want to expose and continue exposing more up into the native element manager of the virtualization platform. Well, I think that's a theme, right? Meeting our customers where they are, you know, later we're going to talk about our evolution with the MCP server. And it's all about how do we, how do we unlock your workflow to be more powerful, not force you into a workflow that, that we've designed. So I think the last piece in here from a, and now this is a little bit of a blend of the virtualization. And then we have a couple other items too, but from kind of the virtualization theme is we have continued to pour engineering resources into Portworx, which for those who are thinking, this is in the context of it's a software-defined data plane or software-defined storage in a Kubernetes environment, right? Kind of thing. So of course there's next generation applications and we can have the whole discussion about the holy war of stateful versus stateless, which is mostly over at this point, or it's just, we know which way it is. So Portworx helps support stateful applications where data lives inside the container. But then you have this whole aspect of the VM, the virtualization landscape of this thing called Kubevert. And if you haven't seen this from your ever pure account team, or I'm pretty sure there's a blog with this in it, I kind of close my eyes and see this table of if you do Kubevert and you're thinking you want to run VMs, here's the capabilities that are in there natively. It's like five or six rows, if you will. And then you get to some things that you just expect would be in a virtual environment around like storage vMotion and things along those lines. And it's not built into the Kubevert layer. And that's not me critiquing Kubevert, it's just what Kubevert is. So Portworx brings in, Portworx Enterprise plus Kubevert brings in the abilities that you would almost just expect as table stakes for an enterprise virtualization solution. So we continued to actually similar to Nutanix one to help people to be able to move more quickly to Kubernetes and Kubevert quickly. We also added in some multi-tenancy pieces that matters in enterprise environments, whether it's regular next generation applications or not. And then actually did some more around insights. I realized that's now like 60 seconds on what you could have. And I can actually feel John Owings and Nathan Wood and some other folks be like, man, you only did 60 seconds. Yeah, it's worth a time. Yeah, but this is also one that we've had great air cover in the last two accelerates, right? Two years ago, SiriusXM was up there talking about their journey, their very successful journey, relying on Portworx, relying on Kubevert and OpenShift specifically. And then this last year we had CSX, right? Transportation, logistics, but most people think railroad, which was kind of fun. We almost had Nicky go out with a railroad conductor type of hat for that interview. But that was another one where these organizations have gone full in on this technology for shifting VMs into containers and it's worked out swimming for them. But it is that combination. My son loves trains. So the keynote about 30 seconds into the CSX kind of infomercial about who they are and what they do. I actually started recording it and then sent it back because I was like, yeah, I think he's gonna like watching this. Like, hey, I'm doing like data technology thing, but hey, there's a cool train video about the links. But also really interesting in speaking with Eric, the customer there, about the notion of managing data for all those trains and all the scheduling and when tracks switch and managing the logistics and what's getting shipped. It's all a data-centric approach. Really kind of cool. So changing gears a little bit. Only three items left in this section. That's kind of all the virtualization-related pieces. But like, for instance, the Portworx one is just about virtualization. But we did also talk a little bit about the way it was tagged was EDC, Enterprise Data Cloud, for file and object, which does really mean more capabilities that we're doing in FlashArray and FlashBurger around file and object. So this is where we were making sure to highlight for folks that maybe not have been paying attention because this is why we do a little bit of maybe inside baseball again, why we do two major launches a year. I know when I was a customer, I didn't pay attention to any of my technology partners every single month. I paid attention once or twice a year, I think. So sometimes we'll talk about things in an announcement that have been out for a little while because, hey, we don't expect that you've got a day job. You've got all the other stuff you need. So, you know, we were highlighting the release of object for FlashArray. This is a file and object category. This is not because object on FlashBlade isn't good. It's amazing. You want a huge scale-out, high-throughput object store or data lake, whatever we're talking about. That's FlashBlade. You want to have a fully unified block file and object. You can do that on FlashArray. And probably most often you're going to want to do that at the edge or some level of edges. Why do you just a little bit of object? Because if you really want object, you're probably thinking a scale-out platform, but that doesn't mean that we shouldn't offer it on FlashArray as well for specific use cases. This is an intentional, not like, oh, you can't figure out where to do object. No, just like with file, we offer file on FlashArray and FlashBlade for different purposes and understanding the use cases Perhaps some fusion implications around that as well, right? But you make a great point though. We had a session with some of our navigators, a group of our customers who come together to give us insights about what we're thinking from a development perspective. And that unified approach you talk about, Andrew, is one of the big pieces of feedback they've consistently given us. But yes, I understand that FlashBlade from a scale-out perspective is a beautiful thing for object. But if I just need a little bit at an edge site, I don't want to bring in a whole separate platform for that. A scale-out architecture, definitionally, usually only scales down so far. That's not a critique of FlashBlade or any other scale-out architecture. It's the approach that you can take with a scale-out versus scale-out from a minimal size standpoint, almost, if you will. On the other two items, and we didn't highlight this as much at Accelerate because it's a huge deal and I think will be a very big item in the upcoming Accelerate London, is around active cluster for file. So this is around synchronous replication for file. Synchronous replication, not quite as common in the US. I actually remember when I was working with a competitor's product that was a metro-y, cluster-y thing back when I was on the partner side. I actually, the smaller part was like, I was getting help from people in Europe. And I was like, that's weird. I was like, oh, because all the people who know this kind of technology are over there. Okay, that's interesting. So active cluster for file. And then last but not least around file and object, specifically really file. Continuing to enhance capabilities around multi-tenancies. So this especially plays for either MSPs or large enterprise companies where you act as an MSP internally kind of thing. So continued richness on those. Two items. The second to last is going to be super short because it's just highlighting EverPure Data Stream Services. We'll come back to this later, a little bit more in the universal data intelligence, but we're highlighting it kind of a platform level, if you will, or data plan level. This is now GA, so officially GA. This is a partnership with NVIDIA. As you think about, you need to have the right data. That's a lead in for data, foreshadowing for data intelligence soon. But then you need to think about how do you vectorize that data so an LLM can consume, you've got all your unstructured data. So that's where Data Stream Services plays. And that is now GA. So more on that later. But just referencing, it kind of fits in multiple sections. So I want to highlight. It spans, yeah. Lynn, would you say it also fits in a little bit with that data primacy messaging we talked about earlier, which is I'm having a universal view of this data that now can be used in applications down the pipeline. Bingo. Last but not least, absolutely not least, is a major enhancement to Evergreen One. So Evergreen One being our true storage as a service. You know it's a true storage as a service because when you finish the contract, there's not a dollar buyout or fair market value. Those are the phrases you know. It's a lease dressed up as a service. Nope, it's not that. Okay, and there's all the SLAs and not SLOs. We stand behind the SLAs. There's actually financial teeth on them kind of thing where if we don't meet those. Okay, so Evergreen One, true storage as a service. Something we've had from the very beginning with Evergreen One is the idea that you can burst from a capacity standpoint. Sometimes I'll use the phrase just to even kind of help people think a little bit is if you're not ever above your commit, you're actually paying too much because you should periodically be bursting above your commit. That's the benefit of as a service offering. That's historically been on the capacity side. And performance, there's different tiers of performance. So we introduced what we're calling Evergreen One Overdrive and that is kind of the analog from a performance bursting standpoint that you can, and I'm not gonna try and lay out the exact details that you'll see in the FAQ around this kind of thing. You can actually burst at a performance level periodically and that actually stays within your monthly bill. So just like you can have capacity bursting, you pay someone on demand, you can have some level of performance bursting, it's on demand before you necessarily have to move up to the next tier. And of course, if you're bursting enough, there's gonna be a break point where you run the math and it's like, yeah, I'd just move up to the next performance tier, you know, kind of thing, giving a buffer there. And look, we've thought this is important for a long time now, but with the conversations that we're having with the shift in memory prices, forcing us to rethink how we leverage and how we optimize taking advantage of the infrastructure that we purchase, this is more important than ever. Look, you know, you and I both grew up as SEs and what was the common approach? Well, let's plan for four years of growth, let's plan for a little bit of capacity buffer, you know, we're gonna go a little bit over here, let's make sure we've got some extra capacity and, you know, it was a normal course of order to oversize because you wanna plan for the future. You don't wanna be too high, but a little high. You don't wanna be coming back and asking, your finance lead year after year, well, I didn't plan for enough, so now I need more, more, more. And we've really, that is a very wasteful approach, where I'm buying, I'm investing so much upfront because I might need it in the future. And so really rethinking the way that we are optimizing for what we need today, but giving ourselves the flexibility to really go and take advantage of that increase in capacity or in performance as we need it is such a powerful thing. I think all the time, but especially now with some of the cost conscious conversations we've been having. Yeah. It's about absorbing demand spikes without either, you know, emergency procurements scramble or even kind of a capacity scramble, because it's there and still mapping to that paying for what you use. And I was gonna end with that, I mean, truly a pay as you use is the core hallmark of that, but now having this flexibility where it isn't that, you know, finger in the wind kind of guessing game. It's there if you need it, it's a risk thing. And I love that you pointed out if you were consistently going above that threshold, let's talk to your CSM and then see if you need to, you know, move up to the next year. Cause it's natural that data usage is gonna grow over time anyway. So look into that. Yeah. The thing about the unified data plane, I still wanna say universal, I can feel it, but you know, someday it'll be really universal, but it's unified. Getting closer every day. We got object on flash right now. Yeah. It keeps coming. So performance pieces, we hit on that a good bit with Exa 190 and with Exa. Broader cloud flexibility and even cloud, I may be like into a little bit of hybrid cloud and the virtualization landscape kind of things too. Efficiency, that's Evergreen One Overdrive. There's efficiency within the various products as well. So hopefully this gives folks a sense, even as we're kind of through part one of the three-ish payload sections, if you will, is that we're continuing to invest in the layers that have got us here as a company, what people often know us for, even as we want you to do more in these other areas too, but we're not taking our foot off the accelerator on what we're doing at this layer. Because there is, candidly, so much more market share that our competitors kindly donate to us over time. It may be a little bit unwilling, but you know, at the same time. And this is where I think we all have friends in the industry. So as long as, unless we're competing, we all know each other and we like each other kind of thing. And then like, hey, may the best man win, best woman win at the time. Well, with all these, and maybe just to put a bookend on it, I mean, you manage a team of a number of technologists who are out talking, right? And I imagine over the last couple of months since Accelerate, where have you been seeing our customers or even prospect engagements find the most interest around this? Or has it just been more broadcast? Like, have there been a couple that have risen to the top where there's just, oh, we're really excited. Like, tell me more about that. Well, I'll be honest with you, Rob, it's very much the where we're going, right? So, you know, the foundational pieces are important, the unified data plane, and the fact that we're doing more, continuing to invest in those products is great. But where we've talked a lot about shifting from a product focus to a solution focus, and I'm really starting to see our customers in the way they interact with us, in the questions they ask, start to shift that mindset. So it's less about how much memory, how many CPU, you know, what is the spec of the box that you're shipping? And it's so much more about how is this going, here's my business outcome that's gonna drive me forward, what are you doing to help me deliver on that? And so the thing that our customers keep asking us about are what's new with data intelligence? How is that fitting into the story that you're telling? What's new with the intelligent control plane? How are you taking those products and putting them together in a way that's making my life easier, right? Those are the conversations we're having. Love that, right. Which also means that they see confidence, right? Which also means that they see confidence, right? That these are, you know. It's a little bit of validation. Evolutionary, iterative. They're no longer kind of questioning, you know, the architectures, the controller model, the evergreen. They just go, yeah, that works for us. Okay, my business really needs to understand the context of data or it needs to automate, right? I had Mike Dahan on the podcast and he said, we use Fusion because it helps me scale my IT team. My IT team, I do not need to add more people. I can do more work with the existing team that I have. That's huge. And the reason I'm glad we still have the 15 Architectural Decision Series out there last year is because there are folks that have been running competitive solutions for years and they're not tracking some of the landscape, nor would I expect them to. Like, they've got a lot of stuff to do. So sometimes the one to two to three to five year ago discussions are still the ones that we have with certain people as they're actually now at the inflection point of let me reevaluate. Like when the facts change, I change my mind or I'm on a three or a five year cycle, et cetera. But even echoing JD a little bit, even with two ways, one is with some of the classically, the folks that we'll chat with often, these are admins and engineers, et cetera, kind of a higher level of trust that you've got a lot of the things that we would need covered. So we need to sometimes dig in at a feature level, but we can have a larger outcome solution focused conversation because we know, we believe that you won't actually fall out with all these items you can't do down here if we start that way. So if it's a startup, we've got to start down here because if we don't, you have so many things missing, it's not even worth starting a solution or outcome focused. And it's validating yourself too, right? It's like, hey, when you're in startup mode, it's very much, can you follow up on the things that you say? Can you deliver on the things that you say you can deliver on? We've spent enough time showing folks that we can do that, that we, I think we've earned the opportunity to have those higher order discussions. Yeah, and then as those folks often in a respectful way, pull us into discussions with their director or their VP, those are folks who won't engage unless they can be at this level and their teams validate that, hey, this is a vendor technology partner, whatever term that they use, that can actually engage and help us. And then it starts to turn into more of a larger business discussion, more of a larger outcome discussion. We're talking about fleet level type capabilities. And I feel like if I keep going, we're going to go right into the next section. That's a great segue. So it's really great time to wrap on this one, on this part one. And it seems like it was the right amount of time to get into the not universal, but the unified. Data plane. Data plane, we're going to not edit that out in post. We're going to leave that just because it's natural. We're human. And it's fun and we're human. And there's a lot of, you know, we use both kind of interchangeably, but that was a really great summary and a good double click because Chad and I covered this kind of at a higher altitude and so it was nice to get your field perspectives kind of double clicking on it. And we're going to come back. We're going to click off here for everybody. And we're going to come back and jump into the intelligent control plane with the connections into universal and ever pure data intelligence. And maybe kind of bookend that with a little bit of the EDC with the enterprise data cloud. It's going to be a good, a good second episode. Thank you both. It's been a thrill to have you both, but it's just natural. It's just easy. It's just easy with you two. Super fun. And hope everybody else out there that you got a lot out of this. Thanks for watching this first parter of our special two part series. As always, tell a friend, tell a colleague, and well, keep the greatest guests on the planet coming onto the program. Yeah, we've done so much together. How could I not say that about the two of you? But thank you again, both of you. We'll be right back. You'll see the second episode of this debuting a week after this one. So check back on YouTube and the normal places that you consume audio podcasts. And with that, we will wrap forever pure for my good friends, Andrew Miller and J.D. Wallace. This is Rob Ludeman saying, don't look back, something might be gaining on you.
