59:47 Webinar
Ask Us Everything about Evergreen//One
In this episode of Ask Us Everything, we’re diving into Evergreen//One™—our storage-as-a-service solution that gives you flexibility, protection, and cloud-ready capabilities.
This webinar first aired on 11 December 2025
The first 5 minute(s) of our recorded Webinars are open; however, if you are enjoying them, we’ll ask for a little information to finish watching.
Click to View Transcript
Yeah, he is. All right, we're at 11:03. I'm thinking it's probably a good time- He is to get going. Yeah, Ta- Tago's with us now. Look. Tago's coming in. There he is. Hey, Tago. How are you? Hey, Tago.
I made it. I made it. Tago made it. Yay. Hooray. Oh, how you all doing? We- Good to see you on the- We have a party. We have a quorum. Oh. This is exciting. Now you can see this. Ah, perfect. There we go.
Yeah, you look good. Yeah. It, It's, it's the best technology. Took me a second. Had to boop, had to come up, but I made it. Sorry I'm late. We do what we take- No worries. Stay awake to, to get us on these Ask Us Everything.
So we were working, Abe and I were working the room. We've got quite a wide- Have, have we already got to, aluminium and aluminum? Yeah, well, we, we brushed up against that. Okay. We have a clearly international audience. We have people from Greece, India, possibly Taiwan, China, somewhere in Asia-Pac.
So- Even Florida. We- yeah, and Florida. And Florida. Of course. I, being the Florida man- I'll lead the charge on that one. So, let me kick things off. Hey, everybody out there, thanks for joining us on this Ask Us Everything on December 11th.
I'm Don Poorman. As you know, I'm usually the host of this thing, except the last time my good buddy Rich Barlow hosted it. I heard he did a great job with regards to that while I was off at some show talking to somebody. But anyway, I'm glad you guys were able to join us today to talk about Evergreen//One.
We're joined today by Tago Navon, who's our Director of Product Management for Evergreen One, and Abe Barnes, who is our Vice President of Subscription Portfolio. And before you guys roll your eyes and, and think that we're just gonna talk about the financial benefits of Evergreen//One for, you know, what your role is as an owner-operator of storage, we're actually gonna put it, a different spin on it.
While there may be some advantages to that as far as financial stuff is concerned, Abe and Tago wanna talk a little bit about outside-the-box thinking when it comes to Evergreen//One. So they're gonna establish a baseline of Evergreen//One, what it is and what it isn't, and then ultimately roll into a story of how you as somebody on the front lines can leverage Evergreen//One to
look like a hero, right? Cause at the end of the day, we want you to look like a hero. We want you to get that promotion, and if you ever wanna rise up the ranks to be a boss, this is kind of the stuff that we think you can leverage to get there. We even had a guy at Accelerate earlier this year who works for the Mississippi Department
of Revenue say that by leveraging Pure Storage and working within all of the stuff that we have available, he rose up the ranks, and he's now the Chief Technology Officer for their entire Department of Revenue. So Abe, Tago, it's in your court. Make these people heroes.
Let's have heroes out of this one. No pressure then, Don, right? None at all. None at all. It's all yours, guys. Take it on. Thank you. Well, I'll jump- You, you go first, Abe.
Well, thank you. Yeah. first of all, it's a pleasure to be here, and it's, wonderful to see, so many our list of attendees here who have chosen to take time out of their day to, to spend with us. Thank you for that. It's a privilege. My name is Abraham. I've been at Pure now
about six and a half years. Prior to that, spent about 20 years in enterprise organizations and in GSIs building as-a-service models and solutions. So that meant I worked mainly with, like, STaaS-type organizations, IT service management teams.
And I spent most of the time with my, in the GSIs I worked at designing and leading, quite large, complex business and technology transformation programs for large businesses like Aviva and BP and Rolls-Royce and E.ON Energy and Deutsche Bank and plenty of other people. I guess, just before I hand to Tago, my, my personal feeling about where this conversation
STaaS is more to do with thinking about what could be achieved with a public cloud-type service but with m- more locations and control and things like that around it. I won't get too, into too much detail, but that's where, that's where it starts for me. There's so much that can be done.
Tago. Yeah. Thank you, Abe, and, I'll, I'll do my quick little intro time up here as well. I go by Tago. I think, that's, that's how Don as well, so Santiago Navon is what you're gonna see up here. Call me whatever you want. I've been at Pure, seven years, similar to Abe.
So we, we both joined the charge at, around the same time, but very different, starting points within Pure. My background is in engineering, in, you know, electrical engineering, hardware engineering. And that's what I did at Pure my first few years, right? So when You know, the, the framing for this conversation and our focus in tech,
really is there. We Our premise for Evergreen//One is a premise of taking great technology and making it shine for our customers, right? So when I approach, our Evergreen//One portfolio, when I approach conversations about Evergreen//One, frankly, the financials are a side benefit, just like Don said. The, the core benefit and the core, value that we're trying to drive for our customers, the
way we're trying to help customers is by building the best product for storage, that we can, right? And the best product really is holistic. It, it starts with, with great technology, with DirectFlash and all the good stuff that, that Pure has built over, over the last, 15 years. And again, I, I used to be a hardware product manager.
I launched FlashArray//C and, and, the E family when those came out, right? So I understand the technology, and I understand how we can make it shine for people when, rather than talking about NDU like we like to do so much at Pure, right, and talking about all the cool things we can do, Evergreen//One is about making you somehow not care-About NDU.
It being so ingrained in the service that it's table stakes for you, and that's the connection to what Abe just said from a, from a cloud type of operation, right? The goal is to make it look and feel as if you had your own private storage cloud in your data center, right? Which operates kind of on its own, which just happens to have as much storage and as much
performance as you need. And Pure Storage in the background is, is taking all of its great technology, all of its great products, and making ensure they're, we are using them appropriately to deliver on what you're gonna consume next. Yeah, that's a, that's a great point, and one of the things that you said, Tago, brushed up
against something that I like to say is that I, I, I like to tell people that Pure Storage might be the first sol- storage solution you forgot you owned, right? And, and I think that really is one crowning moments that Evergreen//One can bring to bear, not just with, like you said, the NDUs and everything else, but extensions of it too.
Like, we already have questions coming in on the cyber recovery SLA and things along those lines. I'll let you guys talk on, but obviously there's a lot of interest out there from everybody to say, "All right, let me know how this thing pragmatically looks to make my world better." Yeah.
And actually, since we've And I'm sure everybody already knows, but it's worth clarifying what that TLA is that we've used a couple of times, isn't it? And it kind of links to, one of the very first questions I spotted in the chat as well, cause I think, our friend, Ken joined and immediately put a question in the, chat which kind of relates. So, so just in case no- people are
NDU, non-disruptive upgrade, the ability that our technology has deployed, perhaps change controllers, change interfaces, change the capacity in the form of the DirectFlash modules, change the operating system, upgrade it, patch it, all, while the business services carry on running, without any disruption. So can't be replicated or emulated, else in the industry, not to the extent that
we can deliver anyway. And that underpins what Tago said about that's why we think of this thing, and when it in the form of Evergreen//One, why of it as more a cloud experience. Because you take that capability and the technology, and all of a sudden you've got a, a non-disruptively scaling cloud that can preemptively and proactively, adapt and,
a- and mold itself to how the consumer is behaving, on it. It will automatically expand, it will automatically increase in performance, all while business services, run online. So Ken, I don't know if that exactly hit the question you were asking, but it seemed like a, a neat segue, and I'm sure, if we wanted to go even more into detail, Tago is our, is our
expert on the call. Yeah. And, and thank you for keeping me honest on those acronyms. I like, we like to throw acronyms at everything. I should, um- we should fact-check each other on NDU being non-disruptive upgrades everything else. The, the reality is that as a, again, an
engineer first, technical user first myself, right, what, what we're trying to, to give you, as an option, as a, as a brand-new thing which frankly doesn't, doesn't exist from anybody else, right? The ability to have your own private little cloud that is managed by, by somebody else in your own data center, same, same, performance that you expect from on-prem infrastructure, but w- virtually unlimited in
a way. To me, it's a huge technical value, and right, and, and it's a, it's the difference Again, hardware guy. I love hardware, but it's the difference between being constrained by the limits of a specific piece of hardware versus being able to seamlessly move across different and multiple hardware platforms to achieve whatever you need.
So when I talk to, to our users, the conversation is just as technical in Evergreen//One as it is when we are about an individual, you know, array the, the great things it can do. The difference to me is what are we focusing on, right? Are we focusing on, things like CPU and memory?
No, we stopped focusing on how many cores the CPU has years ago, right? Nobody really cares anymore how many cores a FlashArray CPU has as long as the FlashArray performs the way it should. We're taking that one step forward, right? And we are, moving away from asking our customers, "Hey, would you like to buy product
model A or product model B?" we're taking that responsibility on ourse- selves and letting then y- all of you as our users just deploy effectively services directly on this private cloud. So you're still, you're still gonna have to specify how much performance you want, right? How many IOPS, how much bandwidth.
You're gonna have to specify, the replication policies. You're gonna have all of the same controls you have today, but what you're not gonna be constrained by anymore is, do I have the right piece of hardware underneath, or do I have enough pieces of hardware under, in my data center to make this work?
That's what we're doing with Evergreen//One. We're really converting or moving your focus, hopefully, from the how, from a hardware perspective you can achieve all of the requirements you have, to the requirements themselves. You enter those into, effectively Pure1, and the hardware is Pure Storage's problem for,
for lack of a better term. So from a- Two from a variable perspective, Tago, we've got two thinkYou know, come to f- the front of my mind, obviously, like you said, performance matters. You know, method of storage matters, right? Scale up versus scale out.
But if we're just talking straight up capacity as far as that's concerned, w- calculate effectively used capacity and, you know, the buffer that comes with the subscription? You know, how are they notified on stuff like that? Yeah. 'Cause storage's a very dynamic environment, right?
And while we're taking on a lot of the glove stuff with Evergreen//One with the subscription, there are certain ground-level things that have to be aware, right? Absolutely. This is a great, a great starting point for our conversation on what does it mean to really operate like a cloud. And I think that's, that's the right, or the easiest way to, to compare the experience you
get with Evergreen//One, to other things you may already be familiar with. So specifically, let, let's start with talking about data, right? When you put 100 terabytes of data on a cloud service, right, nobody's there asking you, Hey, can you please estimate your data reduction for us? Hey, can you please, tell us whether it's this many objects or whatever else?" Hey, a
terabyte is a terabyte. You deal with it, cloud provider, right? We take exactly the same approach in Evergreen//One. We are already We understand our technology. We understand the data reduction capabilities of us.
So think as Ever- of Evergreen//One as effectively guaranteeing the data reduction that we know we have to you as an end user, right? When you put 50 TiB of data on it, it's our problem effectively how much space it takes, just like if you put it on a cloud service, it's the cloud service's problem how much space it takes. All we care, all we then bill you on, all we,
measure size on, everything we talk about with our users as far as data goes is what we call effective data, which is how many, how much data did you actually put yourself, right? If you went to, if it was files, you would go right-click Properties and know, fi- how big the files are. That's the only number that matters to us.
And how do we actually handle it under the covers? I think, Don, you, you snuck something in there, which is the concept of a buffer. Yep. Evergreen//One comes with a set of guarantees, a set of SLAs. They're public on our website.
We talk about them very freely. It's a set of, deliverables to you in everybody's contract that says, "Pure Storage must do X, Y, and Z for me or else, they have to fix it, number one, or they have to pay you, number two. Buffer capacity is one of those things. How can we possibly make a, a, an FlashArray, a storage platform feel,
feel unlimited or virtually unlimited like I was saying before? Well, we do that by always having more capacity on site than you are consuming at any given time. Not only do we have more capacity on site than you're consuming, we also are projecting utilization over time. You've seen some of those projections in Pure1 yourself probably.
But we're using those projections to know when are we gonna have to deploy more capacity, how much more capacity should we deploy, and how can we schedule that work within timeframes that work for you as an end user. So as a very practical example, what we're gonna be doing is looking at the amount of data that you're putting on progressively, drawing a trajectory for how you're growing.
We're also gonna take input on how you told us you're planning on growing, and always gonna be deploying ahead of schedule to make sure that we don't run out because we cannot, as part of one of our guarantees, possibly run out. And the same applies to every dimension, right? When we talk about performance, there's no longer a question of, "Hey, what is the effect
on performance of enabling replication? What is the effect of performance on enabling snapshots?" There's no such You have in your Evergreen//One service a performance guarantee. Performance guarantee means that if it doesn't go that fast, we have to fix it, right? Mm-hmm. You can enable every feature on, you want in
the world, and you should expect it to perform exactly the same. Why? Because once again, we will have sufficient infrastructure based on our projections, and we will always be upgrading that infrastructure, obviously consistent with your own schedules, to make sure that there's plenty. Let me- I'm gonna turn
it over to, to Abe maybe. Oh, go ahead, Don. Let, let me throw a curveball in real quickly though, because you were on a roll. And, and Steve Gray asked a question that I think is kind of interesting because he wanted to know about the buffer, but the thing he put at the end that I thought was really
interesting is, okay, I'm subscribing to you for capacity IOPS in terms of contract, and it's running fine. I suddenly have this new use case that causes a spike in my usage, but here's the curveball. I subscribe to you with FlashArray, and I've got this new AI/ML initiative that scale-out would work better for.
So can I roll that subscription of a FlashBlade type of environment into the master contract? Yeah, ab- absolutely, and Abe's gonna be the expert on this as well because Abe, Abe deploys with our, with our customers and our users so many of these very, let's call them multi-platform, multi, license type of contracts.
We have, before I hand the, pass it on to you, Abe, we have, services, licenses we call them, but services that align to a number of use cases and performance profile. We try to make it easy to deploy new things. So we try to make it easy for a user to come to us and do exactly what you said, Don, say,
I have an AI workload coming on. Can you give me Can we extend this subscription to cover this AI workload?"Very easy. We actually have a service specifically for AI, and it can be added to existing subscriptions, right?
We add, additional, data projections for how much data you expect from AI or no at all in the case of AI/ML. We add additional performance, guarantees for your AI workload, right? Maybe you'll say, "For this AI/ML workload, I need 100 GB/s of throughput, guaranteed, because my GPUs are expensive and I wanna keep them u- busy." We just add that
to the existing subscription, and we adjust effectively the subscription as your needs change, and that's the beauty for me of something like Evergreen//One, something that keeps morphing, keeps adjusting to where your needs are. My biggest thing, and I don't know how you think about this, Abe, but my biggest ask of customers is to not Don't try necessarily to predict where you're gonna be in three years.
Because if you ask me to predict where I'm gonna be in three years, I'm gonna have an assumption. It's not gonna be a very good one, right? I would rather have a discussion about the reality of a customer's data today, say, asking my, our users, "How much data do you have today? What is coming online in the next six months, and what do you want to, to have us, handle
for you?" And then know that you're gonna keep adjusting this without wasting investment, without having to go, you know, justify to somebody, "Hey, why did we buy the wrong hardware platform?" That's not gonna because the same subscription keeps morphing to where you're gonna be in one, two, three years. Well, that, before it goes over to Abe- Yeah that's- Hey, we are not let- we're not letting
Abe talk. It was about- No, Abe's next. I promise. I promise, Abe, we wanna hear your funny accent. So this is hero possibility number one, right? We're talking about how frontline folks can be heroes, right?
And through Evergreen//One, if there is a specific SLA that we're delivering and another workload shows up on the horizon, the best part about somebody on the front line with Evergreen//One is if the boss comes to you and says, "This is a new workload. Can we handle it?" Your only answer is, "Sure. I just gotta let Pure Storage know," right?
And Abe, that, I assume you can expand on that. I mean, sure, I started with this, now I need this. Make it happen. Well, well, this part of the conversation actually funnily enough is why I joined Pure. Because the time I spent in GSIs, but kind of consuming organizations as well,
gonna, I'm gonna bet the experience I had is kind of similar to a lot of the audience, listening to this conversation, and that was whether I was in an organization, you know, building service contracts for customers or whether I was in a consuming organization, it, it all still came down to what Targo just said, the need to agree on this long-term forecast of demand a- and then sign up to that, and then have some essentially rigid
infrastructure deployed, and we all hope that it lasts, long enough. But the problem was it wasn't really about the equipment itself aging or, or becoming less stable or less performant. I mean, yes, those were considerations. The bigger thing was what if the business strategy changes?
What, what if the direction changes? That's the point, right? You s- you sign up to a five-year f- horizon, right? Whether that looks like a contract or an investment, whatever it might be, you sign up to, to a horizon of that long.
You get into year two and the, and the priorities change, and then what do you do with what you signed up to first? And that's the point. That's why I joined, because what I saw in, in Pure was the ability for us to create the scenario that, that Targo just, outlined. And to your point, Don, yes, that means that you can pivot and adapt and change how you
meet the demand of your own lines of business and your application owners who are out there on the front line of each of your organizations trying to keep your own company relevant and current in the market and respond to changes in competition and, and changes in legislation and innovation and things like that. I do feel like we should probably maybe clarify a couple of things just to, 'cause you
know I love a definition, right? Mm-hmm. So we should just You, you said at the top also, Don, that we'd be talking about what Evergreen//One is and what it isn't, and I think we should be very clear, about a couple of things that Evergreen//One is- isn't, that it i- is not. It certainly isn't designed to be a kind of outsourcing, service of any kind.
Nothing that we, design into this STaaS as a service offering that we provide is designed to displace anything that a consuming organization does. Um- So, so to that point, Abe, because this does answer a question, does the storage admin maintain access to the fleet and everything else? We're just behind the scenes, correct?
Yeah, 100%. Okay. And that, that, that was really the point, that the It is not a black box. So we, we, we, we, emphasize things like we don't want you to worry about that doesn't mean it's hidden away or it's a black box solution that you can't have access. The point is that a consuming organization still performs the storage admin BAU part of
running that service. What Pure is providing is the foundational storage services, the block file and object storage services using equipment that we own deployed to hosting locations of a customer's choice, extended into the public cloud as well, underpinned by the SLAs that about and with this non-disruptive scaling capability that So that's the point.
It's there to be leveraged and, to be used by the consuming organization. But operationally, we would still expect the consuming organization maintain control, of where they place workloads and how to balance workloads across Evergreen//One instances, and still remain compliant with sovereignty, with security, and all of the regulatory compliance that might be under as well.
So this is, this is really about, for me anyway, and it might sound like strange language to use, this is solving the riddle of how to contend with, unpredictable demand-'Cause, 'cause I'll tell you, the, the, the, the time I spent in, in any of those organizations, the conundrum was always somebody somewhere had to sign up to a forecast of demand, and it never worked.
It, it was never, ever accurate, right? Right. By definition it can't be accurate. And so if you know you've got inaccurate demand, what do you do? And so our, our kind of philosophy was, if you can't fix the demand problem, fix the supply problem. Instead of, instead of deploying
fixed config, on-prem, rigid, may not suit the future requirements, may not that new AI project you have, may not, you know, be suitable for, for the demand forecast that turned out not to be right. Fix the supply so you're looking after the data that exists today, and fix it in a way that you can adapt it and, and bend it to what might happen tomorrow, or tonight even.
Mm-hmm. So you can just say- Maybe I'm wrong about it you can say, "Sure" . When the boss is like, "Can we handle this?" "Yeah, sure." Well, the- Yeah new, new R&D project needed, right? Sudden, sudden new, new thing happens in, you know, in the industry that one of our friends
on the call works in, they, they need an environment to develop something quickly. You know, what, what would, what would the alternative be? Well, if I haven't, if I haven't got the capability and the infrastructure I've got deployed right now to service that, I've gotta sacrifice something, or I've gotta with, with what I've got. Or I could run up to the
public cloud and get something. Perhaps the data I wanna use for that isn't suitable to, to go up there. Perhaps I don't have that capability, available. So, so that's the alternative. Not, not pretty. And while you're doing those things, you, you,
you know, every day that goes past, every month that, not deploying to that is, setting people behind. So, so to your point, and to the theme of, of what it feels like this conversation is, the ability to say, "Yes." All I've gotta do is email Pure, right? Bang, new instance ready to rock and roll.
Go for it. Maybe to add to that, Abe, it's, it's worth, I think, going through a very short example of the level of control that you maintain and, and the things that we end up doing. Because- Yeah y- you know, the, the, the basic question I often get is, "Oh, is this the same as, as cloud?" Well, it, it's trying to give you that experience, but it's absolutely
within your environment, your control, with your levels of performance that you expect in your location, right? So it's, it's trying to bring you the best of both worlds, and I think, a worth example of how we handle this type of thing is, is probably helpful. When we talk about admin tasks that we made with, with our users, right?
Everything that you're able to do with a Pure Storage product today, things like setting up replication, things like provisioning volumes, provisioning, file systems, setting limits, setting QOS policies, setting snapshotting policies, SafeMode, offloading snapshots. Anything you can think of is totally within your control, right? What, what we are really doing as part of Evergreen//One is turning that one array that
you would have done that, that in, into something that feels virtually unlimited. So what that- so that you never hit the edge, effectively- of what one box can do, right? And that's, and that's really crucial because if you The, the For me, the most interesting benefit of going into some- onto somebody else's platform, onto a cloud or whether or
not it's a service application, is that I never hit a limit, right? Theoretically. Now, we all know that when you go on AWS or wherever, you do actually hit limits sometimes. But, but that wasn't what we were sold. We were sold that you wouldn't hit limits, right? And that's what we replicate here.
How do we do it tactically, right? We're gonna be obviously letting you use the, the platform as you normally do. You would provision your users, access control, everything else. You'd have the same proactive support and front home capability that you have with Pure Storage product. So everything that you're familiar with is
absolutely part of it. Now, like I said, we monitor your First of all, we will always deploy more than you're using. Remember, you're only This is the one financial argument. You only pay for what you're using, right? We won't bill you for what we deploy, we only bill for what you actually use.
But in order to deliver a service, we always deploy more than you are using or projecting to use. And this is true for both capacity, available, available data you can store, as well as performance. What we are going to do is deploy a more powerful box than you probably would have otherwise bought, knowing that as part of your subscription you will only have access to a
subset of that, but you will have the, the ability to expand without having to change any of the hardware. So that brings up a question real quickly that's further down the list, Tago. Somebody asked about, you know, which FlashBlade is best suited for this- Yeah it's almost like, whichever one we think fits, right?
Like, I- That, that's exactly right. So we use, you know, a couple of, of behind the scenes, things that may help understand here. What we are doing, for Evergreen//One is generally more powerful products than our customers would otherwise have chosen to buy, and we're doing that at our expense.
We do that across both FlashArray and FlashBlade, and we generally will, deploy something that's oversized by, at least 25%, generally a lot more, knowing that the, the subscription that you run on that, on that, piece of infrastructure is fully controlled by what you're subscribed to, right? So the, the control, the access to that capacity and performance is driven solely by
your subscription, not by the hardware underneath. So what we want to offer to our customers is-Kind of as big an envelope as possible, to do that. Sometimes, I know the question was about whether we use FlashBlade//S and FlashBlade//E. Sometimes the focus is really on getting the most capacity in the smallest footprint.
E is wonderful for that. Sometimes there is, more, uncertainty around performance needs, and STaaS is actually a great platform for that, right? So we, w- we use both data as well as your projections to, to choose the best platform. But again, the choice of a platform never drives your cost because you're only paying
for the bit you use. Now, just going back to finish my, thought example. Now, eventually, you will probably need a hardware upgrade, right? How does that work? Well, as much as we- Yeah, we've got a couple- optimization as much as possible- We got a couple of questions out there, yeah sometimes
we'll, we'll have to upgrade it, right? Yeah. You guys have questions about that. So how does that work? Where Yeah. Where, where is your responsibility? What do you still do, and what do we do?
Well, we You still have full control of the when, right? We do not touch the equipment without your, permission. It is in your environment. You have the lock and key, and you can absolutely control that. Now, obviously, in order to keep expanding the infrastructure, we do have access, so we,
coordinate with you to find the best time to do it. Now, remember, you only pay for what you use. Do you pay for the new controller? Do you pay for the upgrade work? No, absolutely not. All of that is fully included, right?
So we will coordinate for you to find the best date to come do the, the NDU, the non-disruptive upgrade, thank you, Abe, the new platform that we think you're gonna need to, you know, continue growing and evolving over the next one, two years, whatever it may be. We will agree on a date.
We will send, personnel to perform that upgrade, with you. You will still be able to designate the right time if you have maintenance windows that you prefer. So again, exactly the same level of control that you would have for your own, purchased infrastructure. But we'll take care of it.
Same goes for software up- updates, right? You still have control over what version you run. However, you will be s- recommended the right versions for your environment as well as to get, to enable us to deliver every single aspect of our service. So you'll have recommendations on the versions.
You will have prompts to let us upgrade your software with some cadence, to a version that works for you or to use self-service upgrades capabilities fully available. So again, you maintain control. You maintain security. You just have somebody managing that envelope underneath of infrastructure.
So to dig in a little bit to that, to a question that Joel asked earlier in the Q&A By the way, Joel, shout out to you for one of the questions that, that Paul had put out there. Thank you very much for that. Joel asked a really specific question about controller versions and- Yeah it all and the timing of it all.
Yeah. Can, can you lean in a little bit on that? Coz- Yeah, absolutely. And- This, this is pretty important detail. No, yeah, and it goes back to what you were asking before, right? Do we use an S or an E platform? Do we use W- which model and which revision we use.
It's no secret. We generally will use, like I was saying, overprovision, more powerful products than is actually necessary. Now, the Generally, when you look at generations of, of products, the key, advancement we have from one generation to the next in our traditional purchase products is a bump in performance, right?
So maybe in, Again, I was a hardware guy, so I'll talk hardware for a second. The, the X50 that used to, to perform to this level, when you go to the next generation, it goes up by a little bit. The next generation, it goes up by a little bit. You go up enough generations, it the pr- the old generation X70 used to perform, right?
So what we, what we do is we take advantage of multiple generations So on the For each product, we have access to, both new generation, previous depending on lo- physical location, on inventory, because what we need to do is deliver a performance profile within a specific timeframe. We have a very stringent twen- 28-day guarantee to deploy infrastructure.
If you start a contract for Evergreen//One, or if you need a change, we will deploy the infrastructure within 28 days. You'll be ready to go. 28 days is a really short timeframe- That's an insane- to source- That's insane small whatever infrastructure you may need anywhere in the world and get it going, and we do that by really leveraging a, a diverse and global
pool of inventory. What we're gonna do is, hey, as you, if you need a level of performance that, hey, we would normally need the latest generation product on an X50 again to do, we ask ourselves, "Hey, could we do this on the previous generation one model up?"? Right? Can we do it that way?
May- could that actually deliver you as a customer additional buffer beyond what you would otherwise be getting, right? How can we It's an optimization problem. We're all engineers, right? How can we maximize our time to deployment, make that as short as possible, amount of buffer we make available to you,
while obviously, and I wanna be clear about this, making sure that anything we deploy has several years, before we feel like we need to come replace it, has 100% the same quality that we would deploy for any other purchase product and everything else we expect, you expect from Pure?Yeah, that, that drills into Joel's question quite well because y- you know, y- you think about it, right?
We're getting you to a result, we get to choose the hardware, and sometimes it may not be the bleeding edge, sometimes it might be the bleeding edge, but the onus of those decisions is on our shoulders, right? And I think the best way to look at it is we're leveraging more than just what's coming off the hot line, right, production-wise.
We- we've got a whole bunch of inventory that, that we've used for end users in that are just as good that, that we repurpose in our own house. Look, look, the, the way, i- in a way, a- and Abe chime in on how you would operate on this, right? But- Mm-hmm this is, again, trying to bring you some of the, the, the cloud
your environment, right? When I look at a cloud service, they're gonna have the only way they can expand their capacity, their performance quickly, is by leveraging a diverse supply of, of inventory. And we have exactly, the same challenge that they have. And in these days of, I mean, you've seen the DRAM prices going up, the DRAM- Yeah shortages,
everything else that's going on, we've made a commitment to our customers that even through, through some of the supply chain in the past, we did not increase our Evergreen//One, right? Costs went up, tariffs came into place, there were shortages of var- various components, and we maintained a 28-day, deployment timeframe and the same price on our Evergreen
//One service. And again, I think it's a testament to what you can do when you do actually leverage a very broad, inventory base depending on what the customer actually needs. I agree. I agree. Well, I was gonna jump in on the, 'cause there's another similar, or another question
in a similar sort of topic. So Jose was asking again about the, FlashArray and, and gave us a clue asking, right? Which is his organization has run a concept on a specific model, and they're satisfied with the, read and write speeds they're seeing. Well, that is actually a fantastic way to be
able to benchmark a performance requirement you might have, and operational requirement to, a tier of service. We've got a really comprehensive catalog of services that are like blueprints of service tiers already built with specific, performance levels. So this, and this is why I said at the very top of this, why I, I think that this
conversation for me, it more originates in what is the, public cloud type experience and how do you replicate that using equipment that someone's providing into your hosting location, rather than starting from how is this different to an Array that I might have purchased and am managing by myself? What, what is, what is the step up from that to this?
For me, it's more of a how is this a bit more like the, public cloud? So to, to Joseph Pinto's question, we equate whatever performance you're with to a tier of service, and then we deploy whatever infrastructure from our portfolio we would need to make sure we were delivering that, and then we would guarantee it. And then we would continue to make
non-disruptive changes to, to make sure it was maintained at our cost and, and our risk. I guess the last bit I was gonna say is as well, we should definitely more think of each instance of Evergreen//One like a little cloud as well. For, for example, it could be multiple arrays inside a single instance. There are customers we have with, you know, huge, Oracle Database, deployments, for
example, that span across, tens of, in a single instance being, being managed and being metered and, and supported by a single Evergreen//One instance, but are on multiple arrays. So I guess there's that dynamic to think about as well. We're, we're not just talking about how to make a single array feel cloud-like in your environment.
It's more an instance that could be made up of one or more. And, and then we can still consolidate, in, in the timeframe too. So within a term, we could consolidate down, we could split out, if we needed to. We could add arrays. So we talk about doing non-disruptive upgrades, that could take the form of deploying an another array, adjacent to the one that we
started with, or another two- Yep adjacent to the two that we started with. There, there are, there are so many things, I guess, that we can do- Yeah operationally, all as part of the service that, that don't add- Are there Are, are the, validated designs that we're a part of, are these in play for Evergreen as well? So AIRI, FlashStack, those kinds of things.
I, I assume Evergreen//One is in play for those as well, correct? A- absolutely. Let me, let me talk about that for a second. But by the way, before I do that, Abe, I think I, I need to say the magic word here because how do we do something like what you describe? Pure Fusion is a key part of that.
For those of you- Mm-hmm that have heard about, Pure Fusion and our ability, you unified control plane for our systems, that's what enables us to then treat multiple arrays as well as a single subscription where, you get a unified experience and, and we manage it all as a fleet, right? So again, another example of we leverage in Evergreen//One every bit of technology that
we deploy at Pure Storage. We leverage the latest and greatest hardware where required, we leverage the greatest and latest software features where, where it makes sense. We just frankly don't spend as much time telling our customers about those pieces of hardware and features because they are, to me, a means to an end, and you should hold us
accountable to delivering the end rather than the means. Right, the outcome. Yeah. No- The outcome no, to, to answer your question specific, specifically Don, on validated designs, absolutely. We support things like FlashStack, we do support that as part of
Evergreen//One as well. AIRI, we have actually an Evergreen//One for AI tier of service that delivers incredible value for, for AI workloads. It lets us focus uniquely on, the performance requirements of your AI applications without committing a single TiB of storage. So we have-Uh, we have solutions for every single one of those use cases, whether they're
exactly the same solution or whether we've actually been able to do something that goes beyond what you would have got from a specific hardware appliance- Mm-hmm that's where it may differ a little bit. So again, in the, in the case of AI, yes, you could have bought an appliance for AI that probably would have been a FlashBlade, or you can be on the Evergreen//One for AI service,
which, takes effectively the performance of, of multiple FlashBlades without charging you for the capacity of those FlashBlades. That's what we do cool things like that. Yeah. And I think that's an excellent segue into my favorite one, and we've got two questions on it.
So, the Cyber Recovery SLA add-on to Evergreen//One, I'm not gonna spill the beans or steal your thunder, but it's one of the coolest add-ons, and it's one of the things that I've seen used for people to be heroes in the field outside of cyber resilience, and I'll throw that story out after you guys are done. But let's do a really quick DeepReduce 'cause we did get some people asking about,
you know, what's included in the cybersecurity SLA, and y- you know, ultimately, what does this mean as far as ransomware is concerned, recovery and all of that? Cause it, it really is a cool story. Yeah, Abe, do you wanna talk about it? All right. I'll, I'll make a start.
I think it's a cool one. It is a cool one. I tell you what, Don, this one, this one, and I spotted a, a question about, snapshots, that we make time for in that, Q&A, 'cause that is cool- Yeah as well. Yeah. But no, you're right. You're right to raise it, Don. The, the cyber resilience, recovery add-on.
Now, look, everything that we've talked about so far, everything, is included in the core Evergreen//One offering, and there's only a couple of things that are specific add-ons. The reasons they're not included in the core we can go into, but the, the main add-on we're talking about, cyber resilience, is really designed to cater for the situation where, heaven forbid, an, consuming organization, gets a ransomware attack or potentially even
another kind of disaster that renders their primary storage infrastructure inoperable. In the case of ransomware, it might mean that, the equipment needs to be embargoed for a forensic investigation, maybe an internal one, maybe an insurance organization, or maybe even authorities need to get involved. What does the company do while that infrastructure they've bought for
is not accessible to them? And that could be weeks, it could be months. So they'd be staring down the barrel of looking to, potentially either get some swing kit, get borrowing something from their, their friendly vendor or service provider, or at worst, buying a load more stuff.
So imagine, th- the alternative to that, which is you talk to Pure as a, as an Evergreen//One subscriber where you have the ransomware, add-on, and we simply say, "Okay, that's the trigger of the process that we pre-agreed. We will next business day ship you a new set of arrays that match what you've got in production that's now been embargoed, and we'll send you engineering resources and help
you recover from your last known good Safe Mode Snapshot to restore business within days." you know, let them forget weeks or, or months or anything like that. This is, this is what we're prepared to do to help businesses, get back online as fast as humanly possible, we think, y- at our, at our cost and our risk. W- when that original equipment, surfaces again after its, you know, had its
investigation done and that's all concluded, we'll come recover that and return it to our inventory. So there's no sort of double m- it's not really a migration, but there's no kind of swing back to do onto the original, boxes if you like. We'll just pull those back into inventory, least, disruption possible, and, and have you
carry on running, BAU. Yeah. It's a, it's a- Cool, right? The You got the I mean, to me it, it's a fantastic, backup plan, right? Not, no, not really referring to backups, but really as a plan B, what are we gonna do in case, in case of a disaster, right? Yep.
And we, we review, by the way, that recovery plan, pre-agreed to a recovery plan that Abe was talking about. We review that with you every quarter, right? To make sure we are both holding ourselves accountable to do we know exactly what we would do. It's almost like practicing the motions.
If disaster strikes, what happens when we press the big red button, right? We, we agree on exactly what's gonna happen, then whenever you trigger the as fast as possible to do exactly what Abe said. Frankly- from a, from a, a peace of mind perspective, I think it adds a whole extra layer of knowing that if something were to happen, you're not gonna
incur You're already worried, frankly, getting your your data backup and running, your business backup and running. You don't wanna worry about extra costs that may- you know, you may have to deal with or delays, right? Can you actually get We're gonna get you gear within a day. Yeah.
How else are you gonna be able to do that? And- I'll tell you what- By the way, it's much cheaper than buying a second array, just like Abe was saying. Yeah, indeed. I tell you what, it, it extends beyond that even, right? So, so that- Yeah I think is, is pretty amazing that we can do that.
But it extends beyond that, and this isn't kind of formalized and, or anything just yet, but we are talking to some of the, larger insurance underwriters to say, "Listen, if we, if you, if you're convinced by this, if you, if we've demonstrated sufficiently to you that this really works the way, that we say," indeed it does, and indeed they are, "then would you consider a bit of, alleviating of your customer's insurance
premium," right? Because if they have this, then they're not as at risk as they would be. So as I say, not, not formalized or completed, but, but that's a real For, for other protective or preventative measu- measures that a consuming organization can put in place that has, you know, benefits and implications than just
just, just bringing the business back online.So this is, so this is hero hint number two, right? If you're on the front line and you're in charge of the daily management of the data for your organization, y- you this to your boss and say, "Listen, this is a legitimate extension of Evergreen//One. Clean arrays the next day, dedicated team to help me dig out.
We can get it back in days with this option, as opposed to nail-biting for 90 days hoping the tape comes back and everything." So I think this is another opportunity, right? For the front line folks to leverage what Evergreen//One with the cyber resilience SLA can do to, to look better. And the, the real world example that, that I picked up in one of the channels on Slack was,
somebody had water damage to their array, somebody said, "You know, if you had the cyber recovery SLA, you probably could leverage that until you cut the hair dryer out on your array or whatever." No, we- it's fair. We've been focusing a lot on, on the cyber attacks. Obviously, that's what, what is top of mind for a lot of us.
But- Yeah we do include coverage for any of disaster, anything that may require a new set of infrastructure really. Like, obviously- the, the old piece of infrastructure, you're gonna have to be going through insurance claims and any- and whatever else to, to recover the value of that infrastructure, but that shouldn't keep you from getting back up and running.
Now, the cyber- Yeah the recovery SLA is all about getting back up and running. Now, if I can quick-fire, squeeze in another topic that Abe brought up, which is very related to this one. We talk about how the best way to recover from a cyber attack or anything else is from your last good copy of snapshots.
Okay. So let's talk about snapshots, because number one, there's a question in the chat, but number two, I think snapshots are the hardest thing to size in storage. Maybe that's an ex- exaggeration. I don't know if you probably- No, I, I don't, I don't disagree.
I think we all have your rules of thumb, right? Every single person that I've, that I've spoken to, that when I do it myself, I use rules of thumb. I, you know, estimating change rate. If you think about what snapshot space is, and we're gonna talk technical for a second here. You have 100 terabytes of data.
I'm gonna take snapshots every day for seven days and then keep two weeks' worth of them, right? How much space should I plan for a snapshot? It's very, very hard to account for because snapshots are effectively the integral of your change rate, at discrete points in time with your,
with your rep- snapshotting policy. So it will depend on how quickly your data is changing, and that means really a pr- proportion of overwrites, if you think about it. It will depend on exactly how your snapshot timing lined up with that change rate. And it will obviously depend on how long you keep that, those snapshots for.
Most of us use rules of thumb. Most of us say, "Uh, assume a percent rate, assume 5% change rate, or just 20% for snapshots and be done with it," right? We all have rules of thumb that we've, we've got used to. Those rules of thumb are not necessarily precise, and they'll- there's always risk.
There's always risk that, "Hey, that one workload actually changes faster than I thought. What do I do now? Do I have enough space?" Now, with that premise, Evergreen//One for us is all about obviously taking care of the infrastructure, but we're trying to translate all of that into predictability.
Predictability from an operational perspective, predictability from a predictability from a technical perspective in the case of guarantees How can we make it so that you know exactly how something will behave when, when you consume it? This year we introduced what is, I think, an industry first. I've never seen anybody else do anything like
it, what we call a snapshot, add-on. So snapshot add-on is a flat package for your snapshot, retention policy. What that means is that if you have 100 TiB of data and you're gonna take snapshots and keep them for effectively 14 days, we now have a package.
You can just check that box with your account team. We will not ever measure, in your bills or in your, planning how much space those snapshots take. We're gonna take all of that away, and we're gonna take all of the variability and risk of how much space may be required, how much infrastructure may be required.
What if I didn't plan enough? What if it's not 20%? What if it's 30%? We're gonna take all of those, items of variability on us. We're gonna make it fully predictable.
Say, if you check the 14-day box, up to 14 days of snapshots are on us, and we have an option for 90 days as well, and even for a year worth of snapshots. Whatever your retention policy is, obviously this is your business decision. This is what makes sense to your organization. You can just tell us what that policy is and never have to think about snapshot space again
on Evergreen//One or really on any storage if, if you move it to Evergreen//One. So what we aim for with Evergreen//One is obviously reducing as much guesswork as possible and providing outcomes for the parts where we normally used to guess, right? Which again, that's hero stuff, right? Because you said it earlier, Tago, I can't predict what's gonna happen in three years.
Abe, you said, "We guess 5%," right? And all of it is guessing that never works out. And I think Evergreen//One flips the thing on its head to let a frontline just say, "Yeah, it's covered. Don't worry about it.
I've got the terms in there that we need. Can even alleviate cyber resilience insurance costs. Just, just- Yeah give me the storage, man. That's it."That was, that was the nature of the, the question, was we're having difficulty predicting the EUC impact from the snapshots.
Yep, yep. So, you know, what, what Targo's just said, fine, have a essentially an all you can eat within the context of a fair use fixed uplift on the unit rate. Done and dusted. There you go. Yep. So we're almost to the top of the hour, we've been fielding questions the whole time.
This has been great. Targo and Abe, again, thank you so much for your help and insight on all this. Let's take it home with 1 more potential hero thing. You know, we've talked about cyber, we've talked about no guessing, we've talked about just saying sure when your boss asks if you'd do it.
The, the 25% buffer that comes with it, i- is it possible to repurpose some of that if you find a need that you have to use Pure Storage Cloud for something? Is, is that within the bounds? Because people are gonna ask, "Okay, this is a cloud-like experience, how can I leverage Evergreen//One to extend into what would be a public cloud?" I'm gonna go further, Don.
I'm gonna Let's forget about the buffer. Let's just talk about, you know, the, the, the data that you're consuming, right? Let's say you have Evergreen//One, and you're, you have 500 TiB of data in your Evergreen//One environment, and suddenly, or in a predictable manner, you find yourself needing to shift some of that to public cloud, right?
Maybe there's a hybrid cloud strategy at your organization. Maybe you decide that a, a workload actually needs to be in a physical location that doesn't align with your, where your data centers are. Maybe there's a new office in, you know, the different part of the world that requires some of that data to be moved to that region.
So Evergreen//One, again, forgetting buffer, those 530 TiB as part of your subscription that we have equipment on-premises to deliver, you can consume any of that, of that, of that, data on Pure Storage Cloud as well. So you can seamlessly move, replicate over, of those 530 terabytes, any proportion or add more, whatever you wanna do, to, public cloud as part of your hybrid cloud strategy.
All of that gets counted together, aggregated together. So we don't really care if you have 530 and zero, you know, 200 and 200, zero and 530. You can keep moving them progressively back and forth, whatever works best for you. And your, your cost and, as well as your deliverables in your on-prem environment will
stay exactly the same. That's awesome- And to- flexibility for the Hyperscale- hybrid cloud question, right? To, just if I can, just 20 seconds on top of what Targo said, 'cause what- Yeah he was talking about, th- there's a, there's a fundamental point that I would love everybody on the call, if there were one thing gonna take away from this, would remember,
which is with Evergreen//One, whether we're talking about creating a contract, whether we're talking about metering it, whether we're talking about performance and capacity and all that sort of stuff, the number one thing that we're looking at is data. So the customer is paying for the data that they put on the platform. They're not paying for the amount of capacity that we provision, and they're not paying for
the 25 or more, percent buffer capacity. They're just paying for the data. So I'm, I'm in danger of just paraphrasing what Targo said, but when and that's, that's why this is so different, it's not about pre-provision some capacity and then measure the data. It's pre, pre-forecast, pre-commit, if you
want to, an amount of data that you're gonna store, and then we'll meter that. And so, so that's why he was saying just think about the data and then choose to move the data wherever it needs to go, into the cloud, onto another instance in the same onto another instance in a different place. It's always about the data.
The ultimate flexibility, which is really what Evergreen//One is about. That's my favorite part about it. It just make it happen. That's kind of what Evergreen//One is. So, well, we're at the top of the hour, guys.
This was a very ActiveCluster one. This was one of our more active Ask Us Everythings. I'm glad to see all the questions that came in. Abe and Targo, thank you so much for joining us. I really appreciated you guys taking the angle of how can the frontline people look like
heroes with Evergreen//One, because there's so much in it that enable them to go above and beyond just the usual lever pulling that comes with storage mgmt, you become a hero, that's when you, you can get the promotion too. So remember that, guys. Abe and Targo, anything else you want to say before we shut it down?
Again, thank you for coming. I appreciate it. Thank you, all. And, and, look forward to continuing to chat with everybody on this call. Absolutely. Yeah. Abe, got anything? Appreciate it. Same thing? been, been a pleasure.
Thank you so much. Yeah, and it was active, right? I think we got all the, to all the questions except for maybe Paul- We, we did. You've got Paul Filio's question there Yes. I think we're good to go on that. And, on that note, it's two minutes over.
Guys, thank you again for joining. I really appreciate it. Thank you to everybody who tuned in from the community. Be sure to log into the community website, at any point in time, and going, 'cause Abe and Targo are out there, as well as I and a lot of the
principal technologists. So again, thanks for joining us today. This is Don Portman, and we'll see you next month on the next Ask Us Everything. Thanks, guys. We'll see ya. Thanks. Thanks.
Abraham Barnes
Area Vice President, Subscription Portfolio, Everpure
Santiago Navonne
Principal Product Manager, Everpure
Don Poorman
Sr. Technical Evangelist, Everpure
Got questions about Evergreen//One? Get answers.
In this episode of Ask Us Everything, we’re diving into Evergreen//One™—our storage-as-a-service solution that gives you flexibility, protection, and cloud-ready capabilities. Whether you already use Evergreen//One or are exploring it for the first time, you’ll see how to get more value from your storage—without added cost or complexity.
We Also Recommend...
Personalize for Me
My Updates
No updates at this time.