1:04:22 Webinar
Ask Us Everything about FlashArray File Services
Got questions about managing files with FlashArray? Get answers.
This webinar first aired on November 21, 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
First, welcome everyone to our fourth Ask Us Anything for the Pure Community. My name is Rich Barlow, Principal Technologist at Pure, and I'm happy to welcome you folks here. I live in Richmond, Virgin- Richmond, Virginia, with my wife. I got two kids, 20 and 23, and my background is with NetApp, EMC, and Dell.
And just like I'm gonna ask my other two folks here today to e- share an interesting fact about themselves, I'll share one about me. I'm one of the 3% of the human population who's been north of the Arctic Circle, and next year I'll be part of the .1% of the human population that's been to Antarctica. That's interesting fact about myself.
We're excited today to be talking about files, files on Pure. And, but if you got any future topics you'd like us to cover, please reach out to us and let us know. This is y'all's event. We wanna talk about the things that are interesting to you. So please reach out, let us know what you wanna hear.
And we're excited today to let you guide our conversation. We have slides, we have stuff to talk about, but honestly this is not about this is about you all asking us questions, and us chatting about these things, right? Us chatting about, the questions that, that are interesting to you, and the hard questions you give to John. Just he's the guy you
wanna stomp in this thing. Just kidding. Anyway, you can raise your hand, you can come off of mute, however you wanna ask these questions. Before we start, let's get introduction for our specialists here, our, our smart people. First, I'd like to introduce Antonia.
Antonia, give us a 30-second intro and a, an interesting fact about yourself. Hi, everyone. My name is Antonia. I'm a Field Solutions Architect here at Pure Storage, focused on unstructured data use cases, file and blo- file and objects specifically, not block. I left block a, a while ago.
As far as my background, I have a background in EMC, Dell EMC, VMware, and NetApp. An interesting fact about myself So I'm not gonna let you decide what this accent is, but it's a vampire accent. So I'm originally from Romania, from the region of Transylvania, where all vampires come from.
We like to say don't mess with Antonia during a full moon. Yeah. Everybody knows, don't contact me at that time. Exactly. All right, Mr. Carnes. Yourself, sir? So I'm John McCarnes. I'm Senior Manager of File Product Management here for FlashArray and FlashBlade file.
I've been at Pure for going on nine years. Background is EMC and, working in file products probably since the early '90s, working starting at Novell, NetWare, so it dates me a little bit. And, got a I, I'm based out of San Jose, California now. Used to be in Salt Lake City, but I have moved.
And I have I'm here with my three daughters. You know, and I have a high schooler, a middle schooler, and an elementary child, so And they're all girls. Interesting fact about myself. During my high school years, I was actually an award-winning artist, which is something most people don't know about me. So I actually won quite a few, national and
regional award, art awards during my high school career, and I had actually thought about going art, but, wasn't much money compared to IT. So- Great here I am. What kind of art, John? Was it painting, or- It was everything. I, I w- I had won, I had won, awards in ceramics, watercolor.
A lot of national, a lot of my national awards were in watercolor. I have to admit, I would never have guessed that, John. Fantastic. Fantastic. If you come over to my house, I actually have one of my, award-winning paintings on my wall, which people think m- there's no way I coulda done it with, you know I, I did it, but it, it's there, so. Awesome.
Well, y'all, let's go ahead and get started. Antonia, I think you have some slides to set the mood kinda here, and set what's we're gonna talk about. Um- Yes let's go to here, and I see question in chat. Let's, let's start, and then we'll get into the questions, unless let's get into the topic.
But Steve, we see your question, sir. Give us one minute. Once we get started, we'll dive into this. All right. Everyone can see my screen? Yep. All right. Amazing. All right. So we're gonna talk today about
file services on FlashArray. And we have, you know, the person here that, I, I would say created or it's, bo- birthed the file services on FlashArray, John. But before we get started, I'm really curious, how many people on this call, are using, Pure for block workloads today?
And we have a poll coming up. And how many people knew until today that we're offering file services on FlashArray, and how many are actually running file workloads today on FlashArray? Is the poll running, Rich, 'cause I can't see. There it is. Oh, here it is. Welcome. Here it is.
Okay. All right, so we can't see it- So really, really curious. We need the, we need the music from, what is it, Jeopardy here? Do, do, do, do, do, do, do. We do need some music.
You know that's, you know that's just I'm a Little Teapot? Huh. I never thought of that. Wow. Okay. Yeah. Yeah. All right. And, and- How long do we still have on the
poll, Lauren?Up to you guys. I'm happy to keep it running if, if you'd like 15 seconds. Let's go one minute. Okay, one minute. Well, while we're doing that, why don't we talk about the question that's in the QA? Let's do it. I think there's a question from Steve Gray.
I'll go ahead and, talk about this. So Steve asks, you know, "How does FlashArray File integrate with cloud-based file services like, Azure NetApp Files or AWS FSx, and what are the main advantages of hybrid approach using both?" So currently, FlashArray Services is not based in the cloud yet. Now, it There's, Basically, the architecture we have, there's nothing
us from doing it, so it's more of a when than an if when we will add a file to, the cloud block store or the Pure, data cloud, Pure Storage, cloud, I think is been called now. So I, I, I think the there is advantages to having that where, you know, if you wanna have a DR or if you wanna do some cloud bursting. And it's not necessarily something that we are not going to do.
It's just simply we've been kind of waiting for the right time to do it. So, but if you're interested in that, and I would, I would love to hear, you know, what kind of use case you have for that. So feel free to chime in on chat or on Q&A or wherever. I'd love to hear what, what use cases you would like to run in a hybrid approach, as we
go ahead and, you know, start to prioritize that work. But that's definitely something that will be coming in the future. We just don't have a, a exact date yet. Jon, we have a second one here about, can FlashArray File Service use NVMe-oF or RDMA transport for file workloads, and if so- Sure how does it affect latency
compared to block services? So it does not use the It does not overlap with the NVMe over fabric with RD- RDMA over TCP. Right now that's more of just our simply works on our block workloads. We generally just use, utilize the IP iSCSI cards that the array supports.
And so latency compared to block on file is, is pretty, pretty comparable. And the reason why that happens on FlashArray File is because there's kind of no middle layer between the back end and file. So file services lands on the drives exactly the same way that block data does when it hits the NVRAM pipeline.
So the latency seen from block data is equal to what you see on file. The only difference is file IOPS ha- you know, have a few extra calls that don't have, such as metadata updates and, certain caching, you know, areas. So in that case, you know, sometimes you will see that file, can hit latency, the latency sooner, with a max number of IOPS per array than file block does.
And that's pretty common on most, unified arrays that, you know, file does not share the same max IOPS as block. And so that's probably the only thing that's different in latency on file and block is how many max IOPS you can hit. So Jonathan, you're saying there's no, no gateway, there's no file VM running
front of a, a Pure array? No. That was actually one of the, like, key about, key choices that was made when we built file services here at Pure Storage we did not want this experience of layering, where you were monitoring or controlling two different things. So, you know, I know in the past, like, you
know, I've worked for the competition, and I know our, you know, pe- other com- other companies out there, you know, built one or the other first. You know, they built block first, and then they had file added on, or they built file first, and they added block on later on. So at some point you're kind of managing both protocols to manage one, 'cause you have to
kind of watch the same, hierarchy underneath. On file services you're not doing that. So you're not managing or looking at any block volumes underneath the file system. It's simply a file system that w- lays on the global pool storage, just like you would see a volume on global pool storage.
So in that case, like it's, it's there is no, absolutely no layering here. That's fantastic. I, I love it, coming from NetApp. I'm a NetApp EMC Dell guy, it's just very different way of thinking about it, but it's a very future way of thinking about it, right? It's something that wasn't invented in the early 2000s or the '90s.
It was invented much more recently, so you have the advantage of all that new thinking, right? The cloud and all that new thinking to put into this. Antonia, I, I published the, the poll results. Do you wanna talk us through them? Yeah.
Yeah, so it's, it's really interesting. Of course, we're seeing what we already know, that a lot of people are do, do workloads on Pure. Thank you very much for being customers, customers today. We do see that very few are actually running, file workloads on Pure today, specifically on FlashArray, and that's not surprising.
And then we're seeing that this is surprising that more people on this call than usually in-person events I've seen know that Pure has been offering file services on FlashArray, so this is great. So awareness, higher awareness is great. Now let's talk about, you know, files.
So, I'm, I'm really happy to see that a lot of people do know, because most people know Pure for block, and they should. But a lot of people haven't been aware that we offer we've been offering file services since 2017.And, the investments that Pure made in the platform and, it, placed Pure this year, as a matter of fact, as the STaaS
leader in Gartner's Magic Quadrant for, enterprise storage platforms. This achievement in 2025, is the result of achieving a high score when it comes to Pure's vision, for data storage and, o- our ability to execute on the vision. And we'll talk in just a bit about the vision.
But, so the things that placed Pure as the storage leader, in, in the Gartner Magic Quadrant are our, unified data platform strategy for block, file, and object workloads, eliminating the need for storage silos. Improving customers' operational efficiency by automating as many things as possible operationally from a data storage perspective, and being given customers
manage multiple systems via an intelligent control plane that we call Pure Fusion. And the third thing, but not the least, is, the pro- protecting customers', customers' investments. So future-proofing customer investments in Pure, they And that, that is due to our architecture that we call Evergreen, consists of stateless controllers and modular
component architecture, which means that you customers don't have to refresh or replace, rip and replace your systems every three to five years. As new generations of hardware become available, you can just, non-disruptively replace your controller or the DFMs, the Direct Flash Modules. And additionally, we give customers the flexibility to consume the
platform, as they wish. As a service, SLA-based, no hardware ownership, or customer can purchase their hardware but by c- but consume it as a subscription, or they can own it, and upgrade it, continuously upgrading it without the need for rebuys. So these are the, the things that, place, place Pure in the le- the, as the leader in
the storage market. John, do you wanna add anything, or Rich, on our vision here? Yeah. So I think, Enterprise Data Cloud, you know, f- fundamentally the idea behind this i- is, has a lot of file, ideologies in it, as you see files in the center there.
And, and when it comes to file, right, data sprawl and having your data and not knowing where your data's at and being able to place data where you need it at can be kind of complicated. Because as you grow your, your environment from one array to many arrays, like you're Let's just say you now have 50 arrays to manage within your data center, and you've got
file, spread across all those. You know, logging in and trying to find out, like, oh, where was this share? Where was that share at? Where was that workload? By being able to connect to Fusion and being able to see all those workloads from any one of those arrays, you can go find the workload of any array in that 50 from
You don't have to log And, and then you want, like, copy that workload, you to deploy it a second time, you can do that from one array. Like, you don't have to go anywhere, and you don't have to have anything additional besides a FlashArray in your environment to do this. It doesn't require extra networking, it doesn't require extra VMs, it doesn't require
extra cloud connectivity, nothing. So simply go ahead and set up a fleet, and then go ahead and start deploying file across the entire fleet, and then, you know, seeing where all that, you know, seeing everything's at, which, you know, in the world of unstructured data can be a lifesaver for time.
So I think, you know, this If you're for a low TCO storage, being able to control your data, being able to see it all at once, being able to deploy, multiple copies of it, you know, where you're saying, "Hey, somebody else needs that same app deployed again," and you can do it in a few clicks instead of having to go through a whole bunch of reviews. I think this is where you're gonna see a lot of time and your operation costs go way down
at Pure Storage. And then also, you know, the fact that your data is no longer siloed, right? You have data You know, you can use ActiveCluster to move your block data around and all these things like that. So, non-disruptively in the environment, this allows you to, you know, really, really have your data where you need it to ha- where you
need it at the time you need it. What's interesting to me, John, is that from I- this is my fourth storage infrastructure company, right? And we've all had this message. A- all the infrastructure companies had this message of, "We're gonna make a cloud out of your data. We're gonna make it all mobile and
everything." What's really interesting is, A, Pure is doing it. We're not talking about it, we're actually doing it. And B, it's built in. There's nothing else to buy, there's nothing else to manage, there's nothing else to install.
In fact, it's already running. Even if you're not using it, the stuff's running in the background, right? So Fusion's automatically in there. So i- it's just a different mentality of this, right? We're not trying to make more money by selling yet another product.
It is built into the system. It just works. So I do wanna go and answer a few questions, you guys all right with that? Yeah. So I think we have a, an answer here from David, a question here from David, Rajal- Rajala.
Apologize if I said that wrong. And it says, "Does Pure have official documentation for details on setup of CIFS SMB3 to be fully redundant across both FlashArray controllers? Common use cases from customers, are Windows clients, Commvault, and Rubrik backup." So this is definitely a great question.
I love this questionAnd so we definitely do have a best practice for SMB, out there now published. And we offer, Basically what we did is we took a look at continuous available shares on SMB3, and we implemented an always-on version of it. So actually, if you're on the latest versions of Purity, if you're on, either six, 6.9 or
6.10, you would see a new option in the SMB export, policies that allows you to enable or disable continuous available shares. It's on by default, and the reason why it's on default is 'cause we actually made it so you could run it on any workload. I know in the past, like, you know, as a NAS guy myself coming from another world, we were
You know, we had limits in other places where it says, "Oh, it was only for Hyper-V," or, It was only for this use case or that use case." We wanted you to be able to use it for everything. So we only expect in very rare cases that you would ever wanna disable it. And the same goes for FlashBlade as well.
You'll see that per file system, you'll be able to enable or disable, continuous available shares there as well on SMB3. So this way you can always have a, a transparent failover between controllers for your, you know, SMB clients. So that's really That's one thing that's there available today.
We hope people take advantage of it. Fantastic. Does anybody else wanna take the second one? What telemetry is available for real time IOPS and throughput? So on this one you can You actually get some i- you get IOPS and throughput information right out of the, FlashArray GUI.
If you go to analysis, performance and capacity, you can do that. And we do definitely do have, performance API met-, performance metrics, that can be exported via APIs to, third-party tools. I'm not, I'm not yet very familiar with the ServiceNow Splunk implementation, but would say that we, we can do that, if you wanna, if you wanna do that.
So Jon, Rich, do you have anything to add here on the telemetry? The only thing I would add is that this is on, on FlashBlade this is a On FlashArray this is at a managed directory level. So every managed- Mm-hmm directory has the ability to go see the IOPS, the live IOPS coming into it.
And you can do that through the GUI by exporting or there's APIs available to grab that data through REST. And definitely you can, It's not s- it's not set up directly for, open telemetry yet, but we are working that way to go and have it ready for open telemetry, and you know You, you know, you can import it, into Splunk or into other utilities.
If you You, you just wanna make sure you look at the formatting. So And when In addition to that, you the ability to do auditing, which you at, like, all the file access, auditing events. That is actually on both FlashArray and FlashBlade published via syslog in the native, in the native JSON format that actually ingests seamlessly into Splunk very well.
So we've had a lot of success with customers doing that as well, so they want the next level of, information above, you know, just what IOPS are there. Fantastic. Darryl, since, since we, we see your in chat, but since we've given you talky permissions, why don't you go ahead and just ask your question live? Yeah, sure.
Do you guys got me? Yes, sir. All right, cool. Yeah, the, So, I have two FlashArrays. We're replacing your competitor, that begins with an N out of our environment. Oh, nice. So we did our block, and now we're focusing on the file.
Now, as you know, your competitor started their life doing NAS data, right? So, we have about, yeah, 50 SVMs, you guys all know the acronym, that serve, CIFS and NFS data. Everything that I'm seeing, FlashArray in, its new Or can now support the, multicloud in, like, 6.8, I believe, and upwards, but there are limits on them.
And I really don't need a FlashBlade. I was told the Yeah, I was told the FlashBlade can go up to 200 multi-servers. FlashArray can At this point I've seen documentation for 16 or 20. When is that going to be raised, So we're, we, we, we're gonna be looking at raising it in the next, code release.
So definitely take your limit that you need in place. I think we were gonna go a little higher than your limit. Awesome. We are definitely looking at the scale increases in the next, Purity release. So I'll, I'll bring it back to engineering.
Thanks for letting us know. I mean, we always I knew we would need more than 20 and 16, but as we're building out, you know, these things with replication and everything else, we were making sure we were able to deliver on time, right? And get everything in everyone's hands. Right. Okay. So any idea-ish I'm not gonna hold you
to anything and be like, "Giancarlo said..." I'm hoping to go to 100. That's what I'm hoping to go to. So- That would be- That's my, that's my goal. So, FlashBlade's actually going to probably much higher limit, but, you know, obviously it's different. And we won't That's not the, You know, when it comes to scale and performance, I don't really have, like, the, the very last end thing in
place, because we're gonna be building 300 terabyte DFM, right, on FlashArray. Yeah. So that means we're gonna have to have a lot of, you know, NAS objects for you guys to be able to manage. So we'll keep on increasing that as, as the, scale requires. Okay. That's great. Thanks. And, you brought me back
with the network comment. That was great.So one thing that, one thing that's different about Pure that I've noticed from being here is that we don't beta test in the field. We, we introduce products- with low limits, and we slowly grow them because we wanna make sure all the pieces work together.
There's a lot going on inside the array, so wanna make sure it all works together. And to do that, it tends to We tend to see our limits again in being very small, people are like, "Well, why didn't you make it bigger than that?" We're gonna, we're definitely gonna make it bigger. It's gonna grow. We just wanna make sure that all the pieces work together well before we stress it out a
little bit. Yeah. Makes sense. There you go. I think there's, the one thing that engineering knows, if you talk to engineering at Pure, is that everything we build has to be non-disruptive. Yes, and that's good.
That's the key tenet to, to, designing a storage array. Yep. I mean, they will I mean, and it's, it's It gets, you know If you ever see Pure delay a release of something, it's probably because it, you know, didn't pass that bar, and that means we're not gonna give it to a customer. And so I, I, I really love that about Pure.
I mean, I've, I've worked other places where that wasn't always the, case with Mm-hmm. I know what you're throwing down. Yep, I've experienced that. You need to, Y- you, you guys should all watch, Andrew Miller's, presentation for the 15 architectural choices, and one of them is about NDU all the time everywhere, and it's
something we in Something at Pure we started, John, was it like 2008? This is It goes way back. Yeah. This goes back to the beginning of Pure, it's 15 choices that we made, and only one of them have we, have we changed on, and that's just because we've run into a which we need a different architecture.
But 14 of those things are exactly, we've stuck to them. And it's interesting. We made some good choices in the beginning, and we're, we're rolling with it. Next question from, Keith Smith. Keith, you wanna come on and, and, ask your question live?
Thanks for your question, Daryl, by the way. Yeah, appreciate it. Yep. You're welcome. Thank you. Keith, you there, sir? All righty. Keith, if you, if you, if you, if you come back and you hear your name, feel free to break in.
Antonia, otherwise why don't you continue with the, the presentation, sir, ma'am. I think, Keith, are you on mute, unmute now? Oh, there he is. It could be. I, I hit that by accident. Sorry. It's okay. Oh, okay.
Hi, Keith. What's your question, sir? Hi. Yeah. I didn't have any question. I just, hit the, the raise hand button by accident. Oh, oh. I thought you were saying you hit the mute button by accident. Apologies.
Well, hello there. Yeah, hi. This is great. All right. Well, thank you for joining us. Thanks for joining. Um- Antonia, why don't you continue? Yeah. Yeah, so let's continue.
What I wanna go over next is what, how this vision came about, right? And we have somebody, here, like I said, John, who's been from the beginning on the trenches of building file services on FlashArray. And so, based on conversations, internal conversations, this is how it, it became transparent that this is how Pure, when it came into
the storage market and specifically into the next storage, saw the customer environment. Right? Like, the Rube Goldberg of storage very complex, a lot of moving pieces, and just one failure in everything, the whole chain breaks, right? Am I right, John? That's how you guys- Yeah
saw it at the beginning? I think the early game is, you know, if I, if I could sum it up, the, the one thing we, you know, we got feedback from customers, right? When we, when we were building file services, and I think the biggest there was too many dependencies, on other products.
Like, you're always, you're always managing several layers. You're always managing several different things, and sometimes if you weren't paying attention to one piece of it, you ended up breaking another piece. So, like, and this piece, and this, this is really kind of a prime example of that. If one piece here doesn't move right, this whole giant machine that you see in front of
you doesn't work. And so you have to be checking every You're monitoring and checking every little piece. And this is, this is supposed to be one file system here, you know? Or one set of applications. And so Pure had always been in the world of saying, how do we make fi- you know,
storage simple again? That's really was, you know, the foundation of Pure. You know, that included NDAs and things like that. Obviously Block was super simple. Go create a volume and give it a size and walk away.
Like, don't create pools, don't create anything else. Don't worry about managing the underlying storage. Like, that was a new concept when Pure came out with that. Like, obviously it's been copied and used in other places. And so definitely that's been, you know, one of the issues there.
But File had suffered from a similar thing. You know, there was constraints on file systems, where they were growing at, what, what architecture they were built on. How did you get a file system beyond, back in the day, 100 TiB, now, you know, sometimes 300 terabytes or something like that?
How did you rebalance? All those things were, you know, monitoring lower levels, trying to figure out where you put things next. I called it the game of Tetris as well in the past, you know? You always had to play Tetris. Mm-hmm. And so basically what we came up with is we
just did the same thing that Pure did on Block. We have a file system that has unlimited size, and you can grow it into as much storage as you can put into the FlashArray. And, and that's really the goal there is, you're not, you're no longer worrying about all these underlying pieces or anything.
Just go grow your, grow your data where you need to go grow your data.Just like you would on Block. Actually- What, what I like to say sometimes, Giancarlo, about this is it's 2025 and you're still talking about RAID groups. Really? Yeah. That And that's, you know, file is a little bit more complex in my opinion.
Having done Block for most of my career, I always felt like Block was very straightforward to me, Fibre Channel, right? Even from a troubleshooting perspective, like where the issues were. As soon as I started on the NAS part, on the file, I don't have the slide here with the screaming man with the hair on fire.
That's how I felt when I, when I first started learning NAS and file storage. So, and coming to Pure, and, you know, I joined at the end of January, and start It's Like, the interface, it's so intuitive, and then there's nothing to manage. I tell you, I kid you not, after I took my first course, it was FlashBlade, right? And I was told, "You don't need to manage any RAID groups." I was disoriented for a couple
of days. I'm like, "What do you mean?" Until I went into the GUI and saw what they meant. You don't have to manage anything. So that's, that's I think it's beautiful. So, do we want Do we have any questions?
Do we need to stop for questions, or do we go forward? Let's keep rolling. Okay. So, what is causing the customer's environments to look like this? So when, when you look at the unstructured data growth and the number of use cases that leverage file specifically and object today, we see a massive
growth that's being generated in customer environments. So if you look at this slide, IBC estimates that by next year unstructured data, in purple, will reach 200 exabytes. And if you look at the block data, that's uns- which is structured data, that's very low. It's trailing, behind by a lot.
So customers have to deal with this massive data growth, and that's, you of the, of the, of the Rube Goldberg picture of the NAS storage. And then in addition to that- Can I chime in on that one as well? Yeah. Yeah, let's go back. So, you know, I think, you know, for me, a key here, and, and feel free to chime in anybody on the, on the, on the webinar, and, you know,
you either agree or disagree with me. You know, when you look at what is the biggest cause of an outage on storage? It's when somebody makes a mistake, right? The more storage you have to manage, the more data you have to manage, the more objects and things you have to ma- or that you're forced to manage, means there's
opportunity for more mistakes. And that means there's opportunity for more outages. The goal here, you know, the Obviously, the goal here, the problem we've been looking to solve for customers on file is that you have less stuff to manage, you have less things to worry about. That way there's less chance of a mistake.
So for example, you know, we have policy-driven operations, which means you have these reusable policies, you have less You know, you don't have to have If you have a similar set of, resources that are being accessed by the same set of groups, you have less stuff to manage. You only have to manage the one policy. And likewise, you know, you're not managing, again, RAID groups and everything else.
You're just managing your file system or your directory, and that's it. And so the goal here is, like, as you see the unstructured data grow, in your environment, you're just simply, you know, managing less things, which hopefully will mean you're going to have less downtime from accidents as well as, you know, having the Pure experience of non-disruptive everything.
Plus fewer- Yeah points of management. That's one of the interesting things I think about Pure on the file side, especially through Pure Fusion. You don't have Some of our, some of our Some of the other folks, you manage every controller as its own point of management, right?
And it just leads to a- Sh- a, a spread of management. You got a lot to manage, which is much easier to make those kind of errors again, Giancarlo. So it's just simpler with us. Yeah. Fusion just takes it You know, takes file, the simplicity of file that was already there, to the next level, 'cause now you're Instead of
having few things to manage on a single RAID, now you have few, fewer things to manage across a fleet of arrays, and it's easy to find where those things are at and move them around if you need to. Yeah. And what I feel, too, is that, you know, some of the, file protocols complexity is you don't customers don't have to deal with that.
The implementation is such that it's simplified a lot of the NFS/SMB complexities. And I But that's, that's great. Well, people ask us- Now, au- au- automating complexity is hard, I'll tell you that right now. But when you look at file, right, it's been These protocols have been around forever.
We should know how they operate. We should be able to adapt to your environment and not make you have to go not dip switching. Like, we're not back in the '80s- saying, "Hey, I'm gonna go put a CD-ROM into my box, but I've gotta figure out what IRQ to put it on, you know? So let me go sw- let me go flip 100 switches
to get it, make it work." You know, we wanted to automate storage. So when you're connecting to AD, you'll actually notice that there's not a lot of options. You know why? Because we go discover your AD, we discover the config on the AD, and we adapt to it automatically.
Why make you have to go figure that out? You know, the array should be smart enough to know what's out there. I mean, AD hasn't, I mean, fundamentally, hasn't changed a lot for the last, you know, probably 10 years, 10 or 15 years. Right. You know, since 2012, pretty much.
So-You know, that's where we wanna be at, and that's what we, we build. And, yes, we have given some control, you know, where needed, but minimalistic. Again, we don't want you to have to manage all our stuff. Yeah, and that's- The first time I saw this, I, I asked myself, "Where are And I The first time I saw I mean, be- I've been with Pure for two and a half years, so
the first time I set up a PureArray for Block, I was like, "There's gotta be more than this." And when I got the file, I was like, "Where's the pages and pages of menu options? This is it?" Because our arrays are smart. We know what, we know what needs to happen, and we do it in the background. Yeah, and this leads to, you know, to why customers' environments that run legacy mass
storage look like the Rube Goldberg. So one, they have to deal with the massive data growth, right? And then with the added complexity of legacy mass storage. And their complexity, to be fair here, you know, they started back in the '80s, right? And when they started out, they did, they, they did good things.
They are still doing good things. But initially they started out by replacing snickernet, and being a- people the ability to share files over the network and, send documents to a printer, email receipts. So the whole idea of files and folders is the, the ver- is the digital version of the office
environment, right? So initially, these mass storage systems were created to improve productivity in an office environment, and they were considered secondary storage systems when compared to databases, transactional systems, and mainframe apps. So, then fast-forward to today, you have WEKA more use cases than improving the productivity
in an office environment. You have AI/ML, you have HPC, you have EDA, you have analytics, you have IoT, you have streaming media. So you have all these use cases and massive data, and these use cases are not secondary storage. They actually are, are business critical application that determine a company's
competitive po- position in the market, right? So what happened, the legacy mass storage had to adapt, right? So when a- they adapted, they introduced complexities that customers unfortunately have to manage. So what ends up being is customers have multiple storage silos, each one with their management tools.
Then customers have to You have to manage, clusters of storage, hard disk, SSDs, RAID groups, aggregates, volumes. Then you need to do upgrades to take advantage of any new features, and to Let's say you don't have to take a downtime, but it impacts performance. So you have to do it outside business hours.
You have to do a lot of planning around it. And then, legacy mass storage performance is tied to hardware, so that's why you have to replace your systems every three to five years. And for that's, you Many times you have to do migrations that are disruptive, and particularly file migra- migrations.
I ran block migrations. Those were easy, to me. And when I came into the file migrations, world, and I watched my customers do file migrations, I was like, "What's this?" It was horrible. It was like a nightmare.
So they are, they are hard to do. So that's, that's where our vision came, from, looking at what customers have to do. Wait, can I jump in there, Camilla? Yes, go ahead. Go ahead. Do you want me to go back? Yeah.
So- Okay I would like to see I, I would It'd be interesting to see a raise of hands from people participating in this. How many people have ever sit on a weekend during a storage, upgrade, a hardware upgrade, and having to watch their data you know, as they had to, you know, move it around, to get, you know, hardware upgrade done, or they had to migrate from one array to another array to get the upgrade done?
Yeah. Because I remember, you know, I used to be a, I, you know, I used to be a system admin back in the, you know, early 2000s, and I remember sitting around, you know, spending my Memorial Day weekend watching an array, moving, file systems around in the cluster to try to get things to upgrade. And I, you know, I had to wait for it to rebalance, wait for it to make sure it didn't
You know, non-disruptively It was non-disruptive upgrade, but it was literally, you know, two to three days of work. Then I come over to Pure, and the first time I'd seen a Block upgrade at Pure, It's two hours and nobody even watched it? Is that even possible?" And then we built file services, and we did the
exact same thing as Block. We said, "Well, why, why have to move the data around? Like, just let the data stay there and non-disruptively upgrade the array in place. So not only is it non-upgr- non-disruptive upgrades, it's a data in place non-disruptive upgrade, and that means just go and re- go, go enjoy Memorial Day.
Let the, let the array upgrade. You sh- you, you're gonna be fine. Like, that's the whole goal here. Like, it I would've loved to have a FlashArray or a FlashBlade with file on it back when I was a system admin, then I could take off the weekends when I had to upgrade
like that. It would've been nice. So this is, this is an award I got from NetApp back in the day, back when I was a customer, because it's Individual Innovation Award. And I do remember many weekends sitting there babysitting the upgrades, right? I would always come with my engineers, and I didn't wanna leave them on the weekend by
themselves because I thought it was kind of a mean thing to do as a manager. I should show up and be there with them, right? And I remember all those weekends, and I cannot imagine today I, we have, A couple years ago at our sales kickoff, we had a, the engineering, the, the director of infrastructure for a very large airline who's based in Florida-Um, that is, He came on stage
and said, "We update our arrays with the airplanes in the air. We have no more maintenance windows. We don't talk about that with storage anymore." That's impressive. I mean, that's like, you know, the airplanes are flying, the system's running, and they're updating the storage, and nobody, none of the applications have a clue.
That works for file and that works for block, so don- that's not just a block thing, that's also a file thing, because we're not moving and migrating your data, so it's not migrating anything. There's no file migration. And you can go for, I mean, how long?
I have customers started in 2012 and they're still running the same arrays, quote, unquote. Fascinating technology, man. It's why- I'm not gonna say who does this, but I know there's, you know, defense systems running on FlashArray file, and, you know, they have done several hardware already and not You know, they didn't even plan a maintenance around it.
Yeah. We're talking, like, this is, you know, you know, somebody, some country out there's, you know, civil defense. So, you know, that's I think that's pretty top priority if you think about top priorities in the world. Absolutely. Right. So- Let's go forward this Yeah.
So the You know, this is really good conversation and pointing out the, the differentiation, right? Which is our hardware and software architecture. So the architecture is what is gonna customers, and all our customer, block, object customers, the ability to evolve their platforms without the need to
rebuy new systems to for, to lift and shift their environments, right? To take advantage of any new features, or to scale capacity and performance. So that's number one. And then again, it comes back to, removing some of all those moving parts by consolidating workloads on a single OS for fi- for block and file.
And we have object coming soon on FlashArray as well. And then a single OS for file and object. And, I think that simplifies the environment a lot. You just pointed out as we were conver- you were conversing about this, John. And, the additional differentiation, I think it's our intelligent control plane, which Rich
talked about, and John, so it's Fusion. And what does This means is that you can manage multiple arrays, with, from a single control plane, and this control plane is built into the storage OS. So you don't need to install any, any VMs, any connectors, any collectors, nothing like that.
You simply buy your arrays and you just enable, and you have Pure Fusion. So nothing additional to install. I think this is, this is big. Yeah. So this is, this is the vision. We're, we're very close to achieving the vision, and we have the ability,
as Gartner assessed, and the industry assessed to, to finalize our vision. Awesome. No questions right now, so let's drive forward. This is it. That's all I have. That's the, the- I have other things questions slide, that's the short ones. I love a short slide Oh, th- there, there are more, there are more slides hidden behind,
where we talk about the differentiation, but I think the main things were pointed The hardware architecture, to me is big, coming, from other vendors. The DirectFlash modules as a technology are our differentiators, right? As opposed to SSDs. I don't know if we I know Rich is a fan.
He actually has a tattoo of a DirectFlash module. Please show it to the crowd. So this is a nine TiB Iger, but it's, it's I, I couldn't get a th- I couldn't get 150 for the, for the tattoo guy, so I'll talk to Kaz about that. So he- here's a question for you guys.
I mean, continuing the conversation about this. Mm-hmm. Some other things that differentiate us from our competitors when it comes to file. We have, and we talked about, we talked about, NDU, we talked about Evergreen, we talked about simplicity. What are some other things that differentiate us? So I think that simplicity, it's key, right?
And, that's based on the architecture comes again there. And, simplicity comes from the fact that the architecture allows for a single pool of storage where customers can provi- their block volumes, their file systems, and their objects from the same pool of storage. And by having a single pool of storage, also our data reduction algorithms get to be
applied at the pool level as opposed to, let's say, aggregate level. So not only customers don't have to They You don't have to manage any RAID groups, aggregates, volumes, whatever other construct to be able to provision your So that's, that's key. The single pool of storage also means that you can basically scale your file system size to
the entire capacity of the array if you want to, without any needing to manage any additional constructs. I don't know, did I miss anything, John, here? If I The one thing I would say that Pure is doing, I wouldn't say it's, you know, differentiator, but the way I would look at it is, like, you know, there's an old quote my
dad used to tell me, like, "A smart man learns from his own mistakes. A wise man learns from others."And having, coming, you know, in a modern day Being able to build a file storage solution in these, you know, in the last, you know, five years, 10 years, depending on which, you know, FlashBlade FlashArray. Pure has been able to learn from the mistake of others.
And, and so as we, as we built every feature, as we built every portion of the file, we looked to build it better, smarter, and, and, and less, less complicated. You know, for example, look at multiserver, for example, right? Multiserver in a lot of places, servers own data.
They're hard connectors. Like, you know, your file system will never be able to be moved to another Very easily. Well, imagine you're in an environment where you've got to eventually evolve your AD environment. How are you gonna re-permission that file system?
You can't connect it to two sources at once. You can do a copy, and then you're gonna do a complicated mess there. Well, now imagine you come to file services here, and now if a server is only an access control mechanism. So, you can connect two servers to the same data, and you could re-permission You could
add new ACEs and ACLs and permissions to the data that comes from a new AD, and eventually seamlessly where, you know, you, you didn't kick anybody off the data. They were connected through the old server if they were there doing it. But now they can go over to the new AD and connect that way and get the same access to the data and, you know, that, that kind of thing there.
For example, the other thing I saw in, in Europe, we had a customer who was trying to Forever migrate to move from NFS V3 to V4.1, but they couldn't They didn't wanna spend all the time creating the ACLs. And I said, "Well, actually if you migrate to FlashArray File, we automatically You know, if you leave the cross-protocol access on, and you have it set up correctly following RC2307,
we'll create all the ACLs on 4.1 for you because we are able to do that inside the box." And I said, "It's part of our always-on cross-protocol solution, that we built cross-protocol from the beginning, not at the end." It wasn't like we started off with NFS only and said, "Oh, well, no." We actually The, the system was built for cross-protocol. It's like always on cross-protocol.
You actually have to turn it off if you don't want it. And so things like this where we- we're kind of reinventing file a little bit as we go along, learning from the mistakes of others and learning from what customers are telling us. So hopefully, you know, some You know, most of you in the channel have heard from me before or, you know, talking before.
But one of the, one of the founding that the PM on the file side do is we love, love, love to talk to customers. We talk to Antonio, who gets to talk to a lot of customers as well, and we incorporate all that feedback into the product. It's not like, you, you will Everpure hear us go on deaf ears.
We will always take your feedback back to engineering. We will always take your feedback back to a backlog, a PM request, and you know, yes, it may not be done tomorrow, but we will work it into WEKA, we want our customer We want this to be a customer product, not a, you know, Pure here's, you know, shove down your throat how to do it every time.
There are choices where we made to make things easier and, and hopefully those, you know, those all come through like policy-driven management and things like that. But in the end, this is a very customer-driven product. Jon, one of the things I really enjoy about, about FlashArray File, and this is, this is not just file, this is also on the block side.
But the way we've implemented multi-tenancy, where we've separated isolation from administration and made them totally flexible. It's You don't create like little virtual machines inside the system, and then you put something in a virtual machine, and then it's locked there forever. It's nothing like that.
It's one pool of storage in the back, and it Can you talk a little bit about this? Cause we got a few minutes. I, I think it's an important thing when it comes to file is multi-tenancy when it comes to file. Well, when it comes to complexity, i- in other, other solutions, you're to, to handle all those complexities at once because you don't get to choose what you need.
Sometimes you only need access control. You just need untrusted domains. Sometimes you only need to be able to separate the share namespace, you know, team and your C-level and your regular employees, or that application team you, you need You just want to be able to keep under control.
And sometimes you do. You know, you have a DBA that really needs, you know He wants to create his own snapshots. And so you can go ahead and give him delegated access to go create the just his data. But maybe he has a whole bunch of things to manage.
Well, you don't have to go configure that guy on, you know, five different, you know, virtual machines. You just go figure him on one realm, put everything you want him to mess with and put it in there, and he's, you know You automatically know what he's able to see. And so- And then later you can move things out of that realm and back into that realm.
It's cool. So the thing is, is like you have this It, it, it's all You know, I call it like the combo menu at McDonald's, right? It's like, you know you need certain things to run out of storage. You know, you need a hamburger, you need a drink, you need a side probably. But you don't have to have all of them.
You don't And you get to choose what you want, and I think that means you're not forced to always buy the drink. You could just buy the sandwich and the side, you know? And that's really what, you know, it's Pure is all about, is, is giving admins flexibility with the least amount of complication, right?
Cause flexibility can lead to overcomplicated storage solutions. That, that can happen. Some people say, "We're the ultimate flexibility." But they also come with ultimate complexity. Pure is about keeping things flexible, you know, where you need it, but also making sure that it's not overly complex to do.
So, you know, if you have a new storage admin trying to spin up on Pure- Pure, I guarantee you he's gonna have a really nice day 'cause it's not hard to learn, you know?Yeah, that's one of the interesting things again Giancarlo, it's one of the things I see that's very interesting, and when you go in the interface, there's not a lot of buttons. There's just not a lot of buttons there, because the system is smart enough to figure
things out itself. And, you know, we're very, we're very At Pure, we're very, very, very picky about what we put in the interface, what we expose, because we do not wanna make it any more complicated than it needs to be. That's not because storage administrators aren't smart people. It's because you shouldn't be dealing with this stuff.
There's bigger problems and bigger things, fish to fry, than creating file systems and block array, and block devices. That stuff should be really sim- simple. You should be spending your time doing architecture and the stuff that really matters, right?
The big pro- You know, dealing with your DBAs last 27 problems they've had. Which is all in, which is all related to one index, by the way. I'll tell you the answer for that. All right, folks. We got, we got three and a half minutes left. Any more questions before we start wrapping down here?
This is Ask Anything, so I'm good. This is Ask Anything. Not anything, most things. All right. Daryl, can somebody unmute Daryl, please? All right, Daryl. Have at it, brother.
Hey, guys. I wanted to ask a question, to Jonathan about the, multiservers. I think you said in the next version of Purity. Would that be the long-running version or the feature release? The feature release, so 4.8.
Or, sorry, 6, 6.10. 6.10? Okay. Yeah. Um- I'll let you, I'll let Daryl, I'll get back to you on it. I'll, I'll, I'll, I'll reach out to you. I'll let you know. That'd be great, yeah. I'll- I have to I need
to double-check with engineering. I mean, this is all roadmap stuff, right? So you gotta- Yeah double-check where they were at, where they were planning on pushing it. Cool. Cool, yeah. I have budget, and I, I need to spend it, and
I really wanna give it to you guys, so No problem. We like the sound of that. So, like, just be aware, like So one of the things about these scale we do, like architecturally there's no not really a reason to limit servers probably. The only thing I would limit it is, like, how much we replicate them, but,
uh- Mm-hmm in the future. So we didn't really actually have any, like, hard limits there. So it's, you know, mainly just testing we'd have to go do. So let me, let me talk to engineering about it and get back to you, Daryl. That sounds great. Thank you, Jon.
And just for everyone's, e- edification, we released an enterprise release today. We marked our 6.9 release as enterprise ready. So if you're waiting for an ER, quote-unquote, "ER release" I hate that, by the way, because it denotes that everything else is not enterprise ready, and that's not true at Pure. Our, our releases are very well-tested.
We don't beta test in the field. Not how we do it. So if you've been waiting for an ER release, 6.9, which contains Pure Fusion and FlashArray File, and all these things are available today. And then I'd also just, I just got notification today that the Pure Storage test drive for File has been upgraded to 6.10.0.
Okay. So it has absolutely all the latest features, that if you wanna go and play with it, multiserver, user quotas, any of those things that we just, you know, came out with. Snbca, you can go look at what that looks like. All that's there. So i- if you want, that's, you know, easy to,
you know, reach out to your account team or, or a person in the community that you wanna, you know, you wanna get, access to test drive, and you don't even you know, go all you want in a little sandbox. Giancarlo, we got 60 seconds left. Antonia, we got 60 seconds left.
We have one question in chat. We have a- Mm-hmm use case. We have approximately 800 users of ESRI updating files which need to be shared across Canada. How would Pure do this so files can be accessed with the latest data in real time? I think it's a concurrency question across multiple arrays maybe.
I think that there is maybe, if we can allow the customer to come off mute, off mute, Paul Johnson, because I have a different, hypothesis here. I'm thinking that there's a central location for the files, and then across Canada that need to, locally update the files.
Paul? Hi. Hi. There are multiple locations across the country that actually host storage spaces for these files, and part of the problem we have is they're never in sync. So I'm thinking maybe- And this goes from Vancouver, Edmonton, Yellowknife, Whitehorse, Iqaluit, Toronto, Gander, Halifax, Montreal.
You name it, we're everywhere. And y- today are you currently running them on FlashArray or...? We are not, and that's what I am looking for a better solution, because problem that we've had going on for years. So one of the things that we've been working on, Paul, on FlashBlade, is actually
called Rapid Replicas, and what it does is it allows you to keep data, between different arrays, but be able to see the metadata and eventually do, you know, write back and things like that on it, and through it. So I wouldn't call it exactly a caching solution, because it's, it's a little better I think it's a little better than a caching solution.
But that might be something you wanna talk to your account teams about, Rapid Replicas, and what it might be able to do for you, and your exact, you know, looking at what you have and what your the size of your data and all that stuff. Because that's definitely a utility where, you know, everybody gets to see and access the data.
And they can You know, what happens is read the data, the data is quickly pulled, from the source data, right? So it's, it's pull on access. Um-Which means, you know, you're not, you're not always replicating all the data or trying to synchronize all the data at once.
It's only the data people really need to access. But they can see all the data on all the sites. It's a very intelligent way to, to deal with immediate consistency problems, right? Like, especially with object. I mean, it This works with NFS and SMB though, so don't take this too far, but it's the same
idea of consistency, and this, see, it is consistent. It's all in one place. It just looks like it's everywhere. Yep. But you know the- One place, I mean, you could have, you know, this be cross, you know, across different ways. So you could have data from one site as a Rapid Replicate on a different one,
then the other site have rep- data there that would be rapid replicate to the other site. So, you know, it would be in different file systems, but, you know, each one unique file system, but the fact that everybody would be able to see the data all at once, and then pull data very quickly to what they need, could be a use case. I think that might be something that could solve your problem, Paul.
Antonia, I'll give you the last word, man. Yeah. No, my, my next word was a question to Paul, if this He thinks this would, would help in, in his, with his use case. It potentially could help. My only other challenge at that point would probably be bandwidth- and access-
accessibility to the network, because that's also a challenge when we start getting into the North. And then if I've got field users that could be in some place like Cold Lake, Alberta, or Baker Lake in the Northwest Territories, as an example, because we do have people that travel to those locations on a regular basis. So Paul, I'm gonna hook you up with Dan Cobain.
He's our PT for Canada. I was just in Dawson City not too long ago, so I know what you mean by remote. These places are very remote. It's amazing how remote it is. A- and I think Rapid Replicas could do that, because that's the whole
point of saving bandwidth. You're not replicating the data, you're moving the band You're moving the metadata, not the data, right? So, it may be useful for this. But let me hook you up with Dan Cobain, who is can talk to you into some more detail, and he can refer you to the, the internal file people if needed.
But it's a great question. And I But I From what you're saying, I think it could be It could work. It's a question of scale, right? It's just a question of scale, and it's a question of exactly how your people need to see the data. Okay.
Yeah. But if you have any other questions, folks, after we wrap, please go to the community forums to ask. If you're not signed up for the forum, please do. There's a link in the chat. It's purecommunity.purestorage.com.
Go get signed up. Ask questions there. All the principal technologists are there. The specialists are there. Our product managers are there. Yeah, John's there. So, I mean, if you want to really annoy a
product manager, ask a really hard question in community. It's our favorite thing to do. Our next Ask Us Anything will be on, 12/11. We'll be diving to Evergreen//One. Registration's open. There's a link in the chat, and I'm looking
forward to seeing you all there. Thank you, everyone. Thanks for joining, everybody. Bye. Bye. Bye.
Jonathan Carnes
Senior Manager, File Product Management, Everpure
Antonia Abu Matar
Consulting Field Solutions Architect, Everpure
Don Poorman
Sr. Technical Evangelist, Everpure
Got questions about managing files with FlashArray? Get answers.
In this episode of Ask Us Everything, we’re talking file storage—without the headaches. We’ll start with a quick overview of how Everpure FlashArray™ File Services simplify file management, helping you get up and running quickly, manage data stress-free, scale with flexibility, and protect your files with confidence.
Then it’s your turn. Bring your questions about capabilities like intuitive management, snapshots and replication, or seamless virtualization support. Our experts will help you cut through the complexity of legacy systems and build confidence in keeping your files available, secure, and easy to manage.
We Also Recommend...
Personalize for Me
My Updates
No updates at this time.