47:21 Webinar
Guarantee Data Availability: How to Create a Snapshot Bunker
Everpure prioritizes remediation and recovery as the critical path to cyber resilience. Data availability is the foundation for reliable remediation and rapid recovery, enabling organisations to restore data regardless of the severity of a disaster or cyberattack.
This webinar first aired on 23 June 2026
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
Hello, everybody, and welcome to another expert-led demo. Zane, give me one second to share my screen. I am your host, Jason Langer, and I'm joined by my good friend Zane Allen, as always. Zane, how are you doing today? Great. Glad to be here again with you. Always a pleasure.
I mean, I didn't-- maybe you didn't get enough of me last week at Pure Accelerate, but, you know. Oh, not at all. Anytime I can spend time with you, it's great. Such a busy week. It was actually hard to, hard to spend time with anybody. For sure, for sure.
And we've got this I am just gonna pause here on this excellent headshot slide because I just want everyone to realize, like, Zane is so cool, he is too big for the box. So like his, his, his head is too big to fit in the speaker box, so. Yes. But anyways- Not untrue. Not untrue. Yeah. I'm Jason Langer.
I'm a senior technical evangelist here at Pure Storage. I focus on all things cloud, cyber res, and virtualization, which is kind of a handful if you think about it. And of course, Zane, you are a principal technologist. You wanna kinda touch on what that is just real quick for folks that might not be aware?
Yeah. we're actually principal technology strategists now, but same, same role. Oh, right. so strategist. But that's, that's really where we sit, on the specialist team. Definitely on the front and field facing side, working with sales teams and our solution
architects and helping customers, solve complex problems in their environment, you know, mainly with EverPure technology. Um- Nice and then do a lot of voice from the field, road maps. Sure. And I get to do a lot of events like this. So you're supposed to be a smart guy. Supposed to be. Yeah. Yeah. Yeah.
You are a smart guy. That's why the head's bigger than the box. Right. Exactly. That's what I was trying to lead into. Yeah. But anyways, enough of our long-winded intros for everybody. We are here again to talk about EverPure's layered data resilience strategy, and I'll put
up this slide because everybody loves layers. I've tried to think about what my favorite layers are, Zane, but in this context, it's definitely cyber or data resilience. And for those that might not be familiar with EverPure, or maybe you are, like, we kind of view these things, and granted, these are pretty compact, right?
Because we can't fit everything on a slide. But you kinda look at how we protect your data as you go up the stack, right? At the very bottom we've got data at rest encryption, we've got our immutable snapshots as you keep going up. You've got indelible snapshots with our safe mode, which is where our ransomware protection
comes in. And then we get into the meat of what we're gonna talk about today, which is kind of the top layers, which is really dealing with what I would say data movement or, you know, the, the more common term of replication, where we get into asynchronous replication, which is our Active DR, solution, and then you get into synchronous replication, which is Active Cluster.
And then those, right, Zane, are gonna lead into our, quote-unquote, "data bunker," which always kind of makes me think of a cold, damp data center, as opposed to maybe just a s- you know, a tertiary site. But that's kinda where we're gonna land in this conversation today. So if you're not familiar with the bottom three, that means you missed part one.
Yeah. Yeah. So if you're not familiar, maybe you're new to EverPure, maybe you're researching us, if you're not familiar with, like, our indelibility and our snapshots and our observability when it comes to cyber resilience, have a look at the first video Zane and I did. What was this, like a month or two ago, Zane?
must be warmer 'cause I'm wearing a sweatshirt- Right in that picture and- Might've been, uh- It's like 85 degrees here today, so I'm not wearing that. That's good, right? But scan the QR code. Yep. Go take a look at that. Yeah. Because we're gonna assume some of that knowledge is already there for this next part
of the conversation. So, let's talk about replication. So we kinda, I mentioned we've got the top layer of our cyber resiliency layer cake, which comes into asynchronous and synchronous replication, and in our words or our terminology, that kind of flows into these three pillars. As you go left to right, it gets more, I don't know, it gets le- what's the right word, Zane?
It's, it's better protection. Lower RPOs, lower RTOs. It's just- Yeah more better. It's, it's Well, I think they all fit a use case, right? They all fit a- For sure a need, you know, depending on what you're
trying to protect against. Right. For sure. Um- So that's where we start with is, like, Async DR. that's where your RPO, think about it as in hours, right? So you're, you know, that's not, like, mission critical stuff. Think about, you know, your, you, you could
lose something and maybe brings it a couple hours to bring up. But the RPO can be in minutes, but the RTO is gonna be, like, a manual failover. There's no, automation built into it. You're gonna kinda go in and have to push the buttons and pull some levers. what it, how it works is it's basically replicating some snapshots on a schedule.
for your workloads, though, it's, you know, it's, it's cost efficient, you know, because again, you're just trying to move the data the best you can in the time that you can. so for networking, any WAN link, you know, again, it's, it's, it's all about getting something from point A to point B. It's just not about getting it there, I would say, what, quickly, Zane, really?
Yeah. Yeah. And, and it's, it's exactly an extension, but it's basically a snapshot, but we're sending a snapshot to another array, right? Right. And this could be kind of the beginning of what we're talking about from a bunker perspective. Um- Right I need copies of this data.
I have copies of this data right at the production, so I can do recoveries, right? write over a volume, rescan it, make a new volume cloning, all very thin provisioned with us, right? Zero space. but what if I don't have that array? What if I don't have that site? What if all that's compromised?
This is how you get it to that other side. And, and we'll talk about these other technologies. It's actually- Right the base. I like to-- I'll just keep probably reminding that, is all of our replication technology is really in some way built on top of each other, right?
We, we index data, big key value store, and then it's And as we kinda go through here, it's just the matter in which we get it over. The RPOs, the RTO changes. Um- Right but it's all indexed the same way and kinda- Intertwined, if you will. Yeah, and that's why I was like, it's kind of positioned, you know.
These all work, like to your same point earlier, it depends on what your workload and requirements are, but it's like good, better, best, right? Yeah. But there's always, there's, there's, requirements for each of these. And speaking of good, better, that's where we get into our Active DR, which is continuous replication.
So we get a lower RPO because at the In this one, we're not, like, scheduling the snapshots like we are in Async. We're, we're continuously, hence the word in there, trying to move the data the best we can or as fast as we can based off the network bandwidth that's allowed. That's why we say the RPO can be near zero because, again, it just depends on the
availability of the network and the change rate of the data that you're trying to replicate, right? So- Yeah for some folks it could be near zero or 30 seconds or less, but it can be a little higher. Again, it just depends on your networking and that change rate of data. And what the workload impact is, is that gives you a little bit more, criticality to your
applications that you can protect, right? Because you've got a lower RPO, and then you get into a more automated failover, right? I think about leveraging this with And I know it's not called this anymore, but I can't remember the real name. VM- VMware SRM, right? For folks that are using, you know, SRM- Yeah for disaster recovery. Yeah.
You know, you got a site at You got a data center in Seattle where I'm at, and then you've got a data center, let's say, in New York or something and Where you can't do synchronous rep, but you're replicating that data, and you've got a automated failover part where the VMware's gonna handle the VMs. But we're gonna handle the storage part of that, right, Zane?
So you wanna kind of touch about how we, how we talk about an automated failover for this one? Yeah. So what is a It's a nice in the middle, Well, and it's really a built on It's Async, right? It's Async just like Async, what we call Async DR. It's one way. Well, not completely true.
We, we can reverse that, but it's, it's a one-way replication relationship. we talk about the near zero RPO. I mean, it catches up, right? The intent is that I have, what we call, and we'll show this a little bit later in the demo, a pod construct, and we have two pods that are replicating or have a rela-
replication relationship. So I'm taking one pod. Anything volume-wise that's written to one is batched up and sent to the other. And yes, depending on when you have the failure, you could maybe not have caught up. but eventually all that data catches up.
But, but I think you were getting to, because you can automate the failover is, unlike with Async DR, where you're gonna have to orchestrate it to a greater degree. I'm gonna maybe have to clone that snap. I'm gonna have to write over a volume. I'm gonna have to take those steps, make a new volume, especially at a DR site, right?
There's no volume necessarily there yet. And then I have to attach that to something. And of course, you can script all this, but there are steps. With ADR or Active DR, it, it can be all, what I like to say, pre-plumbed, right? I have those volumes in that pod relationship always established.
They have their, they're, they're routed or they're connected to hosts. I know exactly what that failover plan's gonna be. I can test that failover plan. Mm-hmm. this is what I really like about Active DR, spending a lot of time in data protection, is I can do the same actions, what we call
promoted, to bring that data up, test what that failure would look like, and then shut that down without keeping any changes, and all the while that replication relationship is still happening, right? We're still batching it, and we can commit those afterwards. So maybe unlike Async DR, I might not do the same procedures for a test, right?
I can actually- test this the same way as well. So, it just gives you a lot more predictability. It's, it's, you know, it's consistent. You know, near zero RPO, and you can do those automated failovers. And you mentioned SRM.
it works. It works with that technology, right? We have, we have a, an integration where, hey, that can be coordinated through that as the orchestrator, but it's still using our mechanism underneath. Right. Right? This is It's, it's journaling, but we-- You don't have to maintain the journal, right?
We kind of- Right we kind of do all that behind the scenes for you. And you said something that, you know, I g- you know, in my career, I grew up in backups and DR. We didn't call it DR back then, right? This was the early 2000s. It was just- Yeah, yeah off-siting tapes or whatever.
But, um- Yeah. That was the DR you said testing, right? And that's one thing that's very Obviously, the key part of any sort of resilience strategy is it's kind of reminds me of the Seinfeld ep- episode where you can take the reservation, but can you hold the reservation?
It's like I can take or I can make the replication, but does it actually work? Have I tested it? Uh- Yeah and that's kind of the next thing that's really great is, like, obviously test, test, test, test, right? You just can't always assume that stuff is working. So, that's a key one, Zane.
Thanks for, for mentioning that because I think that's sometimes something that gets overlooked is like, yeah, I'm moving the data, but it's like, well, but can you recover the data? Um- Oh, and then, like, what we're trying to get at here is, like, I might I'm gonna save data off-site.
I might even have what we're calling, you know, in the layered approach, I might even have this bunker site where I'm kind of fading it in there, and I don't use that for anything else, right? Maybe I- Right use it to test the backups or, you know, st- scan them, right, to see if there's an infection or if it's for a DR event. you need a really easy way to, to kind of fire that off once in a while so you can make sure
that everything's gonna work the way I expect it to work when, you know, I don't want to say emotions are high, but when, you know, when there's a lot of stress, right? Where there's a, there's, there's a, there's a down event. Yep. And, uh- And I know we're, we're gonna- And it should be tested when you recover. Yeah.
Sorry, Zane, I meant to cut you off. And then finally- top level, I don't know if you can call it the Cadillac. I know you know that example. We have active cluster, which is our synchronous replication or, you know, some people will call it stretched. You know, that's where your RPO is.
I mean, it's zero. It's just straight up, you know, we're, we're, we're mirroring writes or we're synchronous writing and reading. The RTO is, again, automated. You know, that failover is, you know Again, I'll use the SRM example, right? Like, it's not us that's really the, that, that slows down the RTO.
It's like how fast can you boot up, boot up those VMs, right? Like, the d- 'cause the data is synchronously written and read. Now, granted, this is obviously- Intended for what I would call campus networks or, you know, maybe a hospital, because the networking requirements for this are quite high compared to, like, Active DR and Async, where both those technologies will, again, will work over
any link. It's just, again, how much data can you push through the pipe? We're, we're dealing with active cluster, and how Zain I mean, how Zain will get into it, is since we're, we're acknowledging those rights on separate sides, we need, you know, five milliseconds around trip time. Zain, I believe it's 10 gig connectivity between sites, if I remember right.
You can correct me if I'm wrong there. Yeah, you can- But- It can't be more than 11 milliseconds. Yeah. I mean, that's, and that's the distance. Yeah. So I mean, again, that's why I say campus- Yeah or, you know, or your data center is very close, right?
Met- metro, say metro area, right? You're not- Yeah. Yeah with Active DR in my example, like I said, I was replicating from Seattle to New York. You know, unless you've got a lot of money, you're not doing a active cluster from Seattle to New York- Yeah just because of physics and network bandwidth, right?
And of course, this is for always on applications, stuff that just can't go down or, again, maybe your business, like I s- you know, I've done this in hospitals, right, where they've got data centers, quote, unquote, in building A and building B because, you know, it's a hospital. They, they need their systems up and running. Yeah. so this is basically the, the top tier cream
of the crop. But Zain, we spent a lot of time here, and we will get to the demos, I promise everybody, but when you kind of think about how we put these together, again, the, the first call that Zain and I did is really, like, data center one is the local protection. That's the first couple layers of our layered resilience, right?
And that's where you get into the, the snapshots and the safe modes, and you're really basically protecting the data locally. When we get into what we're talking about today, I'm just gonna call it data center two because, I mean, again, back to whether it's a campus or it's a, a, you know, you're in a building or whatever, you got these WAN links.
It's, it's basically you got a secondary location, and this is where the continuous replication comes in with active cluster, active DR. You get those automated, automated failovers. You've got independent safe mode protection on the secondary site. So again, that's something we talked about in the previous video.
But we're protecting your data in site one from ransomware. If you're replicating that data somewhere else, obviously we wanna protect that data from ransomware as well, right? so this gives you, let's call it site level resilience as opposed to just, like, local resilience.
And again, this is, can be any- either one of those options we talked about, right? At the high level. It could be active cluster, it can be active DR. Zain, it may be both of them tied together, right? Depending. Maybe. And then finally, then you get into the
isolated recovery environment or the d- the damp, drain-y data bunker. That's what, it's just what I keep thinking of. Sorry, Zain. It's just, And that's where you have this air gap recovery, right? So you're using the other technologies in your site one, site two, or data center one, data center two to keep your, you know, keep your
data running and whatever. But then you're using this third site or third location that's off network or, and you're using firewalls, or you're really isolating it the best you can again, so you have this location to keep snapshots and this replicated data for, like This is the full belt and suspenders approach, right? Like, this is protecting you from if something goes wrong in your primary data center, you've
got this off network environment. Something you wanna talk on that, Zain, as well? Yeah. And you This is And you know, you can, you can, you can cover that with network. you pro- you're most likely gonna have Well, you should. You should have compute there, right?
So that you can run these things. maybe a plan and tools to actually clean them. So like, you're gonna have your clean room and your damp, bunker maybe. You know, make sure you have that fenced off. and meaning that nobody can compromise it, run any data, maybe even, you know, let's say
compromise the actual environment and get to the data and change it in any way, right? So also encrypted throughout. so you may take extra steps when you're recovering this on that side. Mm-hmm. So let's say your s- snapshots are probably a great fit for this because I'm gonna just clone them out to a volume and do the steps.
maybe I automate that, but I, but I don't have that pre-plumbed. if I did do something like ADR, it might just not be attached to hosts, but I could s- simply bring that up and then, and make those actions. but it's really in a, environment that's meant for exactly what you're talking about here. I'm gonna bring snapshots back or I'm gonna bring them up there to inspect them so I can
make sure that I, I'm not bringing any, nefarious actors back into the environment. s- I've seen these even if it's just, like, the viable product. Like, hey, what do I actually need to get my business up and running? So I'm gonna, you know, dedicate to that. Where a DR site, you know, if you're gonna separate these, might be, hey, I wanna be able
to run everything, or at least a majority of things here. I might even run here for some time before I fail back. right. So it's just kinda two different strategies. So you definitely have snapshots like we talked about last time, really close to the data.
I'm a big believer in having your recovery close to where you need to run it, so production's great. There's a lot of scenarios where I can just bring that, that back. But there are big scenarios, especially a ransomware event- Yeah where you might have to isolate. That's first to make sure, hey, it might take me a little longer to get back up and running,
but I know when I get back up and running, I'm not just bringing the problem back, into my environment, which is gonna cause further downtime and maybe even data loss right at that point. Yeah. I mean, this is definitely what comes to my mind, and I know I'm an analogy guy. You know that, Zain. This is like the break, break glass in case of
emergency, right? Like you've- Yeah you've had a catastrophic event potentially. Like you talked about ransomware. Like we've talked to customers where, you know, somebody's wanna actually confiscate their arrays, right? Yeah. Because of forensic research. Yeah. Yeah.
And like this is where this comes in, where it's like you still need to recover that data. You know, some- somebody or some agency or something's like, "Hey, we need your, we need your arrays." Which sounds crazy, but true story, right? Like that does happen- It's a crazy scene unfortunately. Right. It's a crazy scene. I've had a customer where the FBI's come in
and said, "You don't get your arrays anymore." Um- True story. So it, the definitely, you know, kind of the- Well, that, that's why we offer an SLA, right, on our Evergreen One contract, where we'll, bring it Well, you know, it'll be kind of preset that you'll have an array that matches. That's actually the technology we're gonna use.
We're gonna do that replication over, get those safe mode snapshots under there, and then use those to recover from it. So this is It's more of a, I don't have everything set up ahead of time, but I have a way out kind of insurance policy. Right. whereas, you know, an isolated environment or the IREs, as sometimes they're called, or
secure isolated recovery environments, you're kind of doing that u- up front so that it's minimizing the time to actually get to action, clean, clean the data. Yep. yeah. So it's just, it's just another strategy to, to fully, to full An encompassing strategy to, to help, uh- Right. Right organizations mitigate against risk.
Great. Well, I think that's my last slide, and I've We've probably bored people for 20 minutes when we said there's gonna be demos. Yeah. But we are to the demo part. Zane, I'll stop sharing my screen and you can take over from here. well, if, if you just wanna go to the next one, we can- Oh.
We can run through. Okay. Oh, you've got it embedded. Got it. Yeah, yeah, yeah. So am I just pressing play? Yeah, you know, we have time here, so I might actually do the next one live if that's cool for, you know, for- Yeah. I- Maybe put it to the chat.
I don't know. Try live. People put in chat, do you wanna see Zane try to do it live or do you want to see him do it- use the recording? So Maybe then we can break down a little bit more of those other technologies. What, what this first one really is, is just an extension of our last webinar. So if you didn't catch it, kind of building on top of those protection groups that, that we
showed last time, right? Where we had a policy that was, taking snapshots, some, some periodic, I think it was like two, two, and two. and then we were keeping that in a safe mode. a safe mode, I guess. Applying safe mode to that protection group so that those snapshots are indelible.
Yeah. Just real quick, Zane. The audience has spoken. They said live. So they're gonna have to bear with us if we mess this up. Okay. But before we do that, so is that the second demo or- We'll play, play this one. I'll do the next one live.
I'll do the next one live. Okay. Do you want me to play- Yeah this all the way through or do you wanna just tell them what- Yeah, yeah, yeah. Yeah, yeah, yeah. Okay. I had set up two demos. I'm gonna pause. We're just gonna talk
about just snapshots here. Okay. So let's, let's just keep this here, 'cause this is kind of building on our last one. So if you remember in our, our, last webinar, if you didn't, this is kind of the volume we were dealing with. And again, everything, everything I was explaining, this is attached to a host group, right?
And then had this protection group. There's volume snapshots, there's protection group snapshots. Protection group being where we made that snapshot schedule, like I said, to two hours, two days. and then keep another one sh- snapshot for two more days, right? This is all local.
This is, hey, this is on that array, now I can do recovery from it. Yep. Clones, whatever I need to do. So to get to a replication, this is that async snapshot part. It's really easy. we just really wanna set a target, right?
Where are we gonna send this? And then we wanna set a schedule. So this is already set up. Should show the replication link after this. So basically there's a replication relationship between the two arrays. Do that very easily with a connection key and an IP address.
So that's available to me to replicate to. And then it's just as easy as setting this replication schedule, right? So this is They're gonna take another snapshot and send it over, right? So this isn't It's building on the same snapshot technology. All the snapshots are from the same place.
Our snapshots are really volumes, right? It's just metadata. but this is taking it a different interval. I can say when it runs, how often, when it doesn't run, and then I give it a retention. You know, how, how often or how many of those do I wanna keep, right? So I'm replicating this, I wanna keep two days.
And then I can do a further retention strategy here, and I got some options. So maybe I wanna take You know, I got I'm running this many in one day. I wanna keep everything for, for this long. So four every four hours, two days. And then maybe I wanna keep a longer term, right?
Almost like a roll-up, if you will. Yep. So I'm gonna say, hey, and then I'm gonna keep only two of those or one of those for the day for this longer period of time. and that's kind of what that schedule will give me there, right? So whatever that policy is, we're talking about snapshot bunker.
Maybe I want 30 days of daily and then I kind of want monthly, or I want weekly, et cetera. I can also set it up where there, there is no retention, like, after that, right? I just have this, the first retention. or I can say indefinitely, right? Hey, after I keep that, that main retention, keep this many forever, right?
Really, really whatever your needs are. You got an error here. I'm gonna pause. Is this something you wanna talk about or a message? Yeah. So safe mode, I still wanted to bring that up. Safe mode still applies here.
I can't go in here and defeat this, to, to take my snapshot schedule down, right? So if I wanna keep a certain amount, I have to make sure that this fits it right. Again, just, just like we built on last time, I can't do a snapshot policy. I can't edit that snapshot policy to run out, you know, safe mode. Safe mode works.
Oh, you paused it. You could, you could keep going. Yeah, I'm just curious. I mean, we've got, I don't know, 50 some odd folks. I'm curious if any folks in the audience are If you're Pure customers, I'm just Maybe give me a yes or a thumbs up or something. Are you using safe mode? Like, this is Safe mode is a key construct in
our cyber resilience strategy, so I just kind of want to It's like if you, if you're a Pure customer and you don't have this enabled, you know, go back and check your arrays, 'cause you definitely wanna turn this feature on. Whether or not you do any of this s- replication or anything else, even for your local protection, you wanna make sure you've got that turned on. Yeah.
So a little PSA there- And that's- Zane, for safe mode. It's the s- it's the same place. You look under there, that ratchet. It's We're doing it by the protection group. So this safe mode is applied at the protection group, so it applies to all the rules inside.
Yeah. Yeah. Yeah, 'cause otherwise, I mean, like, I have admin access, I can't delete the snapshots, but if I can just go edit the policy. Yeah. Whoa. Not only do you have a big head, you got a big mouse cursor. Look at the size of that thing Sorry. Yeah.
Mouse, mouse shaming me. That's what you get for g- letting me control the recording. So anyway, this is just an example of how I was, I was going through all that and trying to set it, to a different policy. So we're just gonna go ahead and take the, take the default there.
but that was just to illustrate that, right? Kind of over- Yeah over-exaggeration. I can't defeat this. I can't give it less protection than I set up upfront. Now, we don't I just like to point this out. It's why I kind of get it out of here.
You'll see it's kind of gray in, in Barker. I set that schedule, but now I'm wondering why I don't have replication. you still have to turn it on, right? I can set this policy, I still have to enable it. I just kinda like to, to point that piece out.
So you'll see it's, it's grayed there. I got the target. And, and I would have to, I would have to ratchet that to, to make that on. So really that's There's nothing more to it. this is the replication connection.
like, you, you do that by making a connection key, putting the IP in on the other side. So it's really easy to build those connections. Once I have that connection, I can start doing all these replications. we'll s- we'll see in the next one where we do, kind of replication links, kind of like that logical link, but really that, that relationship is already built.
So real- I guess one question I'd have, Zane- Yeah is And I guess we didn't cover this before, and I'm, this is my bad. The replication for Async, Active DR, so on and so forth, are those additional licenses, or do Like, how does, how does, how does a customer, if they're new to Everypure or they're researching us, like, how do they get that technology?
they- Is it a license key they gotta put on the box or? They do not have to put a license key. I like to say, like, you know, limited time. If you buy now, it's free. but no, it's all built in, right?
Purity, Purity has this. In fact, what I'll show in the lab, we take We have taken Active Cluster and allowed you So today you would And you could still do both of these, so you could have two, two types of, strategies here. I do Active Cluster. I got that HA relationship that we talked about.
but I can use Async replication, right? I can send to another site a snapshot, right? And that might be enough. Hey, I just wanna have this high availability, and then I just want that snapshot offsite, or I'm gonna put it in an IRE and I don't I'll use it for my DR site in both.
Whatever, whatever the case may be. We've added now the ability to use the other Async technology, which is the Active DR, right? So now I can put that, and it's, it's interesting 'cause those are basically two types of pods or two types of replication, using pods differently.
And that just came in an upgrade, right? It was the last version of 610, which, is already out, and 611's on its way. And you just upgrade, and now you get that feature. So that's just- Nice kind of a testament to, to Evergreen. Hardware and software are just kind of included, so as we give you better and better
features, you can just take advantage of them. Nice. All right. And then now So do I turn it over to you, or is this the live part? Everybody get their, get their nets out to catch Zane. Uh-huh. So I'm gonna stop sharing, Zane, so.
Oh, I have to ask you for sharing. Somebody has to give that to me. You can tell we practiced this, guys. I knew if I took it to a vote, this would be the answer. All right. Do you see the share button then?
I just wanna make sure you're I shared it. Are we not seeing it? I don't see it, but maybe the folks in the chat I'm just seeing you and I. Yeah. See, there Now I'm seeing it. Ah, okay. A little delay.
All right. Everybody else, I hope. Yeah, this is what happens when you do it live. Okay. Well, fun, right? This is what you expect. Jesse says they can see it, so. This is what they expect. Yeah. Yay, you figured it out. Okay.
So, kinda using a different, different lab for this. so it's still But still kind of extended, thinking about, maybe I wanna take that same volume, I didn't name it the same here, and then take this different relationship, right? I wanna use Active Cluster. I wanna use A- Active DR.
I wanna put them together. So in this example, we're actually really kind of more talking about what we talked about before. I have three arrays. So let's think about these as it could be two sites maybe.
Maybe that's in the same site. Maybe it's two different buildings 'cause we're gonna do Active Cluster, and that's a third site. these could be three sites, as long as you can support it. and remember, Active Cluster is, the thing that requires a low latency, but it is committing writes to both arrays.
So it's writing to one, sending it to the other, waits for the commit to kind of come back. And the replication links will be different. They're not gonna be different. We're just gonna see that I have two arrays. So like I mentioned before, you can make these, establish these relationships
by getting that key. So we'll see in this environment one goes to two, but it also goes to three, right? So one goes to two and three. Two goes to one and three, and then three can see both those arrays. And we're actually gonna see the relationship here is just Async.
This one, it's very specifically like, hey, one and two are in a sync relationship. This is an Async, right? That could all be different, but that's how this lab is set up. So what we have here are these things called, pods, right? So this is what we're gonna, we're gonna do here.
Let's call this SME. Oh, that is not typing. There we go. Can you explain what a pod is just real in 30 seconds or less- Yeah just in case folks aren't familiar, Zane? Yeah. So that's exactly. So open, open this up.
What a pod is, is, our construct for fi- kind of replication outside of, outside of just the regular Async, right? We just use the snapshot policy or the protection group, like I showed before, for the snapshots. In this relationship, we're gonna actually establish, hey, I got two arrays And how are they gonna be related from this replication, re-relationship?
So we have active cluster, which is zero, active DR, which is near zero. the difference there is one pod, and this is gonna be stretched. I'm gonna stretch it across a synchronous relationship. And the other one I'm gonna do a replication. I'm actually gonna have two pods, and I'm gonna replicate one to the other.
And now any volume I move into the pod is going to take action on that, right? If it's an active cluster, it's gonna, it's gonna commit the rights of both sides. If it's active DR, it's just gonna batch those rights on one side and then send them over as soon as possible so that that c- that, relationship can catch up. can think of them as a consistency group 'cause really any volume I move in is going to
do that replication action. Whereas opposed before, any volume that's assigned to a protection group, and by the way, I can have volumes in multiple protection groups, are going to follow that schedule, right? So ho-hopefully that makes sense. it's as easy as moving in volumes.
I could have multiples. again, it's like a consistency group. If customers have seven, eight, nine volumes in here, maybe it's all one application. You want it to take replicate all the same, right? You wanna be able to bring it up the same. You want it to have the same HA.
I can do target volume groups here, right? 'Cause now I'm looking at another array, an active cluster. we, we can do what they call non-uniform, meaning you could use it as a failover event, right? The data's in sync all the time, but the host on both sides don't talk to both arrays.
So, you know, I have to fail over over there, but the data's already on that side. But normally, active cluster So it could be a good DR, I suppose. But normally, active cluster, all hosts can see both arrays. and then that way if I lose a host, or I'm sorry, lose an array, the other host just picks up, right? It's just other paths being, being, shown to
those hosts because that volume is stretched, right? It's the same exact signature, same exact volume on both arrays being presented up to the host environment. Hey, Zane, just real quick. We got a question in the chat, and I think it might be re-rel, related to what you're talking about. But Thomas is asking, "Is there adva- is there
an advantage to configuring multiple pods for the same workload type in active cluster?" Workload type. So you're only gonna have A volume can only be in one pod. Well, actually, I'm gonna show how you can use active cluster in ADR, but you can only have this one pod relationship. So I can't have a volume in two pods, if that's what he's asking.
Yeah. Maybe Le- but let me know in the chat, Thomas, if that helps. You can have it in two protection groups because it's just a snapshot schedule. Yeah. But since I'm now building a async or synchronous, you know, near zero or zero RPO relationship, I'm gonna apply that link, that
to one pod. That's kind of why we came up with this feature is, like, hey, I want I love a active cluster, but I also wanna have active DR 'cause I want the near zero RPO. I don't want just an async snapshot once an hour or every five minutes. Yeah. so we, we built this technology so you could
tie them together. Yeah. So I hope that, hope that answers that. Let me know, Thomas, if that, if that helped or not. Yeah. I think we're gone just or mi-mindful of time. Real, real quick, hit here.
Again, safe mode, always important. We're still not defeating safe mode. I have to assign basically a, a similar protection group, so quick, quick hit on that. Protection groups outside of a pod are different than the ones inside of a pod, so you're actually going to build, protection
groups inside to match that same technology. So anything in the active cluster, I might not be doing the async rep anymore, but I do still want the snapshots, right? I still want those local. I might have to write over those.
And so anything written on one side in active cluster ADR is also gonna be written to the other side. So again, it's kind of designated here by these, colons. So it's, that's the pod, and that's the protection group inside. So as long as those match, let's start over there, I can move it in.
So, hey, same protection level. Great. You can move it in and out of a pod as long as you're not defeating safe mode. I guess it'd be the easy answer there. we already have that replication link, like, or that replication relationship between the arrays, right? If you remember, FlashArray-A run.
FlashArray-2 was in synchronous. I'm gonna go up here, and I'm gonna choose this, and I've now committed that action, right? I have that volume. It's going, stretching it across array one and array two. Interesting fact I kind of mentioned before, everything builds on each other.
it's re-syncing. It's syncing. Our initial sync is actually basically using snapshots to get up to the point that it can fail, go over to our active cluster engine, so very efficient. I don't know if it was implied, or it was implied, but if I mention it, you can iterate through all of our replication back and forth, and you'll never lose that baseline as long as
you don't change that data, delete it, you know, read and write it again. So basically, I can go from async to active cluster to ADR and back and forth, and it'll always just be changes that will baseline. So now if we go over here to two, I can see I have this pod. It's online. It's promoted.
It's got the volume in there. It's in the protection group, and bam, we now have a stretched pod, two arrays serving the same data. And again, I can have the s-same host pointed to it. I can have different hosts pointed to it, et cetera.
But now we want to replicate that over, right? And before this feature, we would've gone to the protection group, and we could do an async out maybe to the third site. You have that asynchronous relationship. But what we have added is also the ability to do
active DR, right? So that near zero RPO we talked about, set up that volume, pre-plum it, and then it's basically just promoting it and demoting it. I don't have a remote pod. I can com- I can actually create it here, but basically I'm creating a pod on that third array, right? FlashArray-3 So then just to pause, I wanna
make sure I'm fol- I wanna make sure I'm tracking with you and hopefully the aud- 'cause, 'cause the audience probably isn't, but I'm, I'm not. So in our little scenario of our beginning slides, kind of what you just set up was when I said data center one and data center two, you've, you've picked a replication technology. In this case it's Active Cluster, right?
So you've gotten this site level protection, for lack of a better term. But with this new feature, this feature you're going to show, we're leveraging potentially the data bunker, the third, the IRE, the third site of now- Yep using ADR to move a copy of the data to this air gapped, damped, damp bunker- Yeah for that known good. Just so I wanna make sure I'm, I'm tracking. So.
And before this, you could still do this too. So you could go in here and say, "Hey, I have this Active Cluster relationship, but I wanna just use an Async snapshot." So you would go and, and again, we build a we build on top of each other. This protection group that's inside the pod, I could just set up a replica schedule and give me a one-time snapshot out, or not one time,
but I can do a snapshot on an interval out to- the other site, roll it up, right? Keep it long term. but what this is doing is allowing us to have, Oh, I'm in protection group. A near zero RPO, right? Or maybe I do wanna pre-plumb this, or maybe it's at a site that I don't want the compute
connected, but I can at least have the volume ready to go. When I promote it, I can then just do that connection process, clean it, bring it back online, whatever, whatever the case may be. And again, by everything in that protection group, it's like a local snapshot, will also follow over there. So I'll have vo- I'll have snapshots I can
recover over those volumes. I can clone that data into another, another volume on that array, right? I, I have all this flexibility with it, but I don't have to worry about when did I do my recovery. Oh, I, I run a snapshot every hour.
One hour and two minutes in is when the event happened. You know, I'm great, but if it's one hour and 55 minutes, now I've lost 55 minutes of data. I can get synchronous relationship that's, that's, you know, close to zero as possible. Um- Just wanted to make sure I was tracking. Yep. And so now I can tie these together.
This is the brand new feature. Usually you would either do Active DR or you do Active Cluster, and Active Cluster could have snapshots. Active DR was just Active DR. now we're bringing these together just to give you a better Async, technology. you'll see this, the replication link, it's got this directional.
And if I go over to three, right? It's, it's the other way. So really, really obvious what the replication direction is. now this one is not replicating backwards, but kind of fun fact here is, let's say I do recover to this site. It is a DR event, or I have to run here for a
period of time. I can simply do a demotion. I'll show you a promotion in a minute. And I can reverse that replication right back to production. So another key thing when I was talking about data protection in the back, back in the day was, great, I built this site I'm gonna replicate over.
What's my plan to get back? It's probably very similar to what I did to get there in the first place, and it's probably gonna take some time, right? I'm gonna have to plan it. this one makes it a, a lot easier to kind of bring that data back into your data center, after maybe it's run quite a, you know, some
time over the DR. But talked about earlier that you can test it like you recover, really all that needs to happen to bring this data online. So right now it's offline. we call it read-only, but it's basically hidden. It's not connected to anything, right?
I can have this, I can have this connected to host if I want. you know, we go back in here, I can, I can, I can attach these volumes to, to, to host ahead of time if I need to, but I can also leave it offline and do it at that, at that point. But I don't have to do anything more than promote this volume now. So once I promote that volume, it's actually gonna make it read and write.
If it were connected to a host, that host would be able to start re-scanning it and start utilizing it. If it's not connected, I can go in here and connect it now. This can all be scripted however you want. This is the thing SRM will go in and actually automate as well, 'cause it can essentially
call to our APIs, right? So a lot of flexibility here. Yeah. But once that's promoted, again, I can do read, write. But what I like to point out is that it's still replicating.
I'm not losing any data. So if I'm testing it, I'm just seeing how that works. I'm going through the motions. What would my runbook look like? What does my automation look like?
I'm still capturing changes just in case, you know. It's the inevitable, right? an event happens while I'm in my testing phase, or I can have my testing window be a little wider to really give a full test or, you know, do all my due diligence. I'm not worried about losing any data in that time as well.
So really, really easy and flexible to work with. I'm still doing the Active Cluster here. I have that data replicating over here, and when I'm done with it, I can simply Oh, wrong button. I can simply demote it.
Choose not to save any changes. It'll take the, the changes that have come over in that time that it was offline, it'll roll them up into here, and the next time I promote it, I'll have up-to-date, up-to-date writes to this, to this array. So storage is still really hard.
There's a lot of journaling. There's a lot of things happening in the background, but Pure handles all that for you inside the tech, right? You just make the relationship, put it on the array, promote and demote as needed, and, and there you got it. And then go get your coffee, right?
Go get your coffee. Like I mentioned before, I mean, you can clone these. you can make You can use this for other testing, right? I got this site. Maybe it's, maybe it is a fenced off environment.
Maybe it is an IRE. Maybe I also want to utilize it for some other testing, right? Again, now you're introducing work in that environment, so you gotta be, you know, kind of careful what you're isolating there. But, but a lot of flexibility with this data. Perfect at that site. Yeah.
That wasn't so bad, doing it live So that worked, I think Sorry, what'd you say? So we kind of tried to combine a active cluster and ADR in this one function. So if there are any questions, happy to, happy to answer those. Yeah. Looks like Thank you, Joe. Looks like Joe and Thomas have had some conversations back and forth in the chat.
I will open up, though, for I mean, I'm assuming you're, you're Are you done with the demo, Mr. Zane? Yeah, yeah. Okay I'll leave it open in case we need to- Yeah, I just wanna make sure double down on anything. any other questions from the chat or any other things?
I, I One thing I wanna show or take away from this is it's kinda hard to represent a lot of this in a demo. Zane, you did a fantastic job without kind of seeing it, but, like, it is really just that easy as far as, like, configuring- Yeah and setting up, of s- of course, assuming My network friends love when I use the term plumbing, but as long as the network connectivity is there, it's, it's very easy to
go in and configure- Yeah the replication schedule, the pods, the protection groups, all that stuff. Like, you know, one of the things that I've always liked about a FlashArray, and I go back to 2012 when I touched my first FlashArray, is just the simpl- Yeah simplicity of purity in the interface- Yeah and how it is to configure all that stuff, so.
And we didn't, we didn't have Active Cluster back then, so the original point, and then it came out. And then Active DR didn't exist, and then it came out, right? That came out about right before I got here, about five years ago. So, so that code got better and better, and these features- Yeah got built in, especially as arrays got better, you know, the CPUs got
bigger- Right the performance got better. So we just kinda layer, again, not to, to call the pun, but we layer in these, these capabilities as you build these layered, environments to, to fit your needs. What is your favorite layered thing, Zane? Mine is, like, a seven-layer burrito.
So my f- the thing about a seven-layer burrito, isn't it just really all mixed together? Is it layers? Not if you put it on right. Not if you put it on in layers. What if you unroll it? But if you- so I'm not a big chocolate fan, but I've always liked tiramisu.
But what my wife- Oh, there you go found 'Cause I am actually more like lemon. I like that sort of thing. She found these tiramisu lemon. Amazing. I think they- Go ahead sell them at Trader Joe's, I think, so. There you go. Well, in the chat, I'd like to know what your favorite layered thing is.
Mine is also, like, a Kit Kat bar, 'cause I like the chocolate with the wafers and the layers, but- Okay. I'll give you that. That's good anybody else has something to offer there. Zane, yeah, I'll go ahead and share screen if you wanna go ahead and stop- Yeah 'cause I, we're not getting a lot of questions in, and I wanna be mindful of people's time.
Think you should be seeing my screen. There we go. I wanna thank everybody for joining. I also wanna give a plug to our customer community. looks like there's several customers in the chat, Jesse and, Thomas, or at least people that are looking into it.
This is a great resource for you can go in and actually talk to other EverPure customers that are using these technologies, ask questions, you know, help the community. Some- Zane actually pops in there from time to time to answer questions. Yeah. what I will say, and I always kinda put this out there, is, you know, this is not intended for, like, tier one support.
Like, if you're actually having a real, like, production outage or technical problem, don't go to the communities page. Like, I mean, you can, but just call support or something. But just, just know, like, this is, you know, this is people helping people. But, you know, you know, if you're running your business- Yeah, we- make sure you're calling, calling support or something.
Questions like, "Hey, I wanna know how this works," or, "Why am I having this problem?" But yeah, system down, there's definitely another path for that. Yeah. talked about, integrations coming out. Um- Yeah. Just a good s- it's good, good information on the, on that page.
Good place, yeah. So go ahead and check it out. but I think with that, I'm gonna say thank you all for joining us. Zane, my friend, truly a friend. I always enjoy doing these with you, so thank you for coming on. Yeah. And for those that don't know, there's
producer Lauren in the background. This is her first time working with Zane and I. Hopefully we did not stress her out too much with our- With our- Why is she gone? our planning, air quotes, on some of this stuff. So producer Lauren, thank you so much.
and for, Joe, for covering the chat. but everybody else, I hope you enjoy the rest of your day and the rest of your week. if you're up in the Pacific Northwest like Zane and I, it's gonna be a hot one, so make sure you got your, your AC on or your, your cool hat going. So thank you all, and we'll see you on the next one. Thanks.
Thank you. Thanks for joining.
Zane Allyn
Principal Technologist, Everpure
Jason Langer
Senior Technical Evangelist, Everpure
Without data integrity and availability, recovery is difficult and may take days or even weeks. Everpure prioritizes remediation and recovery as the critical path to cyber resilience. Data availability is the foundation for reliable remediation and rapid recovery, enabling organisations to restore data regardless of the severity of a disaster or cyberattack.
- Why a snapshot bunker is the cornerstone of effective layered resilience
- Key considerations for architecting the bunker to ensure survivability
- How to set up and operate a snapshot bunker
We Also Recommend...
Personalize for Me
My Updates
No updates at this time.