The Enterprise Data Cloud with Pure Storage

So my name's David Staman. I lead our technical strategy for our new product Introduction in Cloud. Um, been here at Pure for about five and a half years and really glad to be here in front of, um, all of you today.

Hi everyone. My name is Brent Li. I'm a technical director here at Pure Storage, uh, with a focus on, uh, pure Fusion.

So I've been at Pure for also most five years, uh, today. Uh, mostly working on architecting and engineering, the pure Fusion, uh, solution. So, um, I helped write, write a lot of the code in Pure Fusion.

If there's anything that you don't like, it's likely it's my fault, so I'm the person that you get to yell at today. Perfect. Awesome.

Thank you so much. Um, so we're gonna kind of kick off today with a little bit about what here at here we call our enterprise data cloud and kind of our updated vision of what we're doing here for the enterprise. And so when we kind of think about this new vision for enterprise data, we're really thinking about how we are changing things and moving forward and really redeveloping how we traditionally think about our data.

And so when we really kind of think about our data, if our data isn't in control, nothing is because if we just have storage that's kind of there. It's just your standard infrastructure is what, it's really the data. It's what's really important there.

And so when it's time to manage your data, it's not just the infrastructure. And so when we kind of think about this is that once you accept the data, isn't that, um, the infrastructure, it ultimately provides the foundation and has this flip. And so when we kind of think about the baseline, automating storage for your infrastructure isn't just this key point, but it's also tuning the infrastructure for that goal to make sure that you're really, um, handling that.

So when you think about the traditional enterprise data center architecture, it's traditionally really siloed. We've kind of had arrays that were provisioned individually. It might've been for your databases or your virtualization or your file shares or for your backups.

And when we kind of think about that, they were in these silos where you couldn't share capacity. It wasn't performance, it was provisioned for an individual object, an individual workload. Anytime a new service was kind of brought, you had to reallocate all of this infrastructure.

It was kinda one of those things. As I remember back when I was working as an administrator, every time we brought in this new application, there's this RFP and we bought all of this dedicated hardware in order to actually work for that service. That's a legacy way of thinking.

We need to figure out what can we do to ultimately pull together a lot of these resources and then ultimately share them. And when we kind of think about things like backup and cyber resiliency, data governance really gets applied manually and can lead to a lot of gaps. And so when we kind of think about this, that ultimately leads to this inconsistent data management, which means that data is ultimately captive to that stack is if you need to share those resources because maybe some are over provisioned or under provisioned, it doesn't really scale very well.

When you think about virtualization, you don't really have a good record of management. Where does your data actually sit? The best part about virtualization is it can really sit anywhere and all of that is completely abstracted.

And when you think about some of your files, that's all raw data and it can sit wherever. But when you need to access that one individual piece, it's not really there and it's not easily shareable with applications. When you have that extremely siloed.

And again, when you think about backup, there's a lot of manual stuff in there. While there's policies, those policies have to be manually applied and have to then be kind of controlled in that that data governances can be completely inconsistent. And so how we kind of think about this is we ultimately need to transform and not just stay stuck in this legacy complexity because when we kind of think about this, we're still going back to the old ways.

Even when we have all of these new sorts of technologies, we're expanding what we're doing, but in the end, it's not staying the same. And so when we think about what that traditional storage architecture was, it was very vertical built for a single use case. It was very physical, it was very rigid, it was very manual.

It required people doing all of these manual objects. And again, it was extremely siloed. And so we think about some of the things that made this even easier.

A lot of that was the public cloud. It's being able to develop the, deliver everything as a service and ultimately allowing the choice of being able to do certain things and ultimately everything being automated and policy driven. And so we think about what this modern data experience looks like.

It's this modern storage architecture where it's horizontal, we can scale out, we can share resources, we can pull all of this together. It's virtualized, which means you're no longer being very rigid and siloed. Things can be shared.

You can pull together all of these resources and ultimately handle that. Lastly, it's automated. Um, that's the key point is everyone is looking at a way to automate.

Because when things are manual, that leads to manual, um, intervention, it leads to error prone things. And again, automation is that key. And the whole key point here is this unified experience.

Being able to manage everything from a single plane, doesn't matter if it's file block object, if it's in the public cloud, if it's on premises, if it's a hybrid scenario, being able to unify all of that together is gonna be this really key point. And so when we kind of think about what we're doing here is there's this new vision for enterprise data. It's not a product, it's this overall solution where we can think about what we're doing with predictive intelligence, with our unified data plane within Pier One.

Um, by being able to do that intelligent control plane, we're able to create policies and workflows. And Brent's gonna get into this here in the next section, but it really provides this simple self-service scalable consumption model that ultimately beneath the covers, you don't know if it's storage array sitting on prem storage array in the cloud if it's a block file or object, but it really provides this ability to share, share and support all your workloads and all your workloads, whether it's legacy virtual machines, whether it's modern virtual containers, whether it's your PAs apps, whether it's your compliance governance, really all states of your entire journey. And so as we continue kind of down this vision, we'll really think about what we're doing here.

When you kind of think about this as you look from the left to the right, this all starts with our Evergreen one that's delivered as a service. We have the ability that customers can buy our raise, but they can also subscribe to our raises and get all of these benefits with purity. It is that unified data plane.

So block file and object across our entire portfolio. Um, it's resilient, it's future ready. It provides cybersecurity and resiliency built in, whether it's our active cluster, which is our replica synchronous replication, our active dr um, our asynchronous replication, our cloud snap or even our safe mode.

Um, it's all native, it's all built in. It's always turned on. These are things that customers do not need to actually opt into.

They're just there. And being able to provide this unified data services, we're removing those silos. We have a single platform that can deliver all of the protocols available to you and then ultimately run that.

And again, the key point here will be on premises in the public cloud or also at the edge. And so as we continue to kind of take forward here, um, a lot of that will add into the intelligent automation and these workflows being able to provide end users that real time observability and that AI ops and a lot of predictive analytics and insights. We don't have to wanna manage all of these individual silos, but if we can handle the automatic workload balancing, the automatic placement, the automatic provisioning of everything, and really tying this in with Fusion, that's really gonna provide this common operating environment that is going to to handle that.

David, Yes, that courier Futurum group. Um, are you implying or or stating in some way that uh, the right way to approach a pure if you're gonna pure storage, if you're gonna buy, deploy and use, you should start with Evergreen one. Is that like a future vision or a current vision or am I reading too much into it?

So it's kind of both. So we kind of think about Evergreen. One is a customer, it doesn't necessarily need to know upfront everything.

What they're doing is say, one of the benefits of the scalable model is when you buy an array, you need to know what you're using it for. Yeah. The benefit up here is we have our Evergreen platform that if a customer buys, for example, an X 20, we can non-disruptively upgrade to the next model up.

That gives 'em additional performance, additional capacity. However, with the Evergreen One service, they're gonna say, I have an SLA of X performance, X capacity, X from a, um, power kind of consumption. And what we'll do is we'll make sure that they always have the capacity available to them with Buffer, they're always meeting their SLAs around performance.

And if not, it's on Pure to actually deliver them. So it's kind of like that cloud operating model to deliver that to the edge. So lemme flip it around and imagine a an acquisition scenario.

I'm a customer or maybe I'm using, maybe I'm not or whatever. Yeah. Um, when I'm working with Pure on, you know, my next six months, next 12 month plan mm-hmm.

I might go in and say, Hey, I need a six one an X one, I need 12 x ones. I need whatever I need. Um, is pure gonna say back to me, well, okay, let's start with, you know, or, or let's start with Evergreen one.

Yeah, we'll get you the X ones, we'll get you what you need. But let's, let's start with this new, you know, unified access point. Yeah.

I think, or are you gonna say, Hey, uh, guess what, you get Evergreen one with it, go ahead and buy your, like is it, is it coming, are you trying to sort of put it first as a strategic question for the customer? Or are you, uh, adding it on and then working through them? Yeah, I think it, it's a little bit of both.

So it's ultimately up to the customer, but again, both options. One, the, the biggest leading question in some cases is one's CapEx, one's opex, and one is more as a service. So if a customer doesn't want to have to worry about making sure that they're buying it and they're upkeeping it and making sure that they're responsible for the SLAs, all of that is now on peer to make sure that all of those SLAs that they're handling are on us.

For example, with our Evergreen one, we also have a ransomware SLA, that in case something ever happens, part of that SLA means that we'll actually ship you an entire array with as well as an FTE, um, very quick. Which means that you don't have to worry about you as a customer purchasing a second array while your other array gets taken away for forensics. So there's a lot of built-in benefits, but when you think about purity, the platform, everything we're talking about in this vision, it doesn't matter whether it's the Evergreen wine or our traditional purchasing model, it's all available to the customer in the end.

It's the, the consumption and the billing and how it's actually purchased. Okay. Thanks.

Cool. Thank you. So when we really think about this, the, the fusion really ties into all of these enterprise and modern apps and really, again, providing that workload, performance and scalability.

That can either be through QOS or through SLAs like we just talked about, but it also provides simplified virtualization support. One of the things we know is that customers are, are very looking at exploring other virtualization alternatives. And so whether it's VMware or HyperV or OpenStack or all of the open source ones, or even with Nutanix, it's a single platform that can work across all the platforms without needing to re-architect rebuy or handle everything.

But it's also providing those customers that built-in provisioning automation and disaster recovery. And the key point here with everything we're doing with Port Works, it's Kubernetes aware storage. And so we really have the ability to scale, balance provision and optimize all of that storage for those strategic workloads.

And so when we kind of think about this vision of your enterprise data cloud, we're now pooling together all of these resources and Brent will talk about what we're doing there. We're virtualizing that thought of data, sharing that capacity, sharing that performance, sharing that features. It's gonna be a consistent environment across all workloads that has built in cyber resilience, data governance, and it's truly software defined and globally scalable.