49:32 Webinar
Microsoft Azure: Get All-flash Storage for Azure Local
Join our webinar to learn how easy it is to deploy Everpure FlashArray™ with Azure Local and how it can completely transform your virtualization journey.
This webinar first aired on May 21, 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
Hey, everybody. Welcome to the May Tech Talk for Microsoft Azure. We're gonna get into what we're doing with Azure Local, with the EverPure FlashArray. My name is David Staman. I'm Technical Strategy Director for Cloud and Virtualization, and I'm
joined here today with Kiel. this is Robert Quimby, and I'm the Consulting Field Solutions Architect focusing on Microsoft On-Premises. Thanks, David. Awesome. So as we kinda think about the environment, there's a lot of platform decisions that are
really driving change. Some of these are increased licensing costs with everything kind of going on with the Broadcom acquisition. There's a lot of bundling choices that kind of limit those choices, especially when we come to some of the challenges with supply chain, and there's also some of this
architectural complexity. And as we kinda sit within this virtualization space, there's a lot of things that we really have to think about choices. at Pure, we've really taken a b- a lot about what we've done with our VMware platform, and we're now looking at how we can expand this to other platforms.
And so with the announcement of GA for Azure Local last week, we're very excited to add another notch, within our belt. And so why Azure Local? Well, Azure Local is a very good infrastructure for customers who wanna be able to run things at the edge.
They wanna have data sovereignty. They wanna be able to run VMs and containers side by side. And the best part about this is it allows customers to consume their Azure services locally. These are things like Azure Virtual Desktop, or we'll call it ADD later, or Azure
Kubernetes Service or AKS. and why it's important is it really allows customers to easy to use an easy and familiar tool set. they might already be using Windows. They may already be using HyperV. And Azure Local takes those and builds upon those tool sets with a very familiar Azure
control plane. And also, it can be managed both in a disconnected, which is in preview, or a connected operation. And what that means is that everything will be managed through that Azure portal. So as customers are building out these hybrid cloud environments, they truly have, that experience.
But one of the things that's really hindered folks for a while is the fact that there was never support for external storage. It was only in a hyper-converged format. You could only use the local storage. However, with a lot of the updated use cases around local AI inferencing or even your
mission-critical applications or even data sovereignty, a lot of the requirements for data at the edge or even within your local sites is growing and growing. And so the ability for EverPure to be able to deliver storage there, is really gonna help. And so when we really think about what are some of those workloads that are going to be beneficial, you might wonder, "Well, why do I actually need a FlashArray to be able to
deliver this?" Kiel's gonna be coming into some of the benefits of what we're doing from a FlashArray perspective, but think about the proven reliability for your tier zero applications. Our platform has a lot of core features that tie in and build upon our enterplat- enterprise-class data security as well as kinda tying into that data sovereignty with
our, multi-tenancy. We have predictable high performance at any scale across multiple protocols. but we also have the ability to offload a lot of the data services to our storage. Get, get those off your Azure Local, your HyperV nodes, and allow you to be able to run those for your virtual machines.
One of the key points is our Evergreen story. No forklift or disruptive upgrades ever again. Non-disruptive software, non-disruptive hardware. and really with everything going on, it helps reduce cooling power and rack space. but let's just not talk about the reduction of node count.
We all know the challenges with supply chain. If we have the ability to consolidate and do more with less, that's gonna make all of our jobs a lot easier, and that ties into that disaggregation of flexibility of really allowing you as a customer and/or maybe a partner, to really have a choice of how you wanna handle these deployments.
because of the capabilities, it also allows you to reuse your existing infrastructure, and this could be, well, maybe you're an existing FlashArray customer and you're running a different hypervisor and you wanna have a multi-hypervisor strategy. You can reuse your existing investment without having to really make any big changes. If you're currently using your servers for something else, if as long as they're gonna be
on the hardware compatibility list, and have supported fiber channel HBAs, you can reuse those. And again, the benefits of what's coming, both today and tomorrow and around the connected and disconnected parity, is really gonna drive a lot of the capabilities that Azure Local can deliver.
And so what we're announcing or what was announced, I think it was back, on the 24th of April, is the ability to attach FlashArray and other external storage to Azure Local. Azure's gonna have two fundamental versions of what they call this. One is a hyper-converged deployment, and that is a traditional HCI deployment. with that deployment, you can have both HCI as well as SAN together.
the HCI utilizes Storage Spaces Direct. with a disaggregated deployment, that is where you can have a SAN-only deployment where you don't actually need any of those individual drives within the servers. both of this support came in with release 2604. And why this is important is the first revision of this support will be fiber channel.
iSCSI and NVMe will likely follow. but what this allows customers to do is provision SAN volumes for their user workload. So whether that's for virtual desktops, whether that's for their Kubernetes, whether it's for their traditional VMs, it allows them to be able to get all of the benefits of what Pure can deliver, which in case, right, high performance for your databases, storage and
performance for your AI, and massive VM consolidation.And so now I'm gonna pass it off to Q to kind of talk about what we're doing from a FlashArray perspective. And a couple of years ago, we really started to lay solid plumbing on all the features that I'm t- gonna talk about on this enterprise data cloud, and it isn't FlashArray specific.
It's really Purity specific, which involves everything we're doing in the cloud, FlashArray, and FlashBlade. And so what is this enterprise data cloud? It's, it's this ability to unify all of your data, across, you know, on-premises hybrid in the cloud, giving you this intelligent control plane that makes management super simple,
makes provisioning super simple, government simple, and it can be delivered as a service. the rate of change in the industry is so fast that if you have the tools and the agility to, make a change because the business environment has changed for your company, we no longer, as an industry, can have a focus on provisioning five, six years in advance and sizing and purchasing for that.
Pure's philosophy the entire time, that I've been here, and at least for a decade, is buy what you need. It's super simple to grow and to, dynamically expand without having to turn a thousand knobs. super simple. And so we're investing, top in the industry on R&D, and, a lot of the, the, the benefits of those investments I'll talk about in the next
couple of slides. So one of the big changes is the ability to abstract individual pieces of hardware, whether they're virtual in the cloud or, or physical, into a fleet. And what a f- a fleet is going to give you is this intelligent control plane where you can manage it as one.
You're able to look for, with workloads and presets, simplification of your deployment. So if there's five or six things that you do, after you, provision a LUN or provision a file system, things like I need to protect it, I need to put a policy on it, I need to have quality of service on it, I wanna replicate it, et cetera, all of that can be a preset, that
you can apply to particular workloads so that they are done consistently and can be, audited, and checked, and you can be alerted if there's configuration drift. So this, fleet management, which we have called Fusion, it enables our systems to have this distributed access to each other. there, there isn't like a gateway and complicated networking involved.
It's unique, and it's built into Purity. Almost everything, that we talk about is built into Purity at a very deep level. It's not a bolt-on, an add-on, and it's a fundamental part of our operating system. this is shipping today. Every month, we're delivering additional features for it, being able
to do things like discovery. I know that the serial number is this on this w- on this Azure local system. which FlashArray is it on? I have a fleet of 30 FlashArrays. Oh, it's on FlashArray 37.
and you can quickly go and, and solve problems. So as I mentioned, you know, th- this, this is fundamentally built into it. Everything from our, unified, file and object to our unified block, everything is, an equal participant in the Purity operating system. everything is built-in.
Things, all-- Everything on the slide, everything from our data reduction, you know, we are leading in the industry in data reduction, not only on, you know, what gets, written, to NAND, but our replication, et cetera. We really focus on a true non-disruptive upgrade. Everybody advertises it.
we have customers, that you will have heard of that, it's mission-critical operations, and they are performing not only software upgrades, but full-on non-disruptive upgrades of controllers to new controller models during the business day. I'll talk a little bit more about some of these features, but all this is just built in. HA is super simple.
You're not typing licensing at a CLI and having to independently license hundreds of different features. Everything's in the box. You have the agility, and the options to engage with the services that you need on-- for the workloads, that you need them for. we really, really work hard on ensuring that it's simple, right?
There-- You're not going to every single, disk device. And in fact, we'll talk about our DirectFlash modules, and creating RAID groups and having silos upon silos within the storage area network. You have one pool. it's hyperconverged simplicity at the SAN level.
all of our features are on by default, everything from, data-at-rest encryption to, all of our security features. you can literally rack and stack and power on and have the array servicing data, in an hour, hour and a half. safe mode, which is something that we'll talk about a little bit more, it's like, let's take it, things that are potentially immutable,
like snapshots, and make them indelible as well, and we'll talk a little bit more about that. The fact that there's no power switch, there's no, guide on safely shutting off the FlashArray, you just pull the plug. no planned downtime, significant number of nines.
The last I looked, it was either six or seven nines i-i-in our fleet of all of our FlashArrays worldwide that were not in a d- in a dark site. pretty tremendous. And all of this, the, the, this, the, these non-disruptive upgrades, allow us to provide these guarantees, within our evergreen subscription op-, op-offerings.
And these have everything from 100% OpEx model to 100% CapEx model and, and a few in between.This is our direct flash module. Y- y- you notice it is a little bit different than an SSD. It's a little bit longer. there's some capacitors there so that, if power is killed, everything will get written to NAND.
they're, they're large, right? They're 150 terabytes, 300 terabytes, either shipped in the last week or in the next couple of weeks, with 600 terabytes on the roadmap. These are large devices that do not have a flash transition layer. They are basically exposing the NAND cells directly to the controllers, and the
controller acts like an SSD. And if you're familiar with NAND, you know that the act of writing, you can't change a block in place, right? You can't change a cell. You have to wipe an entire cell that might be three, three-layer, four-layer, and write the
whole thing. So if you have a bunch of cells that on, on three le- layer, NAND for instance, where two of them are dereferenced, you don't delete them. You wipe the whole cell, which means you have to do this garbage collection, and that action is writing to the NAND, which is hurting the NAND, right? That writing is, slowly killing the NAND.
And so if you have dozens or hundreds of SSD devices in your array, each one of those autonomous units have to do their own garbage collection, and the smaller your container, the more often you're gonna have to do it, which is why we can really say that, you know, it's almost an order of magnitude more reliable. It's like the hop from hard drives to, to, to flash, the, the hop from flash to our direct
flash modules. They're longer lasting. They're denser. it's really an amazing solution. So snapshots are one of my favorite features.
the fact that both a snapshot and a volume is just metadata on the flash array, everything is thin provisioned. it takes microseconds to take a snapshot, regardless of the data set size, right? You Whether we're talking 10 terabytes or multiple petabyte, volume sizes. A snapshot's immutable.
it can be a source, right? So you can mount it by cloning, and copying that snapshot to a new volume or overwriting an existing volume. We have customers that utilize this for test and dev, and they can ad hoc or through storage as code refresh their test and dev as often as they want, nearly instantaneous and
not consuming additional space. All these clones of your production volumes only consume space a- as your test and dev environment is writing a ton of unique data in those environments. And if you're constantly overwriting and refreshing test and dev, it, it, it's pretty amazing.
our snapshots are also the granular block that can be utilized for replication, asynchronously. And, here a little bit more on snapshots, right? The fact that you can and this is focusing on volumes. We can also do snapshots on the file system, file systems as well, with our, with our, with
our, SMB and, NFS file shares. the fact that it's super simple and fast, the fact that you can not consume additional space with those snapshots, that's actually a, a better solution than utilizing application layer cloning and replication where it's having to be reingested because the, Purity operating system has to, look at, at the incoming write stream and try
to optimize and data reduce it, whereas a snapshot and metadata, we're just exposing, additional objects that have references to NAND, right? those snapshots can be replicated to the cloud, to our FlashBlade, to another flash array, to our virtual flash array in the cloud. it is an amazing, solution.
And here to dive a little bit more into safe mode. really cool, feature that saved my bacon a few times. if, if you've ever been involved with enterprise storage and you've started to delete something, and your finger's moving to click it on the mouse, and your brain is screaming, "Stop," and you hit the button, and it's gone, and it's like you can't recover it.
built into Purity since I can remember, there's always been a concept of a recycle bin, a, a de- a destroyed bin, a destroyed bucket. So you first de- just to delete and destroy something, and then it's in that bucket, and by default for 24 hours. I already mentioned that the snapshots are immutable, which means you couldn't mount a
snapshot, change something that does, and unmount it, right? You know that that snapshot is exactly what existed at the point in time of that snapshot. So what if you also could make that snapshot that you destroyed, something that you could not eradicate? So by default, you could go in, you could eradicate it, now it's gone.
Now you've really deleted it. But you can now set And it's not just snapshots. It could be f- both file systems and volumes themselves. So you delete an entire volume or a file system, and it's sitting in that destroyed bucket.
and so now instead of it being 24 hours, what, you can change that date up to, you know, a month, 30 days, and you prevent manual eradication. And so because of that, we've had a lot of customers who have been ransomwared and been able to recover everything.Um, and these bad actors have your administrative credentials. They've managed to get in and get your domain admin credentials, or even if you've siloed,
Purity off and it had local, creds, they've gotten those creds, and they've gone, and they've deleted all of your snapshots and all of your volumes. features built in, no, no charge. and, it's just you're going to just have to calculate a tiny bit of disk space, right? I'm gonna have captured in snapshots X number of days in that destroyed bucket.
And if you ever get into a situation, "Hey, I'm migrating from this platform to this platform, and it's gonna get reingested into this different hypervisor," there's a way for you to eradicate that, right? It's not only multi-factor auth but multi-person auth. and those policies, can be, enforced so that you can,
you know, work with peer support and make sure that you are protected. Awesome. Thanks, Q. I mean, I think the really important is we wanted to, like, make sure that we outlined really all the benefits of what the FlashArray can deliver. Because when we think about the solution is customers might be familiar with FlashArray,
or they might be familiar with Azure Local, and we wanted to make sure that we had the capabilities to kind of talk about what are the benefits of both, but then what can we do that actually ties in everything together. And so the solution, if you're very familiar with Hyper-V, or Windows, is very basic, is from a storage perspective, from a peer side, is you can provision a volume of any size.
You create a host, you create the volume, you connect them together. you'll need to have fiber channel HBAs, as we mentioned. But from a some familiar workspace of what you might be already interested in doing is you just create a cluster and share bond. So you're already gonna have a failover cluster.
You create an NTFma- NTFS-formatted disk, you add it. Once you have that disk, you have a path. And so what Azure Local enables you to do is create what is called a storage path, and that is kind of that storage container, and you specify that local storage path. And so from a perspective of how you configure external storage, it really
is that basic simple. Q's gonna cover in the next slide kind of how we're gonna do this from an array management. But we've already scripted the entire deployment for you of creating hosts, creating volumes, connecting them together, formatting, doing all of that work. It's very basic. From a virtual machine operation, there's a
couple ways that you actually have the ability to consume our storage. So if you're using Azure Virtual Desktop, right now it round-robins between your storage paths. So if you're in a hyper-converged cluster, it might use both your storage spaces direct as well as your external storage.
If you're not disaggregated format, then it's only gonna use your external storage for that configuration. If you're using Azure virtual machines or Azure local virtual machines, when you provision those, again, you can have it automatically be placed on a storage path for availability, or you can pin it down to an individual one.
But also, you have the ability that when you are using containers and you're using AKS, there's really two or three ways that you have the ability to consume your storage, and we'll cover this again in the-- later. but the first one is again, deploying it just as you would. The second one is gonna be using the default storage class that comes in.
And the benefit of that today is that when you provision your storage us- utilizing that Kubernetes, I think it's ACE, HCI, AKS, CSI provisioner, it actually will utilize fiber channel for that provisioning, which is again gonna be high speed, low latency connectivity. We also are working on validation for our PX CSI and our Portworx Enterprise. However, when you utilize those services, all of the traffic will come out of the user VMs,
not out of the actual node. So right now, kind of our recommendation is create those storage classes and utilize those persistent volumes. In PowerShell, we have a couple different modules. we have thousands and thousands of FlashArrays globally, where our customers save time by
automating things with our PowerShell SDK. We have commandletized, hundreds of our API calls, where you can add that into your different workflows and, everything from super simple, I need to add a user, I need to provision something, to very complicated, use cases where we have customers ref-refreshing test in dev or orchestrating an active DR failover
or failback, right? And so, we've done a really good job on this, module where you can install it, install module, you know, peer storage, PowerShell SDK two, that gets deployed. And any of the commandlets, you can do a get help on, and, that has been, human authored and is, is very good.
we have a ton of stuff on GitHub. If you go to our open connect on GitHub, a ton of automation that a lot of our customers use as a template and leverage that to build, what is unique and saving time in their environment. one of the unique things about the SDK two, though, is that you need-- it's, it's a pure API at the back end, so you have to know what it's called, right?
So if you have a workflow and you're going to, you know, initiate a snapshot, you have to know the name of the volume as named, the friendly name on the FlashArray, et cetera. so one of the things that we did to help our Windows administrators that a lot of Hyper-V administrators utilize is our backup, SDK. And the idea there is you're setting a, a set of, volumes, one or more,
that are in a volume set. And how you're declaring those volumes is v- via the, drive letter or mount point in Windows, right? And so in an Azure local environment, you're going to know the cluster shared volume, it's volume one, volume two, volume three, or whatever you've renamed those-You know, see
cluster storage mount points to. You know what they are. Add one or more of them to a volume set, and then you can initiate a snapshot. These snapshots are crash consistent and can be replicated, right? If the volumes that are in the volume set are part, part of a protection group, which is a
construct in the flash array where we're guaranteeing consistency between all the volumes in that protection group, and that a protection group snapshot occurs instead, protection groups can have targets, right? So you could say, "Hey, I want a long-term seven-year retention. I'm gonna replicate this to Azure Blob Storage directly from the flash array, and I'm gonna
initiate it as, storage as code automation that runs periodically, a-against the flash array," and y-you're done. And since we are tagging the snapshots and volumes with the information on what the volume set is called, et cetera, if the primary flash array that you have done this on is unavailable and you've replicated it to another flash array, you can
enumerate, "Give me all the snapshots for volume set SQL 37," and it will show you the snapshots. You can take any one of those snapshots and mount it N number of times in your environment at the DR site or at the other site. And it doesn't matter who set this job up that initially created the snapshots.
It doesn't matter, maybe that person's no longer with the company or out on vacation. Any human at the company can enumerate the source or any of the targets of those snapshots and instantly get at it and instantly solve your problem. Awesome. So let's go into the architecture a little bit. We'll kind of cover the hyperconverged and the disaggregated, but kind of from a
configuration point of view is if you have an existing Azure local cluster, one at a time you'll be able to put those kind of in maintenance, and add in HBAs and zone those for fiber channel. from a hyperconverged right now, there's one to 16 nodes supported, so if you're kind of thinking about scale, take that into consideration.
and this is gonna be a cluster that has an existing storage spaces direct for that clustered shared volume. And so the setup, again, very straightforward after you add the HBAs. You create a host, you create a host group, you create the volume, connect them together, initialize and mount that clustered shared volume, and then you create that storage path.
And then the consumption is, again, specifying the storage path, configuring a storage class, or again, utilizing Azure Virtual Desktop. Key point for this is that customers have choice. If they have an existing S-HCI cluster, they can just add on external storage. They don't have to worry about reformatting everything.
However, we do give customers choice to be able to have a second option, which is brand new. And being able to deploy with disaggregated is exciting because, slightly different networking, right? You're not gonna need dedicated, NICs for the HCI storage replication.
but other than that, it's unlocking a few things. First, without S2D, you no longer are limited d- to a 16-node maximum node count in your cluster. It can go to the full 64-node count that, Windows supports, right? And Azure Local is built on Windows Server 2025.
another thing, that is unique is if you think about deploying your standard HCI Azure local cluster, it needs four devices for S2D. You can have that, minimum. You can have that to two if you deploy in an infrastructure-only mode. If you were to provision a new Azure local fleet, let's say 100 servers, and you were to
put four SSDs in those 100 servers, and the only thing you intended to run on that was the infrastructure that, you needed, you know, things like the Azure Arc Resource Bridge VM. In today's environment, it's not unlikely that this could cost from $5,000 to $10,000 per server, right? And if you had no intent to run, user workloads, VMs, VDI, or containers, and you intended, intended to
connect to, external storage, the flash array, this can save you a tremendous amount of money. We've had a lot of customers who have been stuck with lead times on components for servers, particularly SSDs. And so this is a huge, value to our customers who intend to deploy on external storage.
Reduce your storage costs, take a hard look at, you know, blades that only have boot disks that were always excluded from Azure local. And looking at larger cluster sizes, we have a lot of customers in the, the 30 to 40 cluster size coming from VMware and Broadcom and looking really hard at Azure local. and now that they can not have to double, triple, quadruple the number of clusters to
manage are very excited and moving fast. Awesome. So what does it look like to migrate this to the solution? So there's really kind of two fundamental ways. one is gonna be the traditional approach that customers should be very familiar with, and that's using Azure's own Azure Migrate tool.
What that will do is allow customers to migrate from either Hyper-V to Azure local or from vSphere to Azure local, utilizing a very familiar tool set to allow you to kind of stage those changes over, have a planned migration, handle REIPing, remove some tools, handle the configuration, register with Arc. and the best part about this, again, is it works with any flash array
within our portfolio. some of the stuff we're doing on is kind of, as Key was mentioning, is around the, the capabilities of our automation. Is that what happens if both your source and destination are a flash array? You may not even need to use move, or move the VMs, use Azure Migrate.
In this case, what you actually can do is maybe utilize our snapshot, maybe utilize our replication, and we can actually streamline and automate that. So right now we're working on some of those solutioning to see what can we do to streamline those operations and allow you to migrate and get back to business much quicker.And I mentioned a ton of the replication technologies
built into the FlashArray. one of the options that you have for storage replication is to use an Azure native tool. and it's in preview, the, the disaster recovery blade, which is, once you've clicked on the Azure local cluster itself. And what that is utilizing under the covers is ASR, Azure, site recovery to migrate your
replicate your virtual machine into Azure. And then that machine could then be, replicated back to a different Azure local cluster. but some of the things that Purity exposes is the ability to replicate, you know, asynchronously or continuously, with our active DR,
between FlashArrays, and then you can light up the, DR side and then, import those virtual machines, and with the hydration feature, have it all register up into Azure. one of the things that, that, that David worked really hard on was getting, our uniform active cluster support. This is where you have your Azure local cluster.
Both, FlashArrays are connected to every node in that cluster, and if there were a storage and event, let's say that the rack that FlashArray one is in loses power completely in your data center, it's an NPIO event to the Windows host in Azure local, and there's no disruption. You, instantly have, all of your VMs and, and, and workloads
continuing to run because the hypervisor host, instantly utilizes the othe the other NPIO paths connecting to the live, and remaining FlashArray. Yeah. And one thing to note too is this is between any cluster to any cluster, so it can go between a disaggregated and a hyperconverged, a hyperconverged to disaggregated, two disaggregated to two hyperconverged, right?
that's really gonna be up to you on, on how you wanna use this solution. and then just obviously, we talked about DR, let's talk about backup. all the backup vendors that you're probably using today with Azure local or Windows or Hyper-VR are gonna be supported. So Cohesity, Veeam, Commvault, and Rubrik were kind of some of the four that have been
validated by Microsoft. but again, that's why we have them up here. we're also working on additional solutions validation from an EverPeer perspective, and then also looking at what we can do, from an integration, right? Maybe we can do some of that snapshot offload and continue to streamline your safe mode
snapshots with your backups, and again, orchestrate that full disaster recovery. And so some of the important points here, I mean, clearly we have a bunch of on-premises applications that would run in your virtual desktops, your containers, and your virtual machines. I've worked with customers, particularly in the VDI space, very
interested on on-premises VDI. They love their experience with AVD in the cloud. they're wanting to migrate their on-premises off of, other platforms to Azure virtual desktop. And one of the cool core technologies in AVD is a, is a feature called FSLogic where you
can map data to keep it persistent, right? And so at boot time, if you have non-persistent desktops but you want your home directories, you want the doctor to have access to his hundred gigabyte OST, that's tied to his Microsoft three sixty-five account. So what the, the preferred protocol there that I've seen in the field is that you would host
those virtual hard disk files on SMB directly, right? And so we have a lot of customers that are excited, a couple of years ago when we shipped continuous availability in our S- SMB stack. And so now instead of having to deploy a two or three-node bare metal Windows server cluster to give you highly available file services, you can just, create a
file system directly on the FlashArray, expose SMB, and host all of AVD's, persistent data through FSLogic live on the FlashArray, and that is amazing. And that leads into the fact that the FlashArray is not just block. Our unified storage can give you file and object, which is pretty exciting. you can even configure the file piece of it to be multi-protocol against the f- the, the
exact same file system. So if you have cust- customer, environments that need both, NFS and SMB, you can certainly do that. we are very strong on the ability to persist our array, for a long period of time.
We have customers where the FlashArray has been in production for a dozen years, and there's not a single piece of physical matter in that array that's original, right? We've non-disruptively upgraded everything from the chassis, and our chassis typically have a ten-plus year lifespan, to the power supplies, the NVRAM, the, the direct flash modules, everything.
The controllers, think of the, the controllers on the back can be non-disruptively upgraded. Everything, from a protocol perspective, whether it's Ethernet and MAC addresses and fiber channel and worldwide names, is all virtual inside of our controllers, so that when you switch the controllers and you've now upgraded the, the, the ports are bigger and faster and, and they technically at a hardware level have different device IDs, but what's
exposed to the OSes i- is the same. So now you don't have to go and fix anything if you've, you know, locked things down to the MAC address or changed your fiber channel zoning. You just non-disruptively update, the controllers, and you're ready to go. the ability to grow as you need it-Seriously, do not buy five, 10 years' worth of flash
today and have it sit 80% unutilized for four years, right? it's very trivial. You can do it at a, at a two direct flash module at a time, dynamically expand, just throw in two more devices. very, We're very conscious of, customers getting the most value out of our platform
and to not have assets sitting there underutilized. Awesome. So as you kind of heard, right, there's really a lot of things that really just unlock scale. the benefits of FlashArray really help generate that performance and efficiency and really this overall, the solution with Azure Local and FlashArray, we're very excited about.
And so let's kind of take a look into some next actions. If you're interested in learning more, reach out to your Pure account team or your partner and have a technical deep dive conversation. We have architecture and use case workshops. We have migration readiness assessments.
We're here to help you get to that next step. But also don't think- worry about, we'll be available in Las Vegas June 16th through 18th at our annual Accelerate Conference. scan the QR code to register. We'll have demo fest, we'll have community meetups, we'll have a bunch of breakouts,
we'll have our keynotes. All A lot of our specialists will be there. We'll be able to work with you one-on-one and understand what are your requirements and what can we get, how can we get you to those next steps and adopt these new solutions. Also, don't forget about the EverPure community. we have a digital community available.
Scan that QR code, purecommunity.purestorage.com. Again, to be renamed with our domain name here. But there's forums for solutions, f- forums for our products. There's our Pure User Group. So get engaged with not only us, but also with your peers as well.
and we really wanna thank you for joining us. if we hadn't gotten to your questions that you've asked in the chat, we'll stay on for a bit afterwards and, and get those answered. But, really appreciate you, staying on for the call with us, and hopefully you learned a little bit, so thank you very much.
Thank you very much. Awesome. And now as a couple of folks has, have suspected, we did pre-record this, because David is in, Europe right now, on a business meeting. So, happy, May 21st. I've got a couple of questions, that I wanna answer live, and then if anybody puts anything
into the Q&A, I'll try to watch the chat as well. one of the questions that, has come up is, If you have a FlashArray and you're running other workloads, VMware, you know, bare-metal Kubernetes, Oracle, et cetera, can I share that with Azure Local? And you certainly can.
So most of the FlashArrays come with both, Ethernet and Fiber Channel ports. and you can, coexist with multiple different workloads, multiple different Azure local clusters, et cetera, no problem. another question that's come in is, you know, what's the migration process look like, to Azure Local? And, you know, how much
downtime should I expect? And one of the things that Microsoft has worked on for many years to get, workloads into Azure Cloud is the Azure Migrate tool. And if you've ever used it, it's kind of similar to kind of hypervisor movement, you know, live migration vMotion in that it's doing the bulk of the copy,
and then it has to cut over, and that cut over typically is a minute or two, a couple of minutes, plus the boot time of booting the virtual machine, in Azure Local once you've migrated all the data. so it's, it's, it's pretty fast. and that is something that has been supported for, VMR to Azure Local
for a while now, and it's been supported for, VMR to Azure Local connected to external storage since, the 24th of April when Microsoft announced it. There's a question from, Patrick about iSCSI support, and that is definitely true. So in, this month's drop of Azure Local, every Azure Local drop is the year and the month. So 2604 is technically the build of Azure Local you must deploy if you want to deploy,
to, SAN only or disaggregated. 2605, there's gonna be a preview for iSCSI, so that should drop in the next week. and then July is the scheduled GA for that, but I can't promise, Microsoft's, schedule. I'm not, I'm not a Microsoft PM anymore.
couple other questions that have come up is, how does granular, you know, scaling work with the storage and practice, right? Your, your database needs a little capacity. do you buy or license more compute nodes? How, how, how do you do this?
if you are connected to EverPure, and you have three options, right? You have an existing Azure Local cluster that has been running for years. You can take each node down in the maintenance mode and add a Fiber Channel card, soon an Ethernet card, for iSCSI. And, now you can connect and have both, right?
You've got both storage paths for the hyper-converged storage and a storage path for Pure. When you're in the Azure portal and you're deploying a new VM, you could say automatically place it wherever, it'll round robin to all your storage paths. With a virtual machine, you could say, "No.
Put it on, you know, the, the two petabyte Pure, FlashArray storage path." other things you can do is you're running out of space on your cluster shared volume. So grow it. Just change it in Purity. three seconds later, the hypervisor sees it, is larger, and then you can extend the size of
the partition and you're rolling. Uh-On almost all hypervisors, including Hyper-V, which is the hypervisor itself underneath Azure Local, you can change the size of the VHDX file, dynamically, without taking downtime, so that's no problem. When it comes to the flash array itself, you can add DFMs to your chassis, in a small of an
in, granular, The granularity of two. So you put in two more DFMs, now the pool's bigger. So if you're running out of capacity, you can add another shelf of, potential capacity to a flash array, and so we call that a direct flash shelf. And if you've filled up, you know, the maximum number of shelves or the maximum number of
DFMs that this particular flash array model can handle, you can non-disruptively upgrade the controllers to either a new revision of the controller or a new model of the controller. I recently, upgraded some old, FlashArrayX50R2s to FlashArrayX90R4s in one move. They went from 50 to 90s.
it took a little longer, 'cause there was a lot more firmware to, to deal with. Like, normally it's, like, five minutes to upgrade the firmware, and it was about 30 minutes. But, 'cause normally you're, you're going one generation or you're going, you know, one, one model bigger. I was two gen and two
model all at the same time. so you can definitely scale every part of this non-disruptively, when Ever peer is involved. I've got two more questions, and most importantly if, if, if you've got, some live ones, throw them in the Q&A. so if you experience an outage, right, who, who owns the support ticket?
Do you call Microsoft? Do you call Everpeer? And this is one of the reasons why, Microsoft announced the preview, the public preview at Ignite last November, and it took all the way to April. Half of it was they added some more telemetry, and they were able to pull in disaggregated
SAN only, right? I was If you'dda asked me in March, I was like, "Boy, I hope we get that by Ignite, by November," right? and we worked really hard on that. Because the disaggregated, and I don't know if we really concentrated it, but it, it gives you two things. One, save money, right?
If you're gonna buy SSDs today, they're expensive. They're double or triple. If you, if you look just for home, consumer models, they're double or triple from just last fall, right? And so, if you're just gonna run the Azure Arc Resource Bridge VM, the infrastructure stuff,
save money. You could save two, five grand per node by not having that local disk. The other thing is that if you don't have STV involved, the HCI storage, that means you're not You're no longer bound to a 16-node maximum, right? Storage spaces direct doesn't allow you to grow the cluster to more than 16 nodes.
And if you look in the field, I've never seen a Azure Local cluster, formerly Azure Stack HCI, larger than eight nodes. and I, and I saw numbers from Microsoft, like, last fall, where it was, like, something like 98% of all Azure Local in existence is eight nodes or less. So you effectively have eight nodes maximum.
And if you go disaggregated with SAN only, you now are unblocked. So what is Microsoft's Windows maximum for cluster size? 64 nodes. Does that mean I see 64 nodes? No. But I see 30, I see 35 nodes. I see customers coming from VMware and are looking at Microsoft HCI and recommendations
for, you know, the campus cluster and four nodes, et cetera, and it's like, I've got to increase the number of clusters 500%? And that was a blocker for them. And so for them being able to go and deploy 30-node Azure Local clusters, you can do that now with, with SAN only.
and my last question, and then when I'm done with that I'll give it a minute or two to see if anybody else has a live question, and then we'll end the webinar, and I appreciate everybody's time. this question i- Oh, I didn't really finish this question, which was, if, if we experience an outage. And, and that is, you know, if, if, if there's
something wrong with the storage, call peer immediately, right? the cluster shared volume is offline, something's wrong with the, with the flash array. We're g- we're gonna help solve that problem. If you're not sure, if something's wrong with Azure Local or Azure Local, the nodes can't
communicate with the Azure, Azure portal or Azure Cloud, definitely call Microsoft. this isn't gonna be a, a finger-pointing event. "Oh, we think it's a Microsoft problem," or, "We think it's an Everpeer problem," or, "We think it's a problem with your fiber channel zoning." we have a back-end ticketing system so that we can pass tickets between, each other.
and even if you have a, a server bender that is not Does not view Everpeer in a positive light, right, they'd rather sell their storage, there's defined lines that Microsoft has, is, is built into the support matrix so that, you will get taken care of. the raffle. Yeah, Laura's gonna probably do that in a second.
Let me just, answer this last question from Peter, which is, "Are there any XenServer migration tools?" That is a good question. I don't know off the top of my head. I know that a lot of folks at Everpeer are working with, like, XCP and G, and things like that.
But Azure Migrate, I don't Azure Migrate is the tool to migrate, and it is Microsoft's recommended way of doing it. So you can go to the Azure Migrate page and find out all of the source hypervisors they support. The hydration feature, which is imminent to go live, that's what you're going to want with
all of Purity replication tools. That's what you're gonna want if you use one of the vendors for third-party backup and you've restored it to another Azure Local cluster and you wanna hydrate all that information into the Azure portal. so if you have any tool that will migrate you from XenServer to Hyper-V or a failover
cluster, use it, as soon as the hydration tool drops GA or opt in to the private preview. And then now you've got all your VMs on Hyper-V, on the Azure Local cluster, and then click the Hydrate button, right? that's what I would do. Oh, so Laura just mentioned that a winner was
selected, a- above and sent in the chat. Louis, congratulations. and I'm gonna be quiet for 60 seconds in case there's a last question. And if not, I thank everybody for their attention, and I'm excited about the potential, with Microsoft and Azure Local with Everpeer.
Robert Quimbey
Consulting Field Solutions Architect, Everpure
David Stamen
Technical Strategy Director, Everpure
Azure Local now supports on-premises storage, and Everpure is a certified solution. It offers a new alternative to legacy virtualization platforms, using familiar Windows and Hyper-V technology.
Why add local storage to Azure?
- Delivers data sovereignty as all data remains on premises
- Ultra-low latency storage for your most demanding workloads, such as databases and AI
- Separate compute from storage, to allow granular scaling and avoid over-spending
Join our webinar to learn how easy it is to deploy Everpure FlashArray™ with Azure Local and how it can completely transform your virtualization journey.
We Also Recommend...
Personalize for Me
My Updates
No updates at this time.