Dismiss
Innovation
A platform built for AI
Unified, automated, and ready to turn data into intelligence.
Dismiss
June 16-18, Las Vegas
Pure//Accelerate® 2026
Discover how to unlock the true value of your data.
Dismiss
NVIDIA GTC San Jose 2026
Experience the Everpure difference at GTC
March 16-19 | Booth #935
San Jose McEnery Convention Center
33:25 Webinar
The innovations powering the Enterprise Data Cloud
Prakash Darji, General Manager, DX and CTO Rob Lee unveil new innovations that power the Enterprise Data Cloud.
This webinar first aired on September 26, 2025
Click to View Transcript
so I want to thank you for your time and invite the next speaker out we have Prakash Dargie be speaking to you more about how this works thank you thank you Charlie
so let's get into it we talked about this intelligent control plane now what is this intelligent control plane if you think about it the world historically has had observability and automation
we know that we collect data we do things with that data but they've been disconnected so we're uniquely taking a different approach where instead of worrying about this fragmentation
that exists with your silos where you have different observability and different automation stacks where you have operational complexity where you know
you have disparity in how you do your workflows you don't get to the place where you can actually solve for threats and regulation because of the sprawl you need to bring this together
and this is what the intelligent control plane is about now we didn't start this just yesterday it's not new with just the enterprise data cloud we introduced pure one over 10 years ago now and it's been collecting data from all of the arrays
we collect now on an annual basis over 70 petabytes of telemetry coming off the arrays we now know how our arrays are used we have a lot of data to say how you should optimize your landscape
but if you wanted to do something with it that used to be something that you'd have to go in and discover now you have the ability with the intelligent control plane to take action
with fusion as Charlie mentioned we've networked all of the arrays so you as you think about what networking the erase means it means that oh I'm this signal do this thing
seems like a fairly simple concept I don't know why it wasn't done before but it took a lot of effort to get to this point where you can link together observability and automation
to get to this place so you don't need fragmented silos of utilization you can efficiently utilize all of your assets and the power of the intelligent control planes interesting
we announced this in Vegas with this concept of presets what is a preset it's a set of policy and configuration how you do your replication your snapshot schedules what type of workload it is what type of storage class it is
we've also announced our workflow orchestration capability that allows you to schedule different work workloads and workflows across your entire stack not just tied to storage or Pure Storage's ecosystem but the broad ecosystem
we've introduced an AI co pilot so now that observability is now smarter and can recommend what you need to do you can interact in natural language and on top of that we've created
a built in platform with cyber resilience and trust which Rob is gonna expand on later and now this platform is getting better this is what we did six months ago and six months later we're building on this
now one of the things that you all may kind of have heard about is we've had some customers talking to us about Vmware um and historically we've taken a look at this problem and we're like okay
you have options to move but it's kind of hard you know um damn near impossible if you actually think about it because when you want to migrate from Vmware to Kubernetes
what are the things you have to do well you probably have to get a bunch of people together across your VM admins and database admins and your storage admins
and people who handle your network you need to actually get the data of what configuration exists in each one of those landscape um all of the technologies
disparate with observability stacks and that's for one VM for companies that have a ton like the operational complexity to get off this is actually fairly challenging
but that was before the intelligent control plane because and this is actually evolved when we started with this six months ago I would have been like yeah you're probably stuck but what if you could discover what to do
and what's safe to move because you're not gonna start with like let's say you got 30,000 VM's you're not gonna start with everything day one so how do you figure out what to move confidently
well we're announcing a new capability in data intelligence it's actually gonna crack open data formats and characterize your applications these are your development
postgress and sequel environments or this is your production mission critical Oracle environment so we have application classification those workloads will be tagged in fusion
and will now allow you to discover directly what workloads and associated Vms are ones you should probably take a look at doing now once you do that
you probably want to figure out okay what is all the config associated with that right vambers got a lot of features it's been built over are they using V motion what config exists
so this is where Model Context protocol is interesting for those who aren't familiar rest came around where applications needed to communicate it's for application to application communication
Model Context Protocol is how AI models should communicate with each other it's a new interface it's more uh meant for machine to machine communication but unlike rest it's not a strict format
which means your interoperability and your integration becomes very open to reach into the broad ecosystem so we're announcing extending our copilot with MCP where we can publish into the external ecosystem
and we can consume data from Vmware's MCP server and even Openshift's MCP server to say hey can these Vms move over there and then your storage config so we're extending Fusion's control plane to port work
so you can bring your storage related configuration over now you've done that you now know what to do you wanna go execute it so let's talk about moving the workload
we're delivering now on top of our orchestration platform templates to do these types of tasks so our partners have built recipes we've had Rubrik in the past
VM is a partner here has built um recipes as well into our platform but they're full automation recipes and this VM redeployment scenario is just an example of that
and those recipes are not just built on application connectors we can support AI connectors and pure 1 and fusion are both connectors that allow you to build your own automation recipes
to drive standardization into the landscape and then finally after you're done migrations are risky you wanna confirm everything's good so our co
pilot is consistently trained on more and more data so you can confirm that everything's good we've added Portworks to our training dataset so now you can ask questions of what's the new latency after the migration
um from Portworks because you can see the latency of your stateless containers so let's go ahead and take a look at this environment you have all of these brand new capabilities
where we've enhanced our policy enforcement with data intelligence we've added connectors and templates with a copilot for operational simplicity and you can go ahead and validate
from a compliance and governance what gets better so for those who haven't tried it I know we've got some customers in the audience and some prospects we have an AI co pilot
you can go into the product and ask it questions and normally you'd ask a question and it give you a response but we didn't let AI start automating things now it can trigger governed workflows and templates
so you can only do signed to govern things in a safe way so here ask you know what Vms should I migrate and it'll be like you know
we think these tier 2 Vms based on the data intelligence are things we should migrate and if you wanna double click into that right we're building trust
you can double click into our data intelligence and say no I want these Vms versus those or you can trust kind of R a I you have that ability but then the next thing you want to know is
what is that config so reach into vmware's environment and open shifts figure out whether the config and the parameters will allow you to move these workloads to open shift and we can see here that it is possible to do these Vms
that they look safe the latency the performance there's headspace there's capacity and if you don't trust it to execute just ask we've trained on the workflow so you can ask
what are you gonna do to do the migration what are the steps explain the workflow to me and this builds trust in what AI is gonna do as you take action into the environment you can double click into the workflow
and actually modify it if you're an administrator so you can see every step in the workflow you can see the logic associated with it so you know it's assigned and governed workflow and then say yeah
go do it keep me informed as you're informed you get notified that it's done sometimes it takes longer than this demo depending on the size of your VM and landscape but you'll actually see that hey
here's your source here's your target everything's running healthily so all that stuff I talked about bringing together people process and technology is now largely doable
you don't need to collect everything and know everything in one place you have the ability with AI to federate communication across your landscape this is connecting it
ops in a way that's never been done before and you know few days later you check the environment check the latency make sure you know you can see the environment everything's good this is using open telemetry
for Prometheus and Grafana a lot of people use that in the Kubernetes world um but you have the ability to validate and look at this entire scenario end to end so we're not launching a migration tool
this is just one use case of using the capabilities of an intelligent control plane to drastically simplify it operations and there are more
you're gonna see more templates in the future so this intelligent control plane brings together all of these capabilities and you see how fast it moved in six months imagine where we're gonna be
in the next six or 12 months as you're thinking about how to do this in a trusted governed way you need to reach across the ecosystem you need to reach into the intelligence that's possible and connect your automation
for those who want control prompt engineer it for those who want complete one click automation where you can keep going you don't need a gooey imagine this is a C I C
d pipeline where you hit this threshold do this thing and eventually you're interacting with a copilot just to send new rules into the ecosystem saying check for this and when this happens do that
we're gonna have observability on these workflows as well so in the future you'll be able to benchmark how long these things could take and then use AI
to eventually dynamically author workflows so you can see where this is going very very quickly so with that uh let me introduce your next speaker who's gonna talk to you about the Unified Data Plane
Rob Lee thank you all right good afternoon everyone this is a great turnout here in New York uh good to see everybody here today uh you know
we heard earlier from Charlie about the enterprise data cloud what that looks like how we're helping you to build it we just heard from Pacash about the intelligent control plane powered by Pure Fusion
and how that's helping to bring that cloud operating model into your own data centers but the reality is that you know you don't build a cloud of any sort
on top of a shaky foundation um and this is where pure truly stands apart pure is the only company the only vendor in the market that has built a truly unified data platform based on one consistent software technology
1 consistent hardware technology spanning the entirety of an enterprise's needs right whether that's across protocols block file and object whether it's across uh runtime environments
whether you're looking to run on premises in the public cloud in a hybrid cloud deployment for all of your workloads from AI to archive and everything in between earlier this year in our Las Vegas conference
we announced some significant advancements and additions to this platform today the platform is expanding even further for those of you who are eagle eyed and paid attention during Charlie's presentation
you may have caught the list of items what I'm gonna do now is unpack all of those launch items in gory detail so we're gonna be here for a while somebody lock the doors refill the snacks
alright joking um what I am gonna do though is place those announcements in the context uh that we think of them and we develop them towards which is really thinking about your key workloads
the workloads that we hear the most from you our customers our partners that are driving your thought processes your decisions your technology selection criteria the most severely
whether that's your business critical environments virtualization not sure if anybody's been paying attention there's a lot going on in that space these days uh whether
you know it's thinking about how do you move into the cloud and then certainly AI right and and how that's affecting your plans and so we're going to unpack each of these
and look at some of the announcements that we're adding in the second half that are going to further our lead in in each of these workload areas so let's start with job No. 1
right which is the business critical workloads these are the environments your database environments virtualization your core um it environments that run your business every day um
you know the traditional approaches here uh simply just have not kept up they they they leave a lot to be desired often way too siloed still inefficient
unreliable oftentimes not delivering the performance that's much needed if we look at why right held back by you know whether it's inefficient management
whether it's uh rigid uh and static limitations on performance capacity doesn't really matter why right the traditional approaches simply have not kept pace with uh
modern needs and this is where a pure really stands apart we've changed this game right since the beginning Flash Array and Flash Blade have enabled game changing
consolidation we've done this through a unified approach towards uh management our how our products behave how they're easy to use and frankly by bringing a ton of performance to the table
to enable that consolidation Flashray has been leading the charge when it comes to a lot of these mission critical uh business workload environments the databases uh the virtualization environments uh
and the Flashray XL line has been at the head of that charge it's been at the vanguard earlier this year we announced a significant set of improvements to the Flashray XL line
with the XL R5 with the R5 we introduced about a 70% performance improvement over the prior generation about a 70% performance improvement today
I'm delighted to announce we're setting a whole new bar we're announcing the Flashray XL one 90 the XL one ninety for set of workloads is going to double the performance of the XL
one seventy alright so set back earlier this year a couple months ago we improve performance about 70% we're going to increase that another hundred percent alright guys it's a pretty big deal come on
alright so with the XL one ninety we're for a certain set of workloads doubling the performance or having the latency are really targeting a lot of these you know
read intensive mission critical database environments frankly with the XL one ninety in the whole XL line we're we're setting a new bar and we're bringing the highest levels of performance coupled with the simplicity
the efficiency and the reliability that these environments need so we're ready to take on the largest big iron environments that are out there now speaking of performance
earlier this year we also introduced the Flash Array ST the Flash Array ST is oriented and and really targeted at ultra low latency uh uh database environments
uh applications that run on direct attached storage today we're also uh happy to announce the latest version of the Flash R a s t the Flash R a s t R 5 further improves those performance levels
almost 20 million IOPS almost 200 over 200 gigabytes a second throughput from a single system and so really the way to think about the Flash R s t is it's giving you the best of both worlds
it's giving you direct attached storage like performance we're talking you know we're thinking it we're starting to think about latencies in the microsecond level
uh coupled with enterprise capabilities the management the reliability the data services that you need need it's no longer a trade off it's no longer a choice alright
so these are these are some pretty impressive numbers whether it's the XL whether it's the Flash RS t I want to step back and put this into context alright if we look at where we sit relative to the competitive set
there's really no comparison right we're now delivering over nine times the performance density nine times the performance density we're delivering almost five times
the capacity density compared to the nearest competitor alright um what this slide leaves off is maybe the most important number
which is 10 to 20 times the reliability so these are pretty impressive numbers the comparison gets even more stark if you look at it at the fleet level if I look at a single rack of flash ray XLS
a single uh uh rack of flash ray XLS the competitive set would have to deploy 22 racks of equipment 22 racks of equipment consuming over 12 times as much power just to match that single rack of flash rays
in performance and capacity so you know if you step back from this Charlie made this point earlier um you know there's simply no argument the legacy architectures have not been able to keep up
and the pace of our innovation isn't slowing down we're continuing to innovate we're continuing to push the boundaries every single day so I've hit a bit of the new announcements
and a bit of the additions and innovations in the Flashray family we hit XL R5 from earlier this year we hit XL one 90 the Flashray ST Today we're also announcing the newest versions
of the Flashray C and the Flashray X with these latest versions the C and the X are getting 30 to 40% more performance more capacity you see so you step back from this and the entire flash array line
soup to nuts has you covered for all levels of performance all levels of efficiency for all of your business critical workloads all right
so let's switch gears let's talk a little bit about modern virtualization we know there's a lot of choices in virtualization a lot of you I know I've actually met with a lot of you on this topic
uh a lot of you are probably sitting there looking at different options whether it's uh thinking about how do you go optimize your Vmware footprint
whether it's thinking about alternative hypervisors uh such as Nutanix whether it's thinking about modern virtualization or whether it's thinking about getting to the cloud um our goal
and really our mission at pure is to provide you the best of these options no matter which uh makes the most sense for you today and which makes the most sense for you tomorrow
uh but for those of you who arrived a few minutes early you saw some of the videos uh that Charlie and um uh the Nutanix team did around our partnership I will tell you that here at pure
we're investing heavily in one of these paths that we see a lot of our large customers investing in which is modern virtualization with Qbert with Kubernetes can now manage and orchestrate Vmware VM based workloads
we're working really hard with Red Hat Openshift to enable this with Portworx for Kubernetes um together uh Kubernetes can now bring uh you know bring a single
a consistent management paradigm uh approach to both the the hypervisor based virtualized workloads as well as the container based cloud native workloads you can bring them under one house
uh Portworks obviously provides the storage capabilities for this uh but what's maybe lesser appreciated is only portworks can provide the enterprise uh uh critical features such as live VM migration
Dr replication etcetera that you need to be able to deploy mission critical environments uh on top of Kubernetes by bringing uh
both of these environments the traditional VM based uh uh workloads as well as the cloud native workloads together under one management roof
it also gives customers it gives you a lot of flexibility and optionality to think about hey uh in the future as you think about replatforming or rearchitecting
rearchitecting uh some of your applications you're now able to do that at your own pace without having to change your run books and change your management paradigms
now um some of our customers are also uh looking at how do they best move to the cloud and this is where Pure Storage Cloud comes in right uh Pure Storage Cloud helps customers
unify their experience whether they're running uh on their own premises in a Colo in a hybrid environment or on the public cloud uh since 2018 we've been helping customers
get the best attributes out of the public cloud environment whether it's on AWS whether it's on Azure with our Cloud Block Store technology bringing um you know the same data services
the same reliability the same uh capabilities we provide in our purity software and flash array to the public cloud infrastructure today we're taking a bold new step with the Cloud Block Store technology
with Pure Storage Cloud we're announcing a fully managed service based on Cloud Block Store that's now generally available as an Azure native service
tightly integrated with Azure VM or service this combination now allows you our customers to uh freely and easily move your Vmware based environments into the Azure uh cloud without having to worry about uh
without having to worry about capabilities reliability um you know or or cost optimization uh you're gonna hear a lot more about uh our partnership from uh our Azure partner following uh following my session uh
but suffice to say this is uh one of the fastest growing offers that we have in our portfolio all right AI top of mind for all of us um AI and really technical computing in general uh
is a space that we pioneered uh specifically around unstructured data we started with Flashblade s which really has been the leader in this space and as Charlie mentioned in his opening remarks
we've now found a way to extend our leadership with the Flashblade s technology into Flashblade Exa the combination of these two products now allow us to address the entirety the entire spectrum of AI needs
from the smallest research groups to the largest enterprise needs to the largest uh uh you know foundational model builders and GPU clouds
in the world earlier this year uh we introduced a new version of Flashblade s uh the SR two with the SR two again we almost doubled the performance
compared to the prior generation and we made the fastest file and object unstructured platform in the market even faster we're now about twice as fast on any measure whether it's you know bandwidth performance
metadata performance concurrency compared to the next nearest neighbor customers are choosing uh the Flashblade s for all their enterprise needs uh so certainly because of the performance
you know that's that's uh that's important but also the combination of that with the simplicity the reliability and uh you know and and uh um uh the manageability right as uh
AI projects graduate into production environments uh you simply can't run these things on uh science projects anymore more importantly we've worked closely with Nvidia since the inception of you know AI
to certify our Flashblade s platform with all of the key certifications whether it's you know OVX Base pod Superpod or NCP the Nvidia Cloud Provider Reference Architecture
now there are some environments that demand even a higher level of scale of performance and capacity Charlie talked about this up front uh and to address those environments
we've extended the leadership we've extended the metadata we've extended the software engine uh that is so uh successful in Flashblade s and extended it with an open hardware architecture
and this gives us Flashblade Exa with Exa uh we and Charlie mentioned this as well we're now able to address some of the largest GPU cloud environments in the world
and really the best thing about this is because it extends uh Flashblade s because it's built on the same technology you can start with a Flashblade s which we believe will be more than enough performance
scale and capacity for most customers needs and you if you ever need a little bit more you can grow into an X now the reality is probably most of us in this room uh aren't you know
going to be training trillion parameter models and uh uh building foundational LLMs most of us are probably thinking a lot more about inference and in the inference space um you know
we have a really exciting announcement to make as well which is the latest version of our pure uh KV cash accelerator uh our KV cash accelerator is a mouthful um really
the way to think about this is it's a acceleration capability that provides about a 10 to 20% I'm sorry 10 to 20 times uh performance speed up for critical phases of uh
the inference pipeline natively integrated in our storage stack optimized for S3 and NFS access around the latest LMS uh with this latest release of the pure KV a we're now fully integrated uh
into the Nvidia Dynamo infrastructure the take away from that is this thing is ready to go ready to deploy uh around all of the Nvidia delivered micro services uh and and models
all right last but not least when we think about AI we all have tons of data that we want to apply AI to and our focus on performance and efficiency you know is relentless
so for those bulk data stores uh we're announcing Purity Deep Reduce uh this is bringing uh similarity based data reduction technology into the fold
to make those bulk data stores even more efficient uh this is an area we lead in today with Purity Deep Reduce we're going to be targeting incremental um benefits in terms of data reduction of about 2 into 1
so those large data stores are about to get um even well I guess even more efficient um even more economical and even more ready to be applied to AI
alright so I hit a lot of stuff um I didn't hit everything on Charlie's list uh but I'd be remiss to uh not to mention that all of these innovations right all of the product announcements
the feature announcements all the software ads everything we do is available through Evergreen everything we do is available non disruptively uh you know
you have a non disruptive path to get from where you are to uh um uh deploying these innovations and unlike other vendors who may claim uh similar things we've got numbers to stand behind
we've now done over 30,000 non disruptive upgrades and we're doing more every single day alright so um I talked a lot about workloads
talked about business critical environments I talked about modern virtualization cloud talked about AI and technical computing um there is one I would say horizontal concern
that we all spend a lot of time thinking about which is security and cyber resilience now uh you know if you're like many uh in the market um
you may see your security posture look something like this patchwork of different solutions it's uh you know belt suspenders
helmets shoulder pads duct tape baling wire the whole 9 yards um unfortunately security is a space where uh more is not always better
uh and the issue with a patchwork type approach is you lose sight of what is your true security posture where are your real threats where can things pop up and when they do pop up what are you gonna do about them
well we've got a little bit of a different view as to how to approach cybersecurity and maintain that posture and if I oversimplify it I think a little bit like I think about it a little bit like securing my own home
right if I wanna secure my home I wanna start with solid fundamentals make it hard to get in get a solid door put a door bolt a deadbolt on maybe put bars over the windows
maybe put a fence around the perimeter I wanna have good uh you know visibility I wanna put some sensors out maybe cameras maybe motion sensors
um you know maybe door sensors uh etcetera and then what I'd like to do is put myself in a position to take action when I have uh
you know a threat right it doesn't do me any good to have the alarm guy call me and say hey someone's breaking in and we're dispatching the cops will be there in 25 minutes right
so I wanna put myself in a position so that it's hard to get in I've got good left right visibility to all that's going on around me and then when there is a threat
I'm in a position to act before something uh really goes wrong cyber security is not that different and this is really the approach that we're taking with our enterprise data cloud to enable a resilient and unified uh
cyber uh uh posture for you guys No.1 we start with a solid fundamentals whether this is built in primitives like safe mode whether it's uh
you know whether it's uniformity through the control plane by bringing uh the management and security uh policies all together we're reducing uh
the likelihood of drift we're reducing misconfigurations we're reducing human errors we then pair that with built in native capabilities to proactively and dynamically respond
and then we work with ecosystem partners to connect to the best sensors and the best sensor networks out there so what does this look like and what are we adding so two things on the proactive defense and recovery uh
point of uh um dimension um we've got a lot of signals Procash talked about how we make you know we measure
more about how people use flash than anybody else in the world we've got a lot of signals that are available to us whether it's entropy of data Io patterns access patterns
administrative logins etcetera we're now taking those signals and we're doing intelligent real time scanning to identify malware threats as they happen
now that's only part of the equation right we then take those signals those real time um you know threat indicators and we pair them with a dynamic response
we can uh pair that with Pure Protect and spin up instantly uh auto configured recovery zones so think about it this way something's going wrong we detect there's a signal
we immediately lock down the array we can spin up a recovery zone one for failing over to one in isolation so you can debug it one maybe for forensics all before you're notified
all right and then to further um amplify this we are announcing a set of integrations with ecosystem partners we're working with the top names across the industry to
uh really broaden our sensor network right we wanna get a lot more visibility to the environment around us in your uh in your data centers
to help you identify threats and respond to them more quickly whether it's working with Crowdstrike around seem and soar integration uh to get connected to the best threat intelligence
in real time uh we're working closely with Superna around file and user monitoring uh to detect not just a ransomware uh you know events
but also unauthorized data exfiltration and then we're working with VME to deliver a full cyber resilience recovery scheme delivered as a service through that partner um I don't have a ton of time
but you know all of these partners have a booth I um encourage you to stop by during a break to learn more about these integrations so to bring this home
um you know I think cyber resilience is a great example of where the enterprise data cloud really comes together Charlie talked about um
you know how uh the power right of being able to manage storage as a horizontal um you know as a horizontal resource Procash talked about uh
the intelligence that we can then add with a control plane that spans that and all of these integrations that I just talked about whether it's with Crowd Strike whether it's with VM whether it's with Suprima
are integrated directly into fusion The Enterprise Data Cloud is what brings it all together and hopefully I've given you a sense of how we're innovating on all of
the levels of the enterprise data Cloud from the intelligent control plane our integrations around the environment to further expansions of the core unified platform thank you
Prakash Darji, General Manager, DX at Everpure and Everpure CTO Rob Lee unveil new innovations that power the Enterprise Data Cloud, drilling down into how our intelligent control plane and unified data plane work together and announcing updates on FlashArray™, FlashBlade™, Evergreen® and the Everpure Cloud.
We Also Recommend...
Personalize for Me