00:04
let's get it started please welcome to the stage our CEO Charlie Jane Carlo Charlie come on out well uh thank you Andy
00:19
and thank you to all of you for coming out on such a lovely day as Andy as Andy mentioned uh we know that uh it wasn't necessarily easy getting either crosstown
00:28
or even from even the subway stop uh here during the rain so really appreciate all of your uh all of you coming out to see us so we do as Andy mentioned
00:38
we have a lot to talk about and uh really I think the way I'd like to start is just to tell you where we've come
00:44
over the last several years cause many of you actually have known us for a long time but we the last several years have really brought a lot of innovation and change
00:52
so first off you know it's been a uh 11 year journey of products in the market and consistently gaining share and growth uh as we've uh
01:02
as we've gone through those 11 years and really it's it's thanks to our customers and our partners that have brought us here uh depending on how you look at it
01:10
we're either No. 2 or No. 3 in all flash but really the important thing is that our overall market share in storage has consistently grown year after year
01:20
and really we're the only western company that has grown um in in storage over that period of time if we look at our last Q2 uh we're just shy now uh of a billion dollars uh
01:32
in terms of of revenue on a quarterly basis uh very profitable 13% growth overall uh also almost half now of our uh revenue comes in the form of subscription so this is and you'll hear from Prach later
01:46
our ability to sell our our product not as a product itself but as service level agreements to companies where we just promise performance capacity delivered where and when you want it
01:58
and that comes in as a subscription in addition to that you know we in we have for a long time been you know a very heavy investor in R&D and we continue to be
02:09
so over 20% of our revenue we reinvest in R&D and of course it's not what you invest it's what you get out of it but it has allowed us to continue to gain share we this is a key differentiator for us
02:23
right from the beginning our view was that data storage was high technology not a commodity and we invest in it uh assuming that it's gonna has to continue to scale at the same rate as compute
02:36
as networking and frankly as new use cases such as AI so what is 20% per year of spending in R&D get you well for 11 years in a row we've been the uh we've been the most innovative according to Gartner in the
02:49
Magic Quadrant and while that's important this number is equally important it's gotten us consistently the highest marks in terms of a from you our customers in terms of Net Promoter Score
03:02
so well over 80% we're easy to use highly reliable non disruptive upgrades great customer experience whether it's with our sales or our support organization so so that's you know
03:16
where we've been so what's next well as I'd like to say let's talk about the elephant in the room what is the elephant in the room I think we all know it's got two letters it's AI
03:28
but maybe what I'm about to talk about is not what you're expecting because what AI is really doing in the enterprise environment is it's changing
03:37
the relationship between software and data so before the last a few years it was only a few years ago when a great sage said that software was eating the world right
03:52
and its relationship to data was really software and software applications and the the apps that you were building whether it's for iPhones or the websites or the or or the applications you were building
04:03
for your customers to be able to get access to your goods and services that was eating the world but a funny thing has happened and that is AI has come along
04:14
and now questioning whether software is eating the world or whether data is going to eat software whether data is actually going to replace uh some of the software tools we use every day
04:27
or already in San Francisco believe it or not self driving taxis outnumber the rides on self driving taxis outnumber the rides on on um
04:39
ride sharing apps Uber and lift and self driving taxis were driven by data whereas of course Uber and Lyft was driven by software so it's an interesting transition that we are all going in so what does it mean when software and data
04:54
when the relationship between them change well I think it changes the nature of the way we need to think about enterprise data we as users of data we as custodians of data it changes it so let's describe how enterprise data is
05:11
is used today where it exists how what the architecture behind enterprise data is and to do that we have to talk about the enterprise data center what does that look like well
05:19
typically you build an application let's just take database as an example how do we build a database application well of course you have to choose database software but you place it on a server
05:30
that server might be virtualized may be equivalent to any other server in your environment um preferably it is it's connected of course uh with switches uh
05:40
connected to a network so that it can serve up its uh it its uh service and then of course you choose if you use external storage you choose storage that is designed specifically
05:51
for database performance resiliency etcetera okay then you have another application it might be VDI it might be file sharing it could be email could be backup maybe the servers are are roughly the same they're
06:04
they're virtualized so you can use any server the network is the same but what's not the same is storage storage is chosen for specific characteristics price performance
06:15
to meet the needs of that specific application different storage for different application types so what does that mean well arrays are managed and provisioned individually manual work fingers on keyboards
06:28
you can't share even if you have the same type of application tied to two different um arrays generally you can't share capacity or performance between them
06:39
the array is captive to the app the specific application that it's connected to uh new services are allocated when when you have a developer developing a new Cape uh capability that capability is allocated to a
06:52
an array somewhere and usually done manually and then finally data governance standards whether that is security whether that is resiliency characteristics um whether it's where you want that data to be kept
07:05
or where you want it to be uh backed up it's applied manually alright so that's what the storage characteristics are what does it mean about the data now that you're storing well it's now inconsistent
07:16
your data is captive to an individual application stack not available generally easily to others unless you go through the application stack any movement of the data or any copies of the data that are made
07:29
are usually manual which means there are manual records if there are any records in terms of the copies um or the data movement that's taken place it also means that raw data is generally not accessible to other applications
07:46
you're actually going through the application stack and then finally if you have assuming you have data governance standards when it comes to the data those David data that
07:56
those standards have to be applied manually on keyboards array by array right it is not consistent okay so let's
08:05
just look at a different way of doing these things let's look at the way a cloud operates and by a cloud I'm talking about a hyperscaler cloud or any large organization a large SAS organization that's providing services
08:16
where they want to do it in a standardized uh highly automated way the way the cloud builds applications is entirely different in fact they don't even build applications
08:25
what they build is a flexible cloud infrastructure that can host applications and it starts with consistent compute in the middle tied together so compute services tied together with the network
08:36
of course and then they do build storage services but entirely differently they have a horizontal array all this storage looks the same with the only difference potentially being the performance levels
08:49
so they'll have a high performance layer of storage they'll have a mid performance uh they may even have an archive layer so they will have three or four different price
08:58
performance layers of storage but all of that storage within a layer is available to any application at any time of course they tie this together with a network services
09:07
and a data center management environment such that applications can be placed anywhere data can be placed anywhere networks can be configured for specific environments and it's all software defined
09:20
right so not the typical way that data centers that that enterprise data centers are built there are some that are going in this direction but guaranteed their storage is still bespoke to specific applications
09:33
so what we uh think is that data storage enterprise data storage architecture if you think about it is designed to be vertical whereas cloud data storage architecture is designed to be horizontal
09:46
and that has a lot of different connotations one is that enterprise storage tends to be highly manual you have a lot of data storage admins fingers on keyboards
09:56
whereas cloud storage is entirely automated uh whatever policies they have is done in software uh and applied automatically to the software layer that means that enterprise data storage tends to be siloed
10:12
whereas in the cloud data is accessible to anything that is given permission that is given permission to use it and that's a simple software call and then finally
10:25
enterprise data storage tends to be very physical and cloud is now virtual so I like to say this concept is virtualizes data storage and it's the last of the three major architectures uh assets to be virtualized
10:39
okay so what does that mean for pure what does it mean for you our view is that between AI driving the need for new data architectures and what we can do from a technology standpoint
10:52
that it calls for a new era of data a new type of data architecture more of a cloud operating model inside an enterprise environment so um how are we doing that for you
11:04
and this is the the big reveal if you will that we introduced earlier uh this year and for us it all starts with purity purity is the one operating system that we invest in that supports all different types of workloads
11:18
so think of it as a platform that's a common operating environment and it has all the evergreen properties that we talk about the ability to consistently upgrade and
11:28
and change nondisruptively to the application to always be modern to never become obsolete purity it's not just for block we now support block file and object all in the same purity environment
11:43
so it's consistent whether it's APIs or the way that your um your employees the developers are able to utilize it and now and this has been some of the investments
11:56
we've made over the last few years to be able to penetrate the lower price levels of of traditional hard disk environments we now can operate at any tier of storage whether it's the highest performance tier for AI
12:10
or the low price tier for things like archive and so forth so now you can have one environment across any mode of of um of data storage and across any performance level
12:23
and of course as you know in enterprise storage there are a lot of features for for security for resiliency for performance uh data reduction and all of that is built right into purity um
12:36
overall now this is what really allows us to say that we have a unified data plane it's one software environment for any type of data storage need that you might have
12:48
and the benefit of that is that we can manage it all in one uh one management environment and because of Evergreen we can also provide it as a service so this can all of this is available
13:01
I whether you buy the product and manage it yourself or whether we provide just the capabilities of of this of this system and you just rely on us for service level agreements performance resiliency etcetera
13:17
this is what allows us to support all of your different workloads whether it's traditional enterprise apps or now increasingly modern apps whether that's a IML or
13:29
or a modern virtualization even container environments we can support it all with purity overall now I talked about the big reveal the big reveal is what we've recently added and we needed to have a common operating environment
13:42
in order to be able to do this and it's something that we are calling fusion now fusion adds a intelligent control plane to all of our arrays and this is now built into purity
13:53
so what does a common control plane mean it means all of our storage devices communicate with one another and they instead of operating like individual arrays they operate like a cloud of storage
14:07
workloads can be placed automatically workloads will be able to be um load balanced automatically you can set presets you can set policies and your the workloads
14:20
the data will be managed by policy rather than fingers on keyboards so it automates workloads and workflows it automates how data is managed and it allows you to put your policies into software
14:34
so that those policies can be applied automatically and not with fingers on keyboards and when I say that it uh allows a cloud of data we're talking about not just within a data center we're speaking about a global cloud of data
14:50
uh each each different data center might be a different availability zone but now you can really network and create policies for how your global data is managed and treated and this is what we mean by an enterprise data cloud
15:04
our capabilities are purity infusion but it allows a customer now to start to build and architect their both their storage environment but also their data environment as a cloud
15:16
rather than as individual storage arrays so let's just compare the two a captive array that every the way we've been building storage for the last in enterprise for the last 30 years
15:27
is the arrays require manual provisioning but with the Data Cloud it's automated provisioning based on the policies you set you have dedicated in the case of uh the captive arrays
15:38
it's dedicated capacity and performance rather than shared capacity and performance uh as in auto load balancing governance on the left is manual governance on the right is by software
15:51
by the policies that you and your organization set Protection again is manual in the case of captive arrays but it's by policy again in the case of a data cloud and then finally
16:06
instead of the data being captive to a stack data is now accessible for other application environments with the right authorization um and so it really frees your data
16:18
in order to be utilized by other things and the way I think of that is the enterprise data Cloud really allows you to turn your entire data estate your production data into your data lake
16:31
cause think about it today in order to be able to analyze data you have to move it you have to copy it you have to move it but you have to buy another array that is tuned to being a data lake
16:40
why why shouldn't your production data be available to be your data lake another way to think about it is you're replacing data silos with a data cloud
16:52
okay so um what I'd like to I'd like to switch uh topics now just briefly one is we're gonna talk a lot in in the next with the next set of speakers
17:03
on new innovations that we brought to market and just this year alone as you'll see whether it's on the control plane that I just talked about the intelligent control plane whether it's our unified data plane whether it is uh
17:17
the s the the expansions that we're doing for our service level agreements in Evergreen 1 which is our storage as a service offering
17:24
or whether it's cyber resilience and I had to cut a lot of this short because there were just too many things to put on the slide um we have introduced a lot of capability this year
17:34
and you might ask well how are you able to do that well when I was showing you how far the company has come at the beginning of my presentation
17:44
the one thing that I didn't mention is that is a little a statistic that you have to dig into our financials and that of our competitors to understand and that is as of this quarter
17:57
not only are we spending more as a percentage of sales on R&D we're spending as much or more on R&D in terms of actual dollars as any one of our competitors in storage so
18:08
my view is the only grower and the fact that we're spending already as much if not more than many of our competitors you're just gonna see us accelerate in terms of growth and in terms of innovation uh
18:20
that we can bring to you as customers so I just want to focus on one of these things to show you the flexibility of our platform so earlier this year we introduced a product we call Flashblade X
18:32
now to be honest with you we were behind in the very large scale um AI market yeah we had built products for enterprise Flashblade was our leadership product
18:44
handled AI very very well in large enterprise environments but GPU clouds were something different and we didn't have a product that scaled at that level so we looked at our core technology and said
18:56
how do we stay true to our ethic of always staying focused on purity saying focus on one operating environment and yet scale to these levels and so we introduce Flashblade
19:08
now what we found is that we have in um in our Flashblade product and in purity the world's fastest most powerful metadata engine but we were using Flashblade for two things we're using it as the metadata engine
19:23
but also as the storage engine and we said well what if we separate storage from the metadata use Flashblade Exa and purity for what it's really good at which is meta a metadata engine and then tie
19:35
GPUs to um Jboss to just a bunch of very high speed flash and so we created this Flashblade exit so Flashblade exit is highly scalable not only do we tie it to a bunch of J boss
19:50
but as the as a as a customer scales we can just add more we can and every time we add another Flashblade with another stack of J Boss it just gets faster and faster and more powerful
20:05
and in fact what we've built is a pro is a product with Flashblade Exa that is 5 times faster than anyone else's out there in the market and we did this preserving purity
20:16
we did this preserving Evergreen this is something that a customer that starts with a Flashblade s for let's say a small scale uh uh AI type environment can scale up to non disruptively
20:28
as they as their needs grow should they need this frankly we think this is really GPU cloud oriented uh capacity and performance uh in this
20:37
so the way to think about it is and we've talked about this before this is actually the beginning of an old slide this slide identifies that we really cover now whether it's capacity whether it's price
20:48
whether it's performance you know our product line all based on on purity whether it's file block or object covers the entire price performance base of your needs uh in storage
21:02
what we've just done with Flashblade Exa is scale that a tremendous you know by two orders of magnitude so we can handle hundreds of thousands of GPUs not that that's necessarily interesting
21:14
but if any of you have an interest in hundreds of thousands of GPUs please come come talk to me um but you know at the fastest level 10 terabytes of reads
21:23
5 terabytes of sustained write performance which no one else can do we're already up to 3 trillion objects we think we're gonna get well beyond that so it's an amazing engine
21:33
it just shows you the power of what this model can do the power of what we've done with uh purity the power of what we've done with our platform and now that we're scaling it so that you can build a enter your own enterprise data cloud and uh
21:47
virtualizing your data storage for the first time allowing you to operate a more efficient data cloud one that can be more secure more compliant with your policies uh and more available for analytics
22:04
more available for any AI that you may wanna do we believe that this is the architecture for data in the future
