00:01
Mm Hello and welcome to Tech Best 22. I'm Corey grown and I'm a product manager working on our Evergreen one subscription service And there are so many exciting new things to talk about here at Tech Fest 22 Pierre has made so much progress in the last year on advancing the frontier of technology around storage.
00:27
And one small part of that story is how we are advancing evergreen one to incorporate the awesome new powers of the Flash Blade? S platform into our Evergreen one subscription service. So I and Andy are here to talk to you a little bit about what's changing and to deep dive into what those changes mean for you. So I'll let Andy introduce himself and walk you through some of the great new things with Flash
00:53
Blade? S. Thanks a lot Corey. So I'm Andy Kuttner. I work on the Flash Blade product management team and let's get right into why we are introducing these services. So as our customers look at unstructured data, one of the things that they really find is that the unstructured data is growing, it's growing quite quickly.
01:13
And so what they really are looking for our platforms that can both handle the growth of the data as well as the analysis and extracting value out of that data. Now. Typically they might think of going to a public cloud for something like this but one of the main challenges around public policies, it's not necessarily cost effective for this type of data in order to be able to ingest the
01:33
data process and extract value. It costs money in the public cloud and it usually costs a bit more than what they're willing to spend and look at investing in. And then lastly, if they take a look at their traditional legacy storage platforms. One of the problems they have is that they can't necessarily keep up with the way the data is growing and more importantly, the way the customers want to analyze that data and get
01:55
some value out of it. I think you hit on a really important point there Andy, which is when customers are considering about the right venue for storing their unstructured data, they're often thinking only about costs and performance takes a backseat. And I think anyone who's been a practitioner of data science or ai ops understands the value of fast access to the data.
02:23
They need even data that may not seem value on the face of it. All. Data is valuable when we're talking about machine learning. That's exactly right. And so when customers look at, how do they address these challenges, What they're really looking at is how can they get a storage system that's future proof?
02:40
They know it needs to grow, they know that they need to add value, they know that it needs to be scalable. How do they actually pick that platform and make that happen? How do they make that happen without spending an inordinate amount of time doing storage management. They don't want to learn how to manage these
02:56
systems, they don't want to run these systems, they just want to be able to get the systems in their environment ideally having somebody else operating them and they can just start using them to full value out of them and they want to move off of these inflexible platforms so that when that platform needs to get upgraded or expanded that it's not causing them downtime or pain associated with any type of migrations.
03:20
So if we look at this and we go out and try to look at the markets and say, hey, these are some of our thoughts around it, but really is this really what customers want? The answer is, yes. What we're seeing from customers through market analysis is that they're expecting the volume of data to quadruple over just the next few years. So that's a huge growth in the unstructured
03:39
data of what they need to be able to manage and take care of and again try to get the value out of. But yet almost half of those environments and half of those organizations want a simple consumption based model, right? The idea of cloud is really a good one, right? They just pay for what they used to be able to
03:55
get that value out. They don't want to be expending any more money than they have to. And then lastly again, they don't want to be buying data, but they won't be able to get the data out, but they don't want to be spending time migrating platforms, they want to get out of the business of running storage and they just want to be
04:12
consuming it as a service. So that's that's fascinating. That's fascinating India. I mean we hear all the time stories about data growth continuing to expand. But the most interesting fact here to me is the last one because when we're thinking about purchasing storage technology that focuses on the immediate near term purchase and you
04:36
know, just like when I'm purchasing a car, I think about how much is this gonna car gonna cost me today. But I don't think about how much the maintenance is going to cost me over the next five years. Exactly. So Corey, what have we done with our evergreen one service to help solve some of these problems for customers.
04:54
Thanks Eddie. So let me take a step back and just refresh everyone on why customers choose Evergreen one. When we talk about evergreen one and the core philosophies and values of evergreen one really center around three things. Number one cloud economics we strive to bring all the best aspects of,
05:16
of public clouds to evergreen one. And pure technologies by giving customers a flexible agile financial way to consume pure technologies. And so we allow customers to change their mind over time. We allow customers to focus on only what they know they need today and grow later because we all know none of us have a crystal ball. we don't know for sure how much storage we're
05:45
going to need in three years or five years. Right? Nobody does. And so our solution of that problem says you don't need to know you can be right in that answer And pure will help you make sure you're optimizing your costs and that's the cloud economics of evergreen one second cloud operations, we all know the IT industry is in transition, I thi practitioners are trying to
06:12
spend less and less time on activities that don't actually add value to their business and that includes things like daily care and feeding low level management tasks. And with Evergreen one pure wants to help our customers and our users advance their skills and grow their careers and add more value to their business and we help them in a variety of ways. One of which is our pure one cloud management
06:40
platform, which I'll be talking about later and third we think about cloud experience, we want to deliver a unified cloud experience wherever customers choose to consume their storage technology that can include on premises or in a co location facility that can include public cloud venues like AWS and Azure. And as we all know, the reality of our world today and tomorrow is hybrid cloud.
07:06
More and more customers are managing across a variety of environments, both on premises or in co location and in a public cloud environment. Often customers have key important Tier one workloads running on premises and they have newer exploratory projects running in public cloud and they're both important. And that's why peer wants to meet our customers where they are with Evergreen one and allow our
07:33
customers the flexibility to choose a hybrid cloud configuration. Yeah. Cory. You know, some of that really resonates with me when I talk to customers particularly around the economics and the S. L. A. The consumption based model and the simplicity of it. Could you walk us through a little bit about
07:49
maybe how it actually works? Yeah, sure, customers often are curious about how does Evergreen one work? How much am I going to pay for now and how much I'm gonna, am I going to pay for later? And what's that cloud flexibility that cloud economics you're talking and here's what I'm
08:07
talking about when we think about how customers consume Evergreen One services, there's two components. One is what we call the reserve commitment. So all of us have fairly stable workloads that grow a small amount day by day and a and a relatively small amount year by year. And we have a pretty good understanding of how much storage we're going to need to support
08:32
those workloads today and tomorrow. And for those types of workloads we allow customers to commit to a certain amount of storage consumption and in exchange for that commitment, we offer customers a better price, a discounted price for that reserve commitment and that's reflected in the gray bar at the bottom of the screen. And that sounds wonderful because customers can
08:54
start small. They can start with an initial commitment of as low as 50 terabytes for our evergreen one block service and as small as 75 terabytes for our unified fast file and object service. These uh this service is built in advance and we allow our customers and our partners to choose how they want to be built.
09:17
And that's very important. Different customers and partners have different needs in terms of when, when they want to pay their bills and we want to give customers that choice and flexibility. But the real magic of Evergreen one is reflected in the orange line on this graph. All of us here know that the reality of storage consumption is not the gray bar as we
09:39
learned from Andy earlier data storage, data data storage vines are growing. People are storing more and more data every day. And that's a trend that has been true for a long time. And it's going to be true for a long time into the future. But that's a macro trend right growth over time. But the reality, the daily reality of storage consumption is that it fluctuates.
10:04
It goes up some days. It goes down other days and we want to make sure our customers can efficiently consume their storage in those daily or weekly or monthly fluctuations. And so what we do as we allow customers to consume storage in excess of whatever they have committed to. So they can have that fluctuation above and below.
10:27
They can burst up they can burst down and as they consume more storage in an excess of their reserve. We uh there's a there's a separate non discounted rate that they pay for that on demanding. Now some of you may be asking um That sounds really complicated right? How can I be sure that there's always going to be room for me to exceed my reserve when I need
10:52
it. How can I be sure that there's always going to be enough on demand capacity that I can use it when I want to. And that's your job. That's one of the promises that we make to our customers with evergreen ones. Is that pure will monitor that service level and make sure that if it's getting too
11:12
low for example, we will actually ship out more pure equipment to allow that additional usage of on demand. Were the ones who make sure you never run out of space. And that's one of the promises we make to our customers with here is a service with evergreen one. That's great. So but you know as their on demand usage goes
11:32
up um you know shouldn't they be able to get price discounts through like you know increasing their reserve and their commits. That's a very good point. So as you see on this graph, the orange line does go up over time and it stays above the initial reserve commitment over time over those first two years.
11:54
And so if I were an evergreen one customer, I would say, well this doesn't seem fair. You're charging me a higher rate for the entire area of that orange curve above my initial reserve. So how can I get a discounted rate and how can I optimize my cost. Thinking back to the cloud economics philosophy of evergreen one over the life of a contract.
12:19
We allow customers at any time to increase their reserve commitment as they start to use Evergreen One services. They can, I can see how their usage patterns fluctuate but maybe stabilize over time. And as they see that stabilization, they can come to us and say I want to increase my reserve and I want to pay a lower rate because I'm trying to optimize my costs.
12:42
And so we allow that flexibility for customers over the entire life of the term. That's excellent. But you know, your graph here shows something pretty interesting which is, you know, projects come and go. Sometimes they ramp up, sometimes they ramp down in this case you're showing a project that's ramping down. So could you talk a little bit more about how
13:01
that works at the end of the original contract. That is also a great point and very observant of you. Andy so you can see that for whatever reason maybe I had a cloud migration where I migrated some workloads to the cloud. Maybe I had, I had a different, maybe I decommissioned an application that I'm not using anymore and I put it in cold storage
13:23
or something like that. For whatever reason I'm using less evergreen one services than I was before. And so we at pure storage don't want to lock our customers into long big expensive commitments that doesn't serve our customers best and it doesn't serve us best as pure storage. What we want to do is have our customers have
13:45
the flexibility at the time they renew their contract to reduce the amount they're committed because as I mentioned before, none of us have a crystal ball and so this is how we bring that flexibility and cloud economics into evergreen one services. Excellent. Thank you for this overview of how it works. Let me jump in. Yes. Let me jump over really quickly just to
14:11
introduce a little bit more of the details about the actual service and the tears that we're gonna be offering. So for the unified fast file and object, which again is a platform that provides both fast file as well as fast object. Within the same platform. We now are going to have three service levels associated with it.
14:28
We start with Ultra. That's a service level we've previously had and now what we're doing is we're increasing the performance quite dramatically to be able to give a good boost for customers who have use cases, where they need high performance are actually, I should say the highest performance for scale out file or fast file or fast objects services. We still also have a premium tier and in that
14:49
case we've actually increased the performance there as well. And so those are some use cases around perhaps imaging data or analytics type of data and I'll cover these use cases a little bit more detail on the next slide. And then we've also introduced the performance tier which is still a high performance in particular a scale out here for for those areas where you need good performance across a wide
15:13
variety or a highly concurrent environment. And we'll talk a little more about the ransomware and rapid restore capabilities of that coming up on the next slide. So again, we now have three different tiers will offer much broader differentiation in the services. So it's a good way for you to be able to pick
15:32
and choose what you need for a particular project. When you're looking at the evergreen one service, let's talk a little about those use cases. Um Again, if you're looking at something like a machine learning or ai type of application, you're really looking at application needs to have high throughput and very high concurrency.
15:52
Typically learning models require tens, hundreds or even thousands of servers who are all trying to calculate modeling and doing the data at the same exact time, very high throughput required very scalable, very high concurrency when you start to look at analytics and IOT those are areas where there's large data ingest um data analytics are being run against that log analysis, those types of things which require again a large scalable
16:19
platform. When we talk about imaging and medical imaging or enterprise imaging, those are again large areas where they're doing the data collection on the genomic systems or a large report generation. If you think about a company that doesn't end of month reports or building where they're generating thousands or tens of thousands or even hundreds of thousands of reports all at
16:37
the same time. They need that large scalable environment. And then lastly you have like an EDA or manufacturing environment where they're running large lots of automation jobs, they might even be running 10,000 jobs a week in terms of trying to be able to calculate different models. All of these are really useful scalable use cases for a unified fast file and object
17:00
platform. But across all of them and across general enterprise as well is our rapid recovery solution and rapid recovery provides protection for your backup data against ransomware attacks through our safe mode protection. But more importantly, it gives you a platform that is very scalable for the rapid recovery As Corey mentioned earlier, recovering one or two items at a time.
17:25
Might be useful for any platform when you need to recover 50 or 100 or or hundreds of virtual machines or service at the same time. You need a very scalable platform that can fly rapid recovery. So Corey, can you give us a quick update in terms of how this whole uh service comes together? Absolutely thank you. Andy. So one of the, one of the things that's at the
17:49
heart of Evergreen one is our Pure one cloud management platform and the value of pure 11 of the big values of pure one is that every day, every hour? Actually Pure one is aggregating data from across our entire evergreen one fleet about things like operations, logs, lots of sources of unstructured data and it's applying machine learning algorithms to
18:18
produce really smart things like recommendations. So this is an Ai ops driven feature of Pure one that allows customers to go into Pure one and immediately seek specific tailored recommendations to them about how they can optimize performance or cost or both across their entire fleet. And this is incredibly value because it it takes what used to involve lots of hours of a
18:49
technician's time and digging through logs and spreadsheets and all that stuff. And it makes it really smart and it makes it really fast and it's actually even smarter than any individual customer could get themselves because we're able to analyze across a wide variety of data and a wide variety of sources of data. So that's the Ai ops intelligence that's at the heart of Pure One.
19:14
We also help customers optimize where they're putting new workloads where they might want to consider moving workloads around their evergreen one fleet. So that way they can optimize performance and cost and availability. And then finally Pure One is at the heart of our hybrid cloud vision, which allows customers to manage and monitor all of their Evergreen one services wherever
19:39
they may live in public or private clouds. And you know what an interesting little tidbit is that Flash blade actually powers all of those data analytics behind the scenes. So we actually are using our own system in the exact use case that we're describing right here and now with the and now with the power of flash blade as those recommendations are going to get even faster and even smarter.
20:01
Exactly. So to sum up, we hope you've learned a lot here in this session. We hope you're as excited as we are about the wonderful new things that are coming to evergreen one via flash blade. S and we encourage you to go learn more right now by going to pure storage dot com slash evergreen dash one.
20:24
Thank you for your time today. Thank you very much