00:00
Push any Evergreen One customers out there while we're waiting here. Show of hands, 12, any future Evergreen wine customers out there. Just wait. OK. I think hopefully this is gonna get us going here. I'll tell you who I am while we're waiting for the slides to come up because you're like, what is this woman doing?
00:25
I don't work at Pure, I work at a company called Tier Point and there'll be some slides about who we are. I am not one of you. I'm not a technologist. I'm like, I feel like I'm an elk in a herd of buffalo or something like you guys could take me down at any minute.
00:40
Um I always tell people I'm not in it but I'm it adjacent, right? I've worked in technology procurement for 30 years. So I'm gonna talk about pure as a s I'm sorry. Pure Evergreen one we call it now um from a business perspective because mostly when guys like you need to go spend money on cool new storage things you got to get through people
01:03
like me to do it. So I'm here to help you with that. I did discover this morning however, that with pure new co-pilot tool for natural language A I I too am now a storage engineer. So that was pretty exciting for me. Awesome. Hey, thank you everybody for, for showing up for this. Um I'm gonna talk today about SL A SSL OS
01:30
marketing stuff that's sometimes sounds a little dry stuff that may require you to do a little bit of reading, put on your legal hat. But as we shift to the Sass world or storage as a service world, obviously, you don't own the gear anymore. You don't own necessarily the support contract.
01:51
And the best way to measure storage as a service is through Slas, that's essentially what you're buying, right? And understanding how to kind of see through the facts and the BS with different vendors and Slas um will make you a much smarter buyer, right? As we move into this world. So we jumped around a little bit. Uh My name's Jonas Irwin.
02:15
Uh sorry for the technical difficulties in the beginning, I've been with pure since 2013. I'm employee 120 spent a bunch of years out in the field, working with customers up and down the west was the principal for the West. Um And then I basically built the Evergreen. One business ran that business with a team, a great team around me.
02:36
Over the last 3.5, 4 years, I built all the slas so I can kind of speak to this pretty, pretty credibly. Um And uh uh now I'm in the office at CTO, I work for cause now uh as a platform strategist, lead sort of lead um platform strategist in that in the CTO office. So really cool getting to work with causes.
02:57
Um So uh Linda gave, I wanna say thank you to, you know, our customers make the world go around. I almost wore the exact same thing cos had on, but fortunately changed. Yeah. So agenda today, I'm gonna go through stas a little bit, I'm gonna, I'm gonna try not to be patronizing and like, say, you know, give you all the terms and then all the definitions.
03:21
I'm assuming a lot of you have that kind of general grounding. I wanna talk about architectural essentials. Um And what we, what I was asked to do when we built Slas at for Evergreen one specifically. All right, I'm gonna kind of talk through that.
03:39
Um And then we're gonna take a look at what we've created with Evergreen one or 10 slas that we have and we've done it the pure way, which means we didn't wanna just create a promo and you know, say, oh we can do X and then force customers to read the fine print and really find out, oh, there's really nothing behind this. This doesn't, doesn't really mean anything, this isn't valuable to my business.
04:03
We wanted to do it as I'm calling it the pure way. And then Linda is gonna tell us um all about being an evergreen one customer to reports a large MSP MSP of the year 2024. And um I asked Linda to just say, you know, th this world of, of uh of storage and service is kind of nebulous kind of new kind of different talk
04:33
about what it was like before and you still have other vendors, right? But what it was like managing in the old model and then what it was like with Evergreen one now. So it is connected to Slas, but she's gonna really more talk about what that transformation looked like and then we'll close it out, maybe have some time for Q A.
04:50
Um OK. Going back when II I did leave pure for a small period of time, I went to Azure to lead their solutions architects in the West for a while. But then I realized I missed pure and pure asked me to come back and I said, yes, I will absolutely come back. But I came back to build up the Evergreen one business service architecture, solution architecture, the design.
05:15
And at that time, it was kind of a utility offering kind of a pay per use. And this study came out from uh Gartner that said by 2025 40% of all storage purchased will be storage as a service. And I looked at that and I thought we have this Ragtag Little team in the company, kind of, kind of a,
05:38
a small little group in the company, a small percentage of, of how we worked with and sold the customers. And I thought that's insane. Like there's no way customers are going, are going to be almost half of what they consume as subscription as da I don't see it. Well, it's funny is if you fast forward three years, we've had tremendous growth uh due in
06:00
part to the work of a bunch of phenomenal um people, consultants, like a back there in the room that we have in the field, great engineering, great support, great executive focus and, and, and support. But now it's 42. It, well, actually you ending in 2024 it was 42% of all of our revenue subscription in total,
06:21
right? And, and we're trending very quickly toward half of our revenue. That's a very fast transformation. And it is, it's interesting, right? We beat, we beat that target of 2025 at least that pure. So I'll let you take that picture and then I'm
06:36
gonna scoop to the next slide. I I'm gonna just get, let me get through this, but I will definitely uh call on you. We just have a certain turf we need to cover, we need to get across and then we'll be here. So uh storage as a service, what is a vendor like pure or any of the other players out there? I don't wanna pick on any specific vendor
06:59
although I kind of do, but I won't do that here um For them to, for them to deliver a good storage service offering. It's really about three fundamental areas, right? Operations would be from, from the vendor side would be um can I easily deliver an as a service offering? Can do I have the professional services?
07:26
Do I have the field teams to be able to do it if we're running out of capacity or performance on a license? Can I get capacity to that environment? Can I scale it? Um Is that, is that a natural part of my, of my business of, of my, the way that we interact with customers, the way that we support customers and then
07:44
jumping to the other side commercials like yes, you can package something for a pay per use. Um You know how flexible it is? Is it, are you going to enable a customer to do um to do uh mid contract expansions in a co term manner or you're gonna force them to sign up for a whole another three years? How dynamically can do you do that? And at the core of that is architecture,
08:14
if the architecture you have is flexible and it'll enable you to scale without downtime. If it, it'll enable you to deliver deterministic performance, it'll enable you to maybe never have to do another migration, another data migration again, that enables tremendous flexibility if you don't have an architecture like that and you don't have operations like that.
08:40
Right. Let's say your architecture is 1/20 century disc architecture that's been retrofitted forward. 30 years, an operating system that's 30 maybe even 40 years old. It's gonna be very unnatural, right? But the best way to figure out what these three elements can deliver from a vendor is to look for.
09:02
Are they offering slas, what do those slas look like? Right. And, and, and I'm sure all of you have seen various versions of these slides. But what I was asked to do is look at all of the things that we talk about pure talks about. Look, you never have to take down time.
09:22
You can do intergenerational, extra generational transitions of our technology. I have a very well known coffee shop that started with this system. I put it in, in 2013, we consolidated two mi arrays turned it into that. A few more years later, we turned it into that and so on, right? Until we collapsed.
09:41
Actually, all of their storage infrastructure, a bunch of monolithic disc arrays, they never paid for the same IO more than once. They never did. It had to do a data migration and we did all of the all of these transitions at 10 o'clock in the afternoon on a Tuesday. Um No downtime, no impact, completely imperceptible to business.
10:03
So I know these things I know we can deliver, but pure, makes a lot of claims. We're the most efficient. We can get you 678 nines. Right. So, what I, what I was looking to do in this project was look at the marketing claims that we make and then could I build an sl a out of all of the value props that we have?
10:25
All right. Could we, could we guarantee six nines with no downtime? Could we guarantee that you never have to do a migration ever again? All right. So what did I have to do? I had to look at our fleets which I'll talk about in a minute.
10:41
But it's really been a journey for us over the course of, I guess it's been almost eight years, eight years now. Um We started with the Evergreen storage service, which was basically Evergreen. Uh one V one which I'm saying, which I'm calling stas 1.0 we renamed it briefly to pure as a service, right? Maybe not the best name because Pas was,
11:06
it was the acronym. So that caused some confusion. But in, in 2.0 we delivered a service catalog where you'd have a variety of different price to performance choices that you could choose from initially. It was just here, here's just block storage or here's file storage.
11:21
We found that, that, that wasn't good enough, but we still actually didn't have Slas until Stas three or Evergreen one. That was when we delivered five slas which I'll talk a little more about, but the point is as our platforms have grown, as our operations have matured, right? Um And as we've added more features and capabilities in our platforms,
11:45
the utility service Evergreen one is getting better and better and better. And what we're doing is expressing those value props in the form of slas, right? So this is the part that I just highly recommend everybody who's looking at storage as a service offerings. Take this time to say like, yeah, II, I like the marketing,
12:06
I like the claims you're making. I like your value prop to the vendor, but show me the actual contract, show me what your organization will commit to um and force yourself to read the the legal terms. I've read them all, I've read all of ours. I've, I've helped create all of ours, right? With, with, with council.
12:27
Um But I've also read our competitors and it's really interesting, that's the fastest way to figure out how confident is a certain vendor in the claims they make, how, how confident are they when they actually go to create a service? Because if there's a disparity, if I say seven nines over here and I say four nines in my terms, right? That's actually the four nines is the real
12:53
confidence level, right? So the goal is you, you really wanna find, you know, obviously what you need for your requirements, but you really wanna read those documents and make sure, don't wait to get angry to read legal terms. Uh So if we look at the, the three areas I made sl a the biggest because I'd say um
13:15
it's the most important. But marketing, like I said, you have sort of best case scenario positioning. You see a lot of claims like 200% better performance. They may not say 200% better than you're not even sure. Versus what is it? The last gen is the competitor or they might
13:33
say 20 million I OS, it's, it's basically the sun, the stars and the moon all lining up perfectly. This is what I can get if everything is perfect. And then on the other side, you have service level objectives, which is really an aspirational target. We have service level objectives for evergreen one, like we guarantee six nines with an SL A but we're trying to hit 100% right?
13:56
We wouldn't make those equal because that leaves us no room. And then from an sl a standpoint, the thing is that I would recommend folks look for when it comes to SL A is make sure it's legally binding. Um Make sure you understand across those three areas of architecture, operations and commercials.
14:16
Can that company do this? Can they actually deliver and support the SL A they claiming they can, they can deliver? Um Is the are the terms transparent and easy to understand or do you have to get an attorney and another specialist. And are they nebulously drawn up? Because if they are, that's gonna create some issues down the line because you're gonna have
14:37
one idea in your head of what you think you bought the vendor has another idea and there may be a lot of asymmetry there and, and right, that's fertile ground for uh for conflict later. Uh And then most importantly I'd say is are they willing to back it up with service credits up to 100% of your bill? So for your monthly service credit, like if you miss your availability is the vendor putting some skin in the game.
15:06
And when I say up to 100% service credit, I mean service credits that you can actually use, not a percentage of a percent that you might use in a renewal something you could use for the following month to burn down your service. So uh in terms of your your meter, right? Something you can you can pay for. Can they remediate?
15:24
And uh do they have slis, do you any everybody know what slis are here? Service level indicator? So if I have an sl A, I'm measuring it. The sl I is just a simple indicator. I can go and look at where am I against my SLSLA A? Am I am I meeting?
15:43
Am I short? Am I over? Am I under? Right? When we talk about transparency, the sl I is the best way to, to figure that out. Uh So jumping in a little more, what were the keys? So when I was asked to go build these,
15:59
right? The the genesis of this was we were having conversations with the CIO S CFO S. They were like, yeah, I know that you guys have a great technology. My team, you know, my teams tell me they want, they want pure, we're paying per use.
16:14
But as I looked through your contract and granted, this was a number of years ago, this was like, as we were growing the service, I don't see any hard commitments in here. There's no everything sort of we're gonna try to do this. And he's like, but if you don't, what happens like and, and then he was like, I remember this one meeting the cio said uh
16:38
there's no sls in here like I don't even know what I'm signing up for, right? And that's when Prakash kind of came to me and went, we need to, we really need to get our act together here. We want sl A for everything, right? Which is one of the reasons why we have 10 now. But so when I went to build them, I thought, OK, the first thing I want is I want it to be pure,
16:56
like which is simple, transparent, easy to understand. You shouldn't have to be a lawyer to figure it out. I wanna create a broad spectrum of coverage. So it's not just availability, you can do performance, um anything from efficiency, right? How do, how do we gauge uh how much power rack
17:15
space we take per T uh migration so on and so forth. We wanted to back it up with 100% service credit which really shows our customers and partners. We're very serious about this, about creating these outcomes. If we can't hit the outcome, we don't get paid. Right? So I would tell you as I was doing this,
17:36
this project, um I need to be really careful because if I over promised on what we could do, why I, I probably wouldn't have a job anymore. Right? Because we would have a ton of people flooding in filing for the service credits. So we needed to find just the right level of commitment that we could actually deliver on
17:58
because if we didn't, we don't get paid. Um The next piece was the sli I knew for every sla I wanted to aspirationally deliver an sl I, so that customers could go in and say I bought X. How can I see how I'm doing against it and make it as simple as possible? Like, like, uh you know, uh a speedometer, you can see exactly right.
18:20
And, and while we don't have slas for all of them yet, that's the direction that, that we're working. And then finally remediation operationally, I needed to make sure that we could remediate. Right. Like if you're, let's just say you're not hitting the watts per terabyte metric. We're pulling too many watts,
18:40
right? And, and, and we said we would do three watts per terabyte. We can see you're doing five, you can see that in our console. But we need, I needed to make sure operationally that we could go in there and figure it out and get that wattage down. That means in all regions in the world where
18:59
the service operates. So it's not just technically, can we deliver, it's also operationally like can can we get to you, can we um reduce the like increase the density of the configuration online, right? Which we actually haven't had to do yet, but we need, I need to make sure we could. So I went in,
19:17
I looked at over 20,000 systems over the course of seven years, right? That coffee company example I gave you that was one of the systems and analyze and you know, said, how many upgrades of purity have we done? What's our success rate? Right. How many hardware upgrades have we done in generation out of generation?
19:37
Have we needed a maintenance window or not? Right. What is the actual reality? And I built a model, right? Like a bunch of spreadsheets, uh a bunch of probabilities and things like that. Uh same thing with performance.
19:50
What is our data durability, right? You know, Aws always says 12 nines of durability like, but what's ours? So that required a lot of digging, a lot of working with engineering and it kind of coalesced all back into. Ok. Here's what I think we can, we can commit to for the service um for the for the different
20:08
areas following what I said earlier, the transparency, the coverage and so forth. Um And you know, we kind of move forward, created evergreen one which Prakash sort of touched on earlier today. And every single service offering that we have with an evergreen one comes with this first block of slas out of the gate six nines of availability.
20:33
All right. Uh Performance measured in bandwidth per tab byte and IO per tab byte. It's kind of weird sounding metrics, right? But these are, those are examples of slis, that's how service is measured. Those were the way I decided to do it. Other vendors have followed using the same
20:50
metrics, but we need to make those metrics available to you, right? For Slis, which I'll show you energy efficiency. We're the first vendor to guarantee a certain not to exceed watts per terabyte number. Um because we know we're, we know we're the dentist, right? And the, and the um the most energy efficient platform we're proving it right.
21:12
And we've said to the world, here's this, you know, I've said to the world, here's the standard, here's how you measure it. What's yours like? Tell us yours. Um No migrations if you have to do a migration and you're on our service. I'm talking 10 years. Plus you ever, if we ever require you to do a
21:28
migration, you will not pay for an entire year of service because everything we engineer at pure has to make those non disruptive transitions. We build that into every single thing that we do at the company. Um If you ever lose data, right? Like I could keep going on, but those are sort of the core and then we added some additional ones that have come at a
21:50
slightly higher unit rate. So you'd pay a little more per terabyte per month to have these three capabilities. Cyber resilience has been a really hot one that was called the ransomware recovery uh offering, which means if you ever get hit, we drop ship you a fresh clean R system, we help you with the recovery process, right? With our um our our ransomware uh locking
22:14
capabilities. So no one can actually delete the data, but we help you put Humpty Dumpty back together following all the security best practices, clean room, right? And we'll drop ship that system to you immediately. If freight isn't there on the weekend, it might take an extra day. But if it's during the week,
22:30
those systems should be there the following day. We build a recovery plan with your company ahead of time. We include 80 hours of PS so that if the proverbial, you know what hits the fan, we actually know what to do, right? We know your environment and we know what to execute. So it's gonna be a slightly higher rate,
22:47
but I wanted to sort of point those out. Um And then maybe let's take a look using that rubric of transparency and the Slis and all the, all the stuff I talked about if we look at our competitors, um what do they look like? Which I'll get to you in, in one more slide. But I wanted to show you guys an an example of Slis.
23:07
So this because remember we're providing the infrastructure, you're signing up for a license, a subscription license. It's a little more abstracted, but that's the thing you purchase and the license has the SOAS and in pure one, right? Our global cloud driven dashboard with over 30 petabytes of analytics data. Every log e everything that's ever happened in
23:29
a pure array is all in this giant brain. But what we do is we report in um into one central area. Are I, am I hitting my performance? Am I hitting my bandwidth? Am I, am I getting my energy efficiency? We're still working on the availability indicator which has been a little tricky
23:46
because sometimes customers move arrays and there's reasons, right, why they may, they may require an outage. So we're still kind of tuning that, but you can go in immediately and see is there a problem with my service? Did I buy the wrong one? Is it too big. Is it too small?
24:03
And then what you can do over time is you can look at your consumption model and see how you're doing. Uh your commit here is uh almost 18 petabytes for this example. And what we do with evergreen one is we build in a 25% buffer because customers over buy storage and they never, they never wanna fill an array up beyond say 75 80%.
24:28
People start getting nervous that extra buffer is built in, but you don't pay for it, but it's built into the base unit rate. And it's our problem to make sure that that buffer is always there. That is one of our slas that 25 now you can dig into it, you can use it. It's always there at a slightly higher on
24:46
demand rate and you know, Linda might talk about that but they're, they're, you know, ter point is a big fan of that, of that feature, but that becomes our problem and you don't have to worry about it. You see that you can report on it. Here's the uh sl I of bandwidth per tab byte being reported in. You can see total throughput here is 28
25:07
terabytes uh plus like almost 28 terabytes of total aggregate bandwidth and then read plus. Right. Right. The this customer I think the sl I here is about five megabytes per T and they're well below it. We provide a few other metrics like load. If you're below your bandwidth limit, just as an example or your bandwidth sl A um or your,
25:30
your yeah, bandwidth sl A, you would go look at latency and you would look at load if those were relatively low, it just means you're not pushing enough workload to get to the max. Um But if you were, you would see load is that if we weren't delivering, you'd see load is at 100% latencies through the stratosphere and you're under what we said. But the point is we make it clear, it's not up to you to have to manually calculate it and go
25:55
figure it out in a spreadsheet somewhere. And we even do that with the energy consumption. We pull energy consumption. Um You know, the sl A here, the way I built them, it's like it would be set to two. But most of the time based on the, you know, the way that our platforms deliver, the way that we deliver, there's plenty of headroom in there,
26:17
plenty of headroom, right? And this customer is less than a wad of team. So cool. I'm in a give you a quick competitive view. OK. So this is looking at my three major competitors in the market using the rubric, the measuring stick that I talked about earlier is the claim that's being made around performance availability,
26:42
energy efficiency. Is it in the contract? And is it in sl a is it easy to figure out? Is it transparent? Is there an sl I and so forth? And when I look across all of this, the these three players? Anyway, the answer is no, none of them actually have real slas and I would submit to you that
27:02
customers partners, they deserve better, right? They deserve that. Uh Vendors take a, put a little more skin in the game and commit to actually what outcomes they can create, especially as we move into the, as a service world. Um And uh uh you know, frankly I'm I'm a little surprised but I would encourage you all
27:25
to kind of do that research if this is something that you're looking at either for yourself or for your the customers that you work with, if you're a partner. Um and hopefully you, you got, you all found that somewhat useful. Uh and we're gonna be here after, but I wanna turn it over to one of my
27:46
favorite customers, Ka uh Kathy to talk about um to your point and their experience with Evergreen one. Alright, thanks a lot, Jonas. I appreciate it. So I told you that I am not in it. I'm just it adjacent. There will be no layer three IOPS discussion in this presentation.
28:09
We're gonna talk about the business case of moving from traditional to pure evergreen one first, a little bit about tier point. We are a hybrid it services organization that is 41 data center strong. We have hundreds of it professionals and some it adjacent professionals and we have thousands of customers. What do we do? Well, we do all the clouds, all of them,
28:35
public, private multi-tenant cirrus, cumulus, nimbus, all the clouds. We are security focused. We are advisory and consulting services across the footprint of what we do. And that includes but is not limited to disaster recovery colo location including high density and manage services down to even a platform and an application level.
28:58
We're one of the sponsors of Pure Accelerate. So if you haven't already come by and see us in the exhibition hall or visit our website or call me. But if you mention anything about layer three, I will tell you you have a wrong number and I will hang up on you. So I want to talk about the before my, my presentation is two slides.
29:17
It's the before and the after, but before I talk about the before I just want to level set something we're talking about the traditional way that we've all bought it for decades, right? This is not this before, is not specific to tear point. It's not specific to storage even. It is just the way we buy it over the last. I don't know how many years I've been in tech
29:43
procurement for over 30. So at least that long, you buy the stuff, you put it on a maintenance plan, you upgrade or expand it you have to eventually refresh it. You have to do a migration, then you dispose of the old stuff. And then when you're done with that, you start over, it's like laundry. You just,
29:59
there's always another load waiting when you finish this one. Now, that takes a lot of work from my team. Um, but we're not the only one. So your teams also the, the putter inner, the installers, the configuration, the managers, the capacity managers, the uh migrator that I don't, I can't even make up other titles that
30:20
might be involved here. But there's a lot of work that goes into this lather rinse repeat cycle that we've all done. We call it life cycle management. It's an industry issue and I think at tier point, we do it really well. My team are, we are great stewards of managing tier point spend because we want to make sure that our customers know that they're getting value for their money when they trust
30:45
us with their it solutions, right? But just because we're really great at doing it, the traditional way doesn't mean that that's the best way. So let's talk a little bit about the traditional way. The traditional way means that we have this life.
31:00
We're in constant life cycle management. We're always in this cycle somewhere in the cycle and you may have multiple ones going at the same time. The traditional way means that maintenance costs increase over time, the more your equipment ages the more it costs to maintain it. Jonas referred to that as maintenance extortion. And I've never heard that term before,
31:21
but I love that term. If you're involved in doing maintenance renewals, that's what it feels like you're sitting here in June doing a renewal that's due at the end of July. Say, who knows that vendor knows that you're not going to be able, you have no choice but to do that renewal. If you were going to switch, you would already be most of the way through that project.
31:41
So it does feel like extortion and the pricing is so inconsistent. I've had situations where I've had the exact same make and model controller in two different locations and the pricing was vast different to do the renewal and nobody can explain it. It happens organically over time because the their systems change and whatever it makes it very hard for a company like ours or yours to budget, right?
32:07
And if you can't budget, well, that means that you're going to the person that holds the purse strings and asking for money that wasn't in your budget. Maybe you're asking for Capex to do an expansion that you didn't expect. Maybe you're asking for Opex because the maintenance was more than you thought it was. You're also buying your own buffer, we all buy our own buffer.
32:30
In the traditional model. You, you, I don't know what your threshold is. If it's 80% or 90%. But somewhere back in that data center, an alarm goes off and somebody's like, we gotta get a new shelf and then part of that cycle begins and order the new shelf and get quotes and get it in and install it. Right. That buying your own buffer, you, you're at 80%
32:49
percent or 90%. You're not even using 100% of what you already bought and paid for a year or two ago. And now you're going to go by 20% more so that you can have more stuff that you're gonna pay for now that you're not going to use until sometime in the future. So all of that is part and parcel to what traditional it looks like in storage and even in other areas um enter
33:13
the new world, which is essentially the corollary of what I just told you, right? We didn't just fall into this with pure, we did a lot of work, you know, II I run the procurement group so you can bet there was an RFP and a lot of people at Pierpoint participated in that RFP and it took months pure, had to compete for it on every level and they won it on technology and on the
33:38
economics, it was a, they won it pure and simple, no pun intended. And so once we have that in place, what I could do is go backwards and say we no longer have inconsistent pricing. We no longer have budget challenges. We no longer have to buy our own buffer. We no longer have this lather rinse repeat cycle that we're constantly in over and over and over again.
34:04
We have solved a lot of those issues with a radically new model. Um That makes sense and it's anybody who does storage as a service can tell you. Yeah, we can do that for you, but it has to check all the boxes. And I know Jonas talked about the SL A in my experience when an SL A is contractual, they're not gonna something in the contract that they don't feel confident that they can
34:29
stand behind. As he mentioned. If he did that, he would have lost his job. So having a good sl a having sound economics, having flexibility and simplicity all have to be in place for this model to work. And that's, that's what got us over the finish line at tier point with our partnership with pure and we've
34:48
grown by 300% and we've had a long standing relationship with pure. But in the last three years, we've grown by 300%. So that's a win for both of us. Thank you for your time. That's all I had.
