In this episode of Infrastructure Matters – Insider Edition, Camberley Bates is joined by John Colgrove (Coz), Founder and Chief Visionary at Pure Storage. Their discussion dives into Coz’s vision for Pure Storage and what is next for data storage technology.
Their discussion covers:
- Coz gives insight into the world of storage and what he and Pure Storage set out to change – including the management, upgrades and customer-centric approach
- How the market and technologies are addressing energy and sustainability
- The shift from HDD to SSD and why and how this is shifting. Will this be a similar trajectory of the mainframe and tape?
- Multi-cloud, Generative AI and the future of the Data Storage industry
You can watch the video of our conversation below, and be sure to visit our YouTube Channel and subscribe so you don’t miss an episode.
Listen to the audio here:
Or grab the audio on your streaming platform of choice here:
Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this webcast. The author does not hold any equity positions with any company mentioned in this webcast.
Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.
Transcript:
Camberley Bates: Hi folks. It’s Camberley Bates, VP and practice lead here at Futurum Group with Infrastructure Matters Webcast and Podcast. And I’m very pleased to have an insider edition here with Coz. You may know him better as John Cosgrove. I’ve known him as Coz and John Cosgrove since my Veritas days. Coz is one of the founders and chief visionary officer for Pure Storage. And I’m very, very pleased to have you aboard John. Welcome.
John Cosgrove: Thanks a lot for having me. It’s going to be fun.
Camberley Bates: Great. So let me give you just a little bit background, because some of you may not know all of this history here. Pure Storage was founded in 2009 by John and his partner, John Hayes. This is clearly a situation where one success was not enough in popping out of Veritas and those great days. Pure launched its products in 2011 with a huge amount of acclaim and success. They went public in 2015 when there were about 500 million, couple hundred or a thousand or so employees. Today they’re 2.7 billion, very good, and over 4,000 employees and that’s as of about February earlier this year. So I’m sure we’re going to see some much bigger numbers. We won’t announce because I think as we’re recording this, I think tomorrow they’re doing financial, so we won’t speculate on that.
Today they have two major product lines called FlashArray and Flashlight, and then some others. So we’ll go there. John, or, Coz I should say, I guess when you guys launched, we heard a lot of noise about solid state’s going to replace HDD, et cetera, but you had a bigger vision than that. And I think a lot of people, there’s been more interviews of you over the time, et cetera, especially you’re talking about things that you wanted to fix in the industry. Could you talk about if you fixed those things, what were those things that you tried to fix, and then we’re going to go on from there.
John Cosgrove: Well, I’d say, by the way, I think we’re still fixing them, but no. I’ve been at Veritas as you mentioned, and so I’d been in the storage industry a long time, and I’d seen a lot of things. One of the things I never liked was the cycle. It’s very traditional in tech. You buy something. It ages a little bit. It’s obsolete almost the day you buy it. And so you replace it too soon. It generates a lot of e-waste, which isn’t good, but it generates a lot of overhead and churn. I remember when I was at Veritas, I used to think migrating data from one storage array to another was bad. When I was at Pure, I learned how bad. We actually had one very large customer who described the typical lifecycle of an array as, “Well, we decide we want to buy it. We get it. It takes a few months to get it, get it cabled up, and ready to roll. We spend a year migrating data onto it. Then we use it for a year, and then we start migrating the data off.” And it’s just like you say it like that, it doesn’t even sound rational. So that was one of the big practices, getting rid of data migrations, getting rid of that upgrade cycle. And so our whole, go ahead.
Camberley Bates: Go ahead. Go ahead. No, I’ll come back and say we deal with some big federal agencies on this, and when you deal with big federal agencies, I’m working with RFP, so let’s add two years on the front side of that, so that by the time they get it installed, we’re really talking some really lengthy timeframe. So yeah, huge, huge problem, and very, very painful. But sorry, I didn’t mean to interrupt.
John Cosgrove: Yeah, no, no, no, that’s exactly right. It can be even worse for people who buy on those longer cycles. And so our whole Evergreen architecture and design was designed to get rid of that, to let people stay up-to-date on their arrays as easily as they stay up-to-date on their iPhones or other technology. And that was a huge change. I think another thing that we really wanted to set out to change was enterprise tech did not have a great reputation as being the most consumer friendly. And again, you think about companies you deal with on a personal basis, when they take good care of you, when they provide great service, you want to do business with them longer. And when we started Pure, we said, “Okay, we want to be around for a long time. We want to be a great successful company 20 years, 30 years, long after I’m gone, 50 years from now, not just in the short term.”
And that means providing great customer care, great customer service and satisfaction, and being frankly a good company to do business with. And so really, those were the three things, right? You mentioned the industry transition, all flash. That was one big driver, but the Evergreen architecture, changing the buying cycle and the buying mentality was the second one. And then a third pillar was basically being a great company to do business with, building products that were simple, and easy to use, and being great to do business with.
Camberley Bates: And on that simple and easy to use, one of the things we’ve seen evolved over time is something called Pure1. I believe that’s the interface and the customer success capabilities that you have, and capabilities that you have there. Is that what you were talking about in terms of what you built into the system over time?
John Cosgrove: Well, so that’s one of the things. There’s a lot of things that go into it because it affects every way we think about the system. But consider this. You go on a trip. You go to rent a car. You get in the car, and you drive it, and you do not read a multi-hundred page manual before you do so. You purchase a new phone. You give it to your kids. They know how to use it before you even hand it to them. Okay, you maybe take a little bit longer to use it, but consumer tech, it’s obvious. It’s apparent, and it’s safe enough that there’s no, “Oh, I’m going to push this button, and it’s going to blow up.” Enterprise tech was never designed like that.
When I used to work on mainframes, this is even before Veritas, you had a 10-minute procedure to power it off that was complicated that most people didn’t know because they never did it. And so they’d make mistakes, and in theory, if you made a mistake, you could damage this multi hundred, pardon me, multimillion dollar, not multi hundred thousand, multimillion dollar machine if you powered it off wrong, okay? You take your laptop, and you just hit the power button. You don’t even necessarily hit the power button. Sometimes just close it, and let it go and power off itself, right?
Imagine if enterprise tech was transported to personal tech, and you needed to do something complex, or else you’d break your laptop, right? And it’s just all these assumptions that people had that because they’re working in an enterprise, because they’re in a data center, it’s more controlled that are just bad assumptions. And so we went to change all of those things. And then part of that, to your point with Pure1, which is online monitoring, trying to proactively notice problems with the boxes before the customer notices, providing a central source where a customer can go and look at the status of all of their boxes and see all sorts of things like how many upgrades have they done? What are other people in their industry doing? There’s this feature in there that shows them how are other people adopting features and adopting things in their industry or in the industry at large. And that’s useful because most people, hey, I start using something that meets my needs. As new releases come out, it gets a lot of new features, and they don’t start using the new features.
Camberley Bates: That is a problem. That’s the problem. But with the architecture that you’ve put in place, they’re using those features. One of the areas that I’d like to talk about that you guys have implemented and have led some of the industry in at least visibly talking about it quite a bit, is the ESG space, well or more focused on, I’m going to talk about the environmental situation that we’re looking at. In the environment, you guys have gone and you just put out your latest report on the success that you’ve been having in that space. There is this debate a bit about is HDDs more environmentally friendly than SSDs, et cetera? Can you talk about what you guys are measuring, why you’re measuring that, and how you’re measuring that versus HDD?
John Cosgrove: Sure. Well, number one, the biggest thing that people are focusing on is power, right? It’s the simplest thing. Now, I don’t worry about whether somebody’s using green energy, or the dirtiest coal-fired plants because I can count that they’re all moving towards greener energy over time. But if I’m using green energy, is it better to use a million watts of green energy, or a hundred thousand watts of green energy? The hundred thousand is better. If nothing else, it leaves 900,000 watts for somebody else to use that’s green. So we’re all about reducing power consumption with that. Flash is great for that. A lot of people, they don’t do particularly good comparisons. So for example, when a disk is idle, and a flash drive’s idle, they’re both using pretty minimal power.
Now, when I say, “Hey, I’m going to use this disk,” and I’m going to do some certain workload on it, a lot of people, they then compare flash doing that same percentage workload. But the flash is doing 10 times the IO, so you should compare the flash doing the same workload as disk, which would mean the flash is idle almost all the time. The other thing people build, it’s like you go to buy a car, and everybody’s like, “Oh, here’s how fast it accelerates 0 to 60, and here’s what its maximum speed is.” And I don’t know about you, but I’ve never driven my car anywhere close to its maximum speed, and I seldom actually floor it off, I’m at a stoplight, and I floor it off the line in maximum motor.
You care about what’s your gas mileage, right? Well, in the same token, flash has this interesting characteristic that when you’re writing it, it uses more power. And so when people are out there designing the SSDs that you buy, they want these big hero numbers. So they make them super high-powered so they can get this big maximum, even though people don’t actually use it like that. And the mode it normally runs in uses a ton less energy. So that’s one thing that flash to disk people you don’t compare off spec sheets. You want to compare what people actually use. The other thing is flash has been on this exponential improvement curve that is much steeper than disk. Disk was a fantastic technology for 50 years just getting denser, and cheaper, and faster, and better.
And then it slowed down. It hasn’t been getting faster for 20 years now, and it hasn’t been getting cheaper and denser at the same rate now for a decade about. Flash was you’d buy the biggest disk drive 10 years ago was maybe six terabytes. The biggest flash drive at that time was like one terabyte. Well, now today, okay, the biggest flash drives are a little bit bigger than the biggest disk drives, but the direct flash modules we’re building, which are three times as large as the biggest disk drives, they’re going to get denser at a far greater rate. So in a couple of years, we’ll have flash modules that are 10 times the size of a disk drive. They’ll use the same power they do today. So they’re already more power efficient, and they’re going to get 10 times better.
Camberley Bates: So I think today you’re shipping-
John Cosgrove: A huge difference.
Camberley Bates: Yeah. Today you’re shipping 50 terabyte drives. Is that correct?
John Cosgrove: So today the biggest are effectively 48.
Camberley Bates: 48. Yeah.
John Cosgrove: 75s are coming out in a matter of just six more weeks, and then a year later, we’re going to go to 150, 300 the year after that. And the disk drives, you think about it, they’ve been about 20 terabytes for a couple of years now. Now they’re coming out with 24s and 26s. People are slow to change, but there’s a huge jump in the flash that is not happening in the disk right now.
Camberley Bates: Which is why we may see that overlap or whatever coming over the next five years or so. It’s a slow moving process that’s going on. One of the other things I like about this solid state that goes into that factoring number is I will keep my solid state drive, or should. You should be keeping your solid state drive longer. So maybe I have a 7-year, or a 10-year lifespan of a solid state drive. And many of the vendors like yourselves have got those longer warranties there. And when you have the designs that you have, then I can just swap out my controller, and I get that longevity. So that has to go into the overall impact because then we have the other piece of it, which is the waste side of it, which is difficult to measure really.
We have recycling and all that stuff, so we’re looking at the numbers from all sides of it. And so you’re right. There’s a whole lot of different ways to report on it. We’ve looked at is there a way to standardize reporting? No, because it’s pretty difficult measurement that everybody has to have.
John Cosgrove: It is a difficult measurement, you’re very right. That said, a flash drive weighs one third as much as a disc drive, one quarter, and that is not a bad approximate measure in the long term because when you recycle these things, you don’t reclaim a whole lot, it’s too contaminated with different things, so you don’t really have easy reclaim of too much. But again, when the flash drives are five times the size they are now, 10 times the size they are now, they weigh one quarter as much, so that means, gee, they’re going to be 10 times the size of a disc drive, one quarter of the weight, one-40th of the weight, they’ll last twice as long, make it one-80th the weight. That’s a huge difference. And yeah, we don’t know exactly how it’ll recycle, but that’s a gigantic benefit to get on with.
Camberley Bates: So one of the things we talked about earlier, because both of us date back to that as the mainframe and of course the tape. So we’ve been saying the mainframe is dead, we’ve said the tape is dead. Tape has had this new life to it coming on boar, thank you very much hyperscalers. And the mainframe has got its life of its own as well, especially with Linux going onto the mainframe, it changes things quite a bit. So are we in the same situation as tape and mainframe? Is this still going to exist? And so we’re not really getting rid of anything. We’re just going to have just another layer on here, or do you really think we can get rid of spinning-
John Cosgrove: I think we can get rid of disc.
Camberley Bates: You think so? Okay.
John Cosgrove: Mainframe hangs around because of all the software. Tape, there isn’t a competitor. I can take tapes that were written in the 1970s, and if I have a tape drive, I can actually still read them, the data’s still… some amount of it is recoverable, there’s some of them I won’t be able to read. And that’s something that tape has, that flash and disc don’t really have. If I don’t spin a disc for 50 years, I guarantee you the motor will never come up to speed, the lubricant will have seized up, it’s not working.
And so the tape has that sort of longevity too, completely powered down, put it in a box, stack the boxes in my closet. Okay, really in a mountain somewhere, and get them back 30, 40, 50 years later, as long as I preserve the drives, which are the hard thing. The SSDs, direct flash modules, which we ship, they can be a complete replacement for disc. So once you reach a point where the flash is cheaper than the disc, it performs better, it’s more reliable, it uses less power, generates less e-waste, there’s less maintenance because it’s more reliable and longer life.
The only thing the disc has over it is the cost. And once we get that cost close enough, now you start to say, “Wait, the cost of all these other things means my total cost is better for flash.” That starts to flip. And then as you flip more and more workloads to flash away from disc, you just accelerate the gap between them, and flash will take over completely. And by the way, I think the technology that will replace tape is something like you’ve seen some of these glass etching projects or things like that. Your DNA. I don’t think DNA is happening for a long, long time, but that would be a more suitable replacement for tape because it’s about keeping it for the long term, and about reading it back slowly, not about rapid fire changes.
Camberley Bates: Yeah. And I think the big thing we need to think about as well is understand is that we do move slowly in replacing things. We’ve got five-year cycles that we look at, and since the hyperscalers are probably the biggest consumers of disc drives right now, they’re not going to replace it until things die, or sometimes die in place, if you will, at least on certain levels.
Okay. So thinking about that, let me to another topic, which is multicloud, that’s the other big… two other topics, the multicloud kind of thing. When you talked about reinventing or relooking at the data storage piece of it, on the data cloud piece of it, it’s not talking about necessarily hardware as much, but infrastructure matters in terms of software, in terms of data management and those kinds of things. What are you envisioning right now in terms of how do we operate in this world where data has gravity, so it’s going to go wherever it goes? We think, okay, so it’s going to maybe move twice, and then it’s not going anywhere because it’s too heavy to move. That’s one of our theories of thinking about that. But at the same time, we end up having copies, and changes in data management issues, and geopolitical stuff, etcetera. So what kind of things are you thinking about in terms of what do we need to do to manage in this multicloud environment?
John Cosgrove: Well, I think the biggest thing that people need is agility. If you lock yourself into one solution, you’re likely to have made a mistake or find that you have needs that you can’t do, so agility is the number one thing. If I’m running a hospital, that data can’t leave the hospital. Could you imagine a hospital like, “Oh, sorry, the network’s down, patients are going to die.” Forget it, you can’t do that. And that’s the thing. There’s reasons why data will not all migrate to the cloud. There’s reasons why some data should migrate to the cloud and others shouldn’t.
And the cloud is great, for example, at being not on your premises. So data that you want to get off premises instead of shipping it to a mountain somewhere, ship it to the cloud. And it’s a question of economics and things like that, but there’s reasons for data to go to the cloud, reasons for data to stay on-prem, reasons for data to float around in between. I think the biggest key thing is to have the agility to be able to easily move the data because you mentioned this notion that, “Well, hey, maybe we will create it and move it once or twice, and then it’s locked in.” That’s because there’s a certain cost and effort to moving it.
And so if there’s a value to moving it, as long as the value outweighs the cost, you go and move it. And as an engineer, we like to think, “Well, gee, it’s all about the technology.” No, it’s about the economics. Money drives the world, economic benefit drives the world. So if I can maximize the agility, minimize the cost for the data to flow to where it’s most useful, then I allow people to extract the most value from it because they’re able to say, “This data’s going into an application that’s super bursty. I don’t want to buy all that infrastructure, I’ll move it to the cloud, I’ll burst in the cloud. This data’s much more static, I can control it better. I need it here because I can’t afford to not be here. Okay, I’ll put that there and I can manage it efficiently. And as my needs change, I can move pieces of it or all of it back and forth.” That agility is something you cannot put into words how valuable it is.
Camberley Bates: So what are you guys doing to enable that? Especially like in FlashBlade, which is your file object piece that’s going to have the bulk of the video and the imagery and the kind of stuff that we’re using nowadays.
John Cosgrove: Well, so one of the things we’re doing is we’re trying to put our software stack in the cloud in such a way that you can run on premises and in the cloud the same way. When you think about running stuff on-prem, and you generate most of the data on-prem, let’s be fair, whether you have video cameras there, sensors, algorithms that are producing data monitoring things, you generate most data not in the cloud. And you have a certain infrastructure that processes it. If you have to change to a completely different kind of infrastructure to go to the cloud, that’s a gigantic barrier, I have to change my application stack. It’s a huge barrier.
Why are mainframe still around? Because the software runs in those applications. And so I think the most important thing is to be able to say, “I can go from on-premise into this cloud and not have to change my VMs, my containers, my software packages. I can go from this cloud to this other cloud, and I don’t have to change my VMs, my containers, my software packages. And I need to be able to go from cloud to cloud without actually having to make a bunch of changes. That’s what will enable me to have the maximum agility.”
Camberley Bates: Okay. So let me shift to one other topic, we’ve just got a little time left, generative AI. Say we’re talking a year from now, what’s Pure’s direction with generative AI? Where are you expecting you guys to be next summer, or maybe at Pure Accelerate, which is about nine months away probably. And what do you expect will have happened with generative AI in that time as a visionary kind of person, and that you could tell people that you want them to prepare for?
John Cosgrove: Well, okay, let me start with that last piece first.
Camberley Bates: I know those are big… You and I could probably talk forever, so I’m throwing up some very big topics to you.
John Cosgrove: So on the generative AI, so look, we’re looking at a lot of ways of using it, for example, inside Pure, and that’s a great way for us to think about how customers and others will want to use it. And what we’re finding is it is great at, let’s call it, giving everybody a personal assistant that is kind of junior, a bit of a novice, can sort of do the busy work or the first pass, but you still have to take responsibility for the whole thing. I can send the AI off to go look at every win-loss report we have on all of our deals and give me a whole bunch of analysis and filtering and insights into it, but I then have to go and really look at the work it’s done myself because it can miss very obvious things or make some very naive mistakes.
And I can have the AI do all sorts of work in HR, writing employee development plans for career pathing and analyzing our recruiting efforts and things like that. But again, somebody who really knows what’s going on has to check it and still be responsible. So it’s a productivity enhancer, and I think a lot of people haven’t come to grips with that yet, they think it can do too much. And you see these stories, “Oh, here, I asked the AI to go give me…” A lawyer who asked the AI to go give him a bunch of cases to be cited, and it made up cases.
I’ll give you another example. We were talking about advantages of flash over disc. Out of curiosity, we asked the AI give us 10 advantages of flash over disc. We also asked it give us 10 advantages of disc over flash. We looked at the list, we were like, “Okay, eight of those are definitively wrong.” So you have to check it and you have to learn how to use it right. I couldn’t go and say, “AI, write me a set of software to be a fantastic storage array,” and have it go do it. I can say, “Write me a piece of code that will do this very constrained thing.”
So people have to learn how to use it, that’s the biggest thing. Right now you’re in this hype cycle of it’ll do everything. It really will make people more productive when they use it properly, but we still have a ways to go before it will really produce and finish stuff with no human supervision. I do think it’s going to drive a tremendous amount of need for storage and compute because that’s one of the things that it does. And I think the biggest thing there, again, I will say is flexibility and agility. It’s evolving so rapidly.
I have a friend that has an AI startup that has been around for probably about a year and a half now. And well, when he started this, everybody’s like, “Okay, there’s no way you can possibly do what you’re saying. It’s never going to work. It’s impossible.” And he went and he did a little prototyping and demonstrated to people, “Wow, you’re getting fantastic results.” Well, guess what? Today I look at it and it’s like, “There’s 20 companies doing the same thing you are, and you’re falling behind.” The pace of innovation around this, and the pace of… it’s gigantic.
And again, I’m going to go back to having storage, having infrastructure, compute networking storage that is agile, that can adapt is so critical because the one thing you know is you don’t really know how this is going to be accessed. And so that’s, again, if I go back to FlashBlade for a minute, one of the great things that FlashBlade has done from the beginning, you can do large accesses, small accesses, metadata-intensive accesses, you can be write heavy, you can be read heavy, large IOs, small IOs, it doesn’t matter, it just gives you great performance. And that’s something that’s tremendously valuable in an environment where the way you’re accessing the data in a year or two years will be potentially completely different from what you’re doing today.
Camberley Bates: One of the things I was last… just recently I was at the VMware Explore Conference, and they outlined their intelligent assist offering that they’re going to be bringing out. And especially as the tools that you already have for your clients, I would expect that that’s going to roll out. And watching the demos of what this potentially is doing is like now we’re used to having words fill in the blanks as you’re typing, or Google fill in the blanks as you’re typing, you just click it and go.
And the similar kind of stuff because you guys, you’ve gathered so much data that you have within your database that the people are bringing up there, so it’s the best practices are the best people to be able to implement on it. So it’s going to be an exciting next year, I think, as we will see individual generative AI language models coming out that are taking on these core little pieces there that make our lives really totally easier to manage in the world. So we’ll be very excited to see how you guys roll out with that in that timeframe. So any last words for everybody?
John Cosgrove: Just to emphasize something you just said? I think it’s not so much even an exciting next year, I think it’s an exciting next 20 years. Ever since the advent of the, let’s call it the smartphone, I’m going to use that as the drawing line, we’ve carried around with us basically access to the entire world’s intelligence and information. And I think we’re really only 10 to 15 years into what’s going to be a 30 or 40-year journey that completely changes the way things are done.
And those of us that are a little older might be a little worried about some of the changes. It’s the way the world works, things change, and you move forward, and it’s going to be really exciting. I’m looking forward to the day when I don’t have to drive, I can sit in the car and trust it, and my car can fly too, by the way.
Camberley Bates: I don’t know about that because my mom’s 97 years old, and she’s got her iPhone, it frustrates the heck out of her. She can barely use it as a phone, but she can talk to it and have it text, and that is pretty cool.
John Cosgrove: That is. And I will tell you, every time I get in my car, my phone tells me, “Well, it’s going to take you this long to go somewhere and here’s the good route to take.” And I pretty much ignore it because I know where I’m driving around here, but it’s accurate a lot. And it’s not like I told it, I’m going to get in the car and drive here now, it’s got a pretty good guess. And I see those guesses getting better.
One of the things I do is I follow these algorithms, I study them, whether it’s the algorithms for how the elevators in the building run or for when I put my car into the auto drive mode, how it’s going and where it makes mistakes. And they’re getting better at a fantastic rate. And it’s just going to be really cool to see what it can do for you. And if you embrace it’ll make your life better.
Camberley Bates: Yeah, it sure will. So Coz, thank you very, very much for joining me, it’s really been great to talk to you and catch up with you. And maybe we’ll do this again at Accelerate and kind of like saying, “Okay, so what did we talk about then and where we are now?” So if we’re going to be moving that far fast, and so-
John Cosgrove: Look forward to chatting with you again.
Camberley Bates: Absolutely. And folks, thank you very much for tuning into Infrastructure Matters. And don’t forget to push the Follow button that’s down there, and we’ll put some links into some of the things that we’ve talked about in here as well, especially on the latest report that Pure has sent out on their ESG reporting. So thank you again, Coz.
Author Information
Camberley brings over 25 years of executive experience leading sales and marketing teams at Fortune 500 firms. Before joining The Futurum Group, she led the Evaluator Group, an information technology analyst firm as Managing Director.
Her career has spanned all elements of sales and marketing including a 360-degree view of addressing challenges and delivering solutions was achieved from crossing the boundary of sales and channel engagement with large enterprise vendors and her own 100-person IT services firm.
Camberley has provided Global 250 startups with go-to-market strategies, creating a new market category “MAID” as Vice President of Marketing at COPAN and led a worldwide marketing team including channels as a VP at VERITAS. At GE Access, a $2B distribution company, she served as VP of a new division and succeeded in growing the company from $14 to $500 million and built a successful 100-person IT services firm. Camberley began her career at IBM in sales and management.
She holds a Bachelor of Science in International Business from California State University – Long Beach and executive certificates from Wellesley and Wharton School of Business.
