Patrick Smith, Pure Storage Field CTO for EMEA, on a sustainable, agile, and evergreen future for data storage

Speaker A: Hello and welcome to CEO Extra, uh, a very special episode of the CEO Digital Show. Today we're going to be thinking about the data store of tomorrow with a very special guest, Patrick Smith of Pure Storage.

Speaker B: I am Patrick Smith. I'm the field CTO at Pure Storage, covering the EMEA region.

Speaker A: With high level tenure at Goldman Sachs, Credit Suisse and Deutsche Bank, Patrik is a vastly experienced expert in both technology and financial services. Drawing on the wide breadth of his insights, we're going to be discussing the not too distant future of enterprise data, uh, and how it's set to play an increasingly pivotal role in the next generation of innovation and sustainability. And so Patrick, welcome to CEO Extra. Thanks for joining us. So Patrick, your financial services expertise is um, incredibly impressive. Uh, I was just wondering what inspired you to make the switch from finance to data storage.

Speaker B: Yeah Todd, it's a good question. I spent 25 years in financial services and within most of my experience was looking at infrastructure engineering and looking at the futures and driving change within technology. And so in my last role, one of the big uh, change projects was to introduce flash based storage technology. And so I became aware of Pure, worked with Pure and had a good understanding of Pure as a company, uh, the technology and the people and culture and found that actually there was a really interesting roadmap and a really interesting future for Pure. So the opportunity to join as field CTO for the EMEA region came up and um, it was one that I leapt at.

Speaker A: What an interesting, um, divergence from what might be the more traditional path. Um, do you find that your finance expertise, uh, lends you, um, any extra benefits in your role as field cto?

Speaker B: There are a couple of areas where I think that helps. Firstly, I was previously the customer, um, and so I have that customer perspective. Um, hopefully I understand what's on the mind of a potential customer of Pures, um, and can help put at ease any concerns that they have about. Typically if you're looking at, at um, embarking on a partnership with a new vendor, you're going to have some concerns. So I've been through that process, so I have that sort of previous um, experience which I think really helps. But I think one of the other things that especially coming from financial services is an understanding of how important technology is to the business. Um, you know, my early days in financial services were spent sitting on a trading floor supporting traders. Um, and that, that really makes you realize how critically dependent they are on technology to do their job. And uh, so having that awareness really makes you realize that technology Needs to be, you know, available, always work and supports critical business services. So you are coming from financial services. I have that sort of firsthand experience.

Speaker A: What a great asset to have that sort of sheep dip of experience that really helps you gain the extra insight that you need. I can see how that would be really, really useful. Um, so, uh, one of the areas that we've been talking about quite a lot on the podcast recently is sustainability, which is one that I know that you yourself care about and that Pure care about as well. Um, so, uh, moving us on to that subject more broadly, I was wondering, from an environmental and business perspective, what would you say isn't with the old way of doing data storage?

Speaker B: I think this is a really interesting area. One of the obvious aspects of technology and storage is this idea of built in obsolescence. There are very few technology platforms that you run for three, four, five years or more. They become obsolescent, they become too slow. Um, and the vendors of those products encourage that because you're going to get rid of the old product by the new version. And so we're in this perpetual cycle of buying a piece of technology, running it for a certain amount of time and then getting rid of it. If it's my iPhone, then it gets passed down to my daughter who'll be very happy to have a newer version of an iPhone. But, but typically if you're looking at business, that doesn't happen, that reuse of technology doesn't happen. And vendors are even more extreme, uh, in the storage space. There is this rip and replace approach to storage technology where you move, um, all of your data from an old system to a new system. And the manufacturers of storage hardware typically encourage that by increasing the cost of maintenance to encourage somebody to buy the latest and greatest storage system. So that's how the industry's been for the last 20 plus 25 years. And what I think is really interesting is a company like Pure coming along and saying it doesn't have to be that way.

Speaker A: It's such an exciting idea, this concept of um, being never obsolete. Um, you know, this sort of, the wastefulness of obsolescence is really, really becoming apparent. Um, you picture landfills full of old mobile phones full of, uh, defunct media, CDs, VHS, things like that. Um, and why wouldn't it apply to all kinds of software, hardware, all of these projects? Um, so it makes a lot of sense to me and it's a really, really exciting idea. I'm going to jump ahead a little bit because I'm so interested in Being never obsolete. Um, what does pure's never obsolete offering mean for enterprises looking towards longer term data solutions?

Speaker B: Yes, so this is really interesting. One of the biggest challenges if you look at data centers today, um, and it's been the case for many years, is life cycle management is really difficult. I describe it in terms of it's very easy to put something new into the data center, but it's very difficult to take something old out of the data center. And so that means that lifecycle management is, to be honest, it's a pain. And so anything that can be done to simplify lifecycle management is a huge benefit to organizations. And so what we've done is introduce a concept called Evergreen. Ah, it was built into the platform from the very beginning, um, which actually is important because it's the kind of concept that can't be retrofitted. You have to build a system to support an evergreen concept, you know, a forever modern concept, right from the beginning. And so it means that we can do upgrades of the important components of our system, the intelligent bits. The bit that gets better, the bit that follows Moore's Law, we can upgrade those components independently of the wider system, which includes the chassis, the power, uh, supplies, the cooling fans and the flash storage that a customer's data sits on. And so you're only replacing the intelligent bits. The other components have a much longer life cycle because they're not evolving as quickly as Moore's Law. And so it means that we firstly can do that in place in the data center. To secondly we can do it without any downtime and uh, actually without any performance impact. Even when you're midway through the upgrade process, there's no performance impact. And I think the other thing that's really important about it is it reduces risk for our customers because when you're doing a forklift migration from an old storage array to a new storage array, there's lots of change activity that goes on. You have to wheel the new array into the data center, you have to connect it up to your network and your power. Um, so lots of facilities changes, then you have all the changes associated with migrating data from one system to another. And these are all have administrative overhead in terms of having to get a change request approved, having to get a change window sorted out and we eliminate that.

Speaker A: Um, what a really innovative concept. It's genuinely very, very impressive. Just to very quickly touch on something there, you mentioned Moore's Law. For any of our listeners familiar, would you mind, uh, explaining what Moore's Law is?

Speaker B: Yeah, Moore's Law, um, Moore's Law is uh, the concept that came out of intel that the processing power, or probably more accurately than the number of transistors on a piece of silicon would double every uh, over a certain period of time. And we've followed Moore's Law very uh, well for a long time. Although there were recent comments that Moore's Law has ended and now Nvidia are the ones that are uh, really driving the number of transistors on a piece of silicon. So um, yeah, um, it's the increase in performance of processors over time.

Speaker A: I mean things are certainly changing at a very staggering pace for a lot of people, which I know, um, so many thinkers, uh, and thought leaders are really struggling to keep up with the pace of change. And it takes a lot to kind of admit that and to admit what you don't know. And sometimes we don't know what we, we don't know. Um, thinking ahead to the future and the future of data storage, our topic for today, um, if you could look into your crystal ball, what do you think the data landscape looks like in sort of 10 years from now or even 20 years?

Speaker B: I think that's a really interesting question because lots of people who look into the future also are uh, historians. And if you look back 10 years the amount of change we've seen has been absolutely phenomenal. Um, ten years ago pure had been shipping products for two years. Um, um, ten years ago we only had one product. Now we have a portfolio, um, that forms a platform for uh, all data usage. And so it gets interesting as you look forward 10 years, certainly a few things that we expect to happen. Firstly over the next 10 year period we would expect that the flash technology that we use to store data will replace in its entirety the use of mechanical hard disk drives. We've seen over the last several years the increasing capacity of flash drives, um, far exceed the capacity increases of uh, mechanical hard disk drives. We've also seen performance of flash drives, um, increase, um, and probably most importantly we've seen the um, cost efficiency of flash based storage platforms on a dramatic downward curve. That means that it simply won't be economic, especially in 10 years to be running hard disk drives. And some of this also ties back into the sustainability angle that we were just talking about. Flash allows people to lower significantly the power consumption of their systems, uh, compared to mechanical hard disk drives. So that's definitely one thing we'd expect to see happen. Also with what we've seen over the last 18 months with generative AI, I, um, think we're seeing no end to the insatiable appetite to store data. Phenomenal, um, uh, growth in data across all aspects of business, uh, and social, uh, life. So I think that that growth will uh, continue. But equally over the next 10 years we're going to see even more pressure applied from a sustainability perspective to both lower and increase the efficiency of uh, all systems that consume power. And you know, whether that's in our home life, whether it's in our business life, um, I think we're going to see even more focus on sustainability, which has to be a good thing.

Speaker A: I completely agree with you, Patrick. Yeah. Um, from the conversations I've been having on this podcast and even just more generally, I know, uh, that sense of pressure that leaders are feeling to hit their sustainability targets, and not just about hitting targets, but just to do the right thing, um, to make real change where change is most needed. Would you say that data storage is, um, one of the many bricks in the road to helping these leaders achieve net zero, for example, and uh, is it something that can help them reach net zero fast enough to actually make a difference?

Speaker B: I think it's an interesting one. Clearly we've seen the massive growth in the number of data centers in every geography, um, that you can think of pretty much, uh, whether it's the challenges that Ireland are facing with massive growth in uh, data center building there and consumption of power, whether it's the data center growth down the M4 corridor in the UK, um, which means that we're seeing residential power demands, um, competing with data center power demands. And so people are really looking at the power consumption of data centers, which is depending on whose numbers you believe in. Around about 1 to 2% of global power consumption is taken up with data centers. And then within the data centers Today around about 20% of the energy consumed is on storage. So you can make a meaningful difference by adopting the most power efficient storage that you can buy, um, both in terms of greenhouse gas emissions but also increasingly in terms of energy costs, which

Speaker A: helps the bottom line and who wouldn't want that? Absolutely. Um, that sounds great. I know that one of the main things, um, that are sort of slowing people down from getting to that meaningful point of change quickly enough is purely just not knowing what the priorities are, not knowing where to begin. It feels like there's so many different plates that we're all having to spin to try and get us over the line. Um, do you have any advice for any of our listeners who might be struggling to determine priorities with regards to sustainability?

Speaker B: There are A couple of key things that we see when we're talking to customers, um, firstly that old adage of if you can't measure it, you can't improve it has never been more true or more relevant. Uh, and so understanding where the major power consumers are within an organization, uh, so that you can actually focus on those, the right areas, um, but equally going after the low hanging fruit, um, within whatever landscape you have of power consuming systems. And then I think what's interesting in the world of storage is

Speaker A: you know

Speaker B: as we mentioned earlier, storage manufacturers have been encouraging people to move their data from an old system to a new system every three or four years. And so even though it's painful, people are used to that process. But because they're used to that process, the ability to move to a platform that does lower the energy consumption and that does put an end to that for annual or every three year rip and replace process or cycle, um, has to be a good thing. And so actually it's not one of the harder challenges to adopt a new storage platform.

Speaker A: It's great to know that um, yourselves at Pure have solved that conundrum for people. I know that you recently uh, announced a stack performance benchmark result for high frequency and quantitative trading. Um, how do next gen solutions like this separate themselves and why is that such an exciting thing?

Speaker B: This is an interesting one because you know as we look at our uh, systems and the use of Flash, we're now seeing Flash being able to address data requirements all the way from you know, low performance requirements for archive data all the way up to the highest performance, um, requirements for you know, AI machine learning, but also real time analytics. And this is an area that we've been focusing on with one of our platforms, uh, which we call the flashblade. And we worked with uh, Stack to use uh, their benchmark, um, the stack M3 benchmark which is for uh, um, tickdb data, uh, and it really demonstrated the performance capabilities of the flashblade platform. And it's quite unusual for us actually to publish uh, publicly uh, performance data because it can often be a constant leapfrogging process. But we were so happy with the results um, of the Stack benchmark that we decided to um, put them in the public domain and this. Why is it important? It's because tick databases are uh, the foundation of financial services organizations. It's where they hold all of their market data and do all of their analytics on market data to try and predict future movement in the markets. And so being able to do that analysis incredibly quickly and faster than the competition gives them a competitive advantage in the industry. And so it's one of the areas where performance really makes a business difference.

Speaker A: Of course, I mean sometimes that's really all it comes down to is just performance, speed and resilience as well. Um, so with uh, the speed and the pace of change that we've already talked about, um, a little bit today, do you have any tips for people uh, within organizations who are looking to bolster their resilience and gain that competitive advantage?

Speaker B: It's really interesting at the moment. If we'd have been talking about resilience several years ago, you'd look at uptime and availability, you'd be looking at disaster recovery and business continuity. Um, what we saw during the pandemic, um, was another angle on resilience, which was um, supply chain resilience, being able to get systems when you need it to support your business. Um, too many people were having to wait a long time to be able to order systems. Fortunately we were able to hold our um, shipping times throughout that period, which definitely benefited our customers. Um, and one of the other aspects that helps customers on that side is a trend that we've seen accelerating over the last several years, which is people consuming their ah, storage services not by buying systems, but by consuming a subscription so that they actually don't own the systems that they run in their data centers, but they consume them as a service. And that becomes interesting if you look at that whole resilience play because it means that they can be confident that they don't over buy systems based on what they think they might need in one, two or three years time. But equally by not over buying they aren't putting lots of technology in their data center consuming power producing greenhouse gas emissions unnecessarily. But because they're consuming a service, they can react very quickly to business demand, whereby we will make sure that new capacity is brought into their data center to support changing business demand and to make sure that there's always headroom. And so that then gives them a very resilient service that they can consume. And it's backed by a set of SLAs that talk about capacity, performance, availability and give a customer the confidence in that service. So that's becoming a really compelling proposition for customers across all industries at the moment.

Speaker A: Um, and what uh, a great benefit to have. Um, you mentioned there, um, trends touching on emerging trends and this sort of subscription as a service trend that's rising. Um, this is definitely one that I think is um, getting a lot of buzz at the moment and people are sort of talking a lot about and where that might be going. I was wondering, with your expertise, are there any sort of other trends or exciting things on the horizon that you think maybe aren't yet getting the attention that they deserve? Anything that's very cool and positive and exciting that's not yet hit the mainstream?

Speaker B: I think it's, you know, it feels like uh, 2024, um, is the year of AI in all its forms. Um, and that's been talked about so much. But one of the areas that I think people are rapidly understanding is that part of the tool chain, um, for a lot of the data sources within AI are increasingly moving towards containerized platforms using Kubernetes. So the world of microservices, the world of containers and that has seen I think a tremendous march over the last few years where organizations are looking at modernizing their applications, or if it's a software house, then um, providing software as a service, then modernizing the applications for their consumers. And that's delivering benefits in many different ways. The adoption of containers means flexibility in terms of where you run your application so that you can run it cost effectively in the public cloud, but you could also run that same application in a colo or in your own data center. So you have that flexibility to run your application where it needs to run. Um, and that application and the use of Kubernetes leads to a much more agile development model and faster release cycle which is essential to organizations today who are operating in incredibly competitive environments where releasing the next new feature um, can give uh, them advantage over the competition. So I think the big, the continued push towards containers, microservices and Kubernetes is a big trend that we continue to see.

Speaker A: Sounds like that truly is the future of data storage. Thank you very, very much Patrick. Um, we've talked about some really interesting stuff today and there's an awful lot that we've touched on. Do you have any recommendations for further reading for anyone listening who'd like to learn a little bit more?

Speaker B: So certainly anybody, um, who wants to learn more about pure storage, purestorage, uh.com uh, and especially check out the blogs, uh, section on that website. Um, it really highlights a whole load of the new trend areas that we're observing, um, across our customer base. And then for people who really like reading, um, I've just finished a ah, really great book called Chip War by an author Chris Miller. And it's a fantastic history of semiconductors, people, companies and politics. And whilst the idea of reading a book about semiconductors might sound really dry. Um, it's genuinely unputdownable, um, as a book. So I'd strongly recommend that Chip War.

Speaker A: Thank you very much Patrick. That sounds right up my street. I'll definitely check out Chip War and also our listeners. Please do check out Purestorage.com as well. Thank you so much for your time, Patrick. I've really enjoyed talking to you.

Speaker B: Thanks very much Todd. It's great to talk to you too.

Speaker A: I'd like to thank Patrick Smith for joining me on CEO Extra for the CEO Digital show. And thanks to you for joining us. If you'd like more information, Please do visit PureStorage.com or the Special Innovators section of CEO Digital where you can find a whole wealth of extra information about the great work that Pure Storage are doing. Don't forget to subscribe to the show wherever you get your podcasts. Until next time, thanks for listening.