00:05
Ajay Singh: A lot has been said about this journey, we are all on digital transformation. In fact, I think we look back one day, and realize that this was the defining moment of our careers. Hopefully, for most of you the defining opportunity of your careers. Many are guilty of viewing digital transformation,
00:34
as something cliche. But I want to get very specific with you today, I want to break down the steps for a true digital transformation journey, and how a modern data strategy can either accelerate or sync your digital journey. And finally, I'm delighted to be joined by a team of Pure engineers. As we
01:02
share how key innovation, that Pure portfolio can be powerful tools, and accelerators for your journey. So let's start at the top. digital transformation is a journey, not a destination. It's called a journey for a reason, you'll never be done. So embrace it as a continuous growth and learning exercise, continually
01:30
improving and extending your digital ambitions. Today, getting to the cloud, and automating infrastructure, maybe your definition of digital transformation. And that's great. But once you get there, you will be able to see a few more steps down the path, perhaps to AI and machine
01:53
learning, or better real time customer experience, or new ways to empower developers. Every step gives you value and the visibility and the courage to see a bit further and be more ambitious in your goals. That's our goal as your partner in this journey to help you confidently take steps as you pick up the
02:22
pace a bit faster. Now everyone's digital transformation will be a bit different. But we believe there are three really big and pretty universal steps in the journey. First, modernizing infrastructure. Second, modernizing processes. And finally, third, modernizing
02:52
applications. modernizing infrastructure is about embracing hybrid cloud in taking advantage of technologies like flash, faster networks, and in memory computing. modernizing processes happens at many levels, from driving, DevOps and ci CD to automating infrastructure delivery. So your
03:17
developers are never waiting. modernizing applications is all about leveraging cloud native architectures, microservices, containers and Kubernetes. To take advantage of the new software tool chains, like scale out databases, real time analytics, and data streaming. These three overarching steps in
03:42
the digital journey are huge. And they of course go far beyond data and storage. But data and storage impacts each one in a profound way. And any digital journey must be built on a solid strategy for modernizing data. modernizing data infrastructure, modernizing data management processes, and delivering data
04:16
services to power modern applications. At Pure, we call a strategy for this, the modern data experience. So let's go deeper and explore each of these areas starting with modernizing infrastructure. Pure pioneered the vision in 2009 of the all flash data center. People thought we were a bit crazy back
04:47
then. But look at the data center today. Not only did we prove that storage arrays could be all flash at the performance tier, but we brought flash Do big data analytics, data lakes, tier two applications, and even backup and recovery. We often pioneered these use cases for flash and
05:12
technologies like NVMe and Q LC years ahead of a competition. In fact, most of our competition still sells spinning disk for many of these uses. All of this is born on the back offer simple idea, or really an innovation commitment at Pure. We have to understand flash better than anyone else. We have to learn
05:44
flash, and write our software to talk to flash directly optimizing everything from end to end. So who better to talk to you about flash innovation than one of the folks who actually does the real innovation? I would like to welcome one of our talented engineers Zoltan Dewitt, who has been
06:09
instrumental in pioneering the use of qlc flash in our systems. Welcome, Zoltan. Zoltan Dewitt: Thank you, Ajay. Ajay Singh: So Zolton as QLC is such a game changer in the industry, enabling the next generation of apps to be able to
06:26
afford to run on flash. Aside from the obvious differences around density and cost, how is QLC different from other NAND? When it comes to integrating into a storage device? What thoughts Did you and the team set as a North Star for the project.
06:44
Zoltan Dewitt: So QLC, NAND is a very different beast than prior NAND generations, it has a very difficult and complicated programming model, you sort of have to imagine that it gets exponentially harder every time you want to squeeze another bit inside of a flash NAND cell. And so it really was a big team
07:07
effort from you know, all of the teams that are from the hardware team, the firmware team, and our direct flash module software team, that we really wanted to make it just as transparent to the customer experience as possible. And, you know, other companies, they tried to use QLC, you know, by throwing a
07:32
bunch of persistent memory in there, or, or just use off the shelf SSDs. But it really destroys the cost profile of QLC, you know, off the shelf, SSDs, you could get, you know, 20% overhead in the storage efficiency that you're getting out of it. So, for us, we really focused on making 100% QLC
07:54
solution, solving the problems of QLC in software, and just keeping it to the highest enterprise standards, you know, as good or better rate overhead. And just the same reliability standards that we have. And, you know, I think the adoption of the FlashArray//C line in the field is validation of that end
08:17
to end approach. Ajay Singh: That sure has been amazing thanks Zolton. Building great products is not just about using the best components. There's a lot of work that happens to make sure these components perform the best and give you years of use. Our
08:36
original vision with Evergreen was to deliver an array that can last a decade in the data center. Flash plus the right software approach is actually making this possible. Sustainability is a core theme and differentiator for us at Pure meaning longer life of the product, and less environmental
09:00
impact in your data centers. On that topic, Paula Zubriri has been instrumental in designing and testing mechanicals. Paula, I appreciate you joining us at accelerate. Paula Zubriri: Thanks Ajay. Ajay Singh: One of the more surprising artifacts of the rise
09:17
of the cloud giants is a renaissance in hardware engineering. When you run at cloud scale, driving efficiency through unique hardware and software optimization is key. At Pure, we view our opportunity as democratizing the same level of hardware optimization for everyone. You don't have to be a
09:42
cloud giant to run super optimized hardware anymore. Paula, let's dive into some of the details of how this is done. Can you take a moment to explain what a mechanical engineer at Pure does? And how does that work translate into creating more efficiency, sustainability and a better customer
10:02
experience. Paula Zubriri: Sure, I get to design technologies that are integrated in our system on a daily basis. As you may know, when you look at increasing the density of storage and the compute, the biggest barrier often becomes a thermal design,
10:18
you can only cool things. Sorry, you can only design things as dense as you can call them. My team worked on a data aggregating script that I use to analyze the temperatures at multiple sensors on the entire flash blade system so that we could build a rough heat map of the blades and the networking
10:36
modules. On the current flash blade, I leveraged that data to design multiple heatsink prototypes in order to run a more efficient system with better temperature margins. improving our systems power and efficiency is key to making a more sustainable product. Another area we focus heavily on
10:56
is continuously increasing quality and reliability. Even after a product is in production, we work constantly on improving the design. For example, the current flashlight system was so popular among our customers that a system manufacturing quantities increased, we persistently
11:14
worked on troubleshooting and streamlining the assembly process. We did so by tweaking the tools and the fixture designs as well as updating the assembly instructions to seamlessly accommodate higher production volumes than we had originally anticipated. So as you can see, we're never really
11:30
satisfied. We're constantly optimizing our hardware and passing that benefit to our customers as part of the Evergreen experience. Ajay Singh: Thanks for sharing that with us, Paula. And thank you for your passion in building sustainable solutions for our
11:44
customers. After modernizing infrastructure, the next logical step in the journey is modernizing data management processes for how you deliver storage. And the key focus here is automation, and intelligent prediction. Pure brings a cloud like operating model to the data center that prepares you to
12:09
easily move traditional and modern Kubernetes based workloads into the cloud, out of the cloud and across multi cloud environments. Automation delivers the storage as code experience for developers across traditional and modern workloads. This enabled integration with automation
12:30
frameworks like terraform, Ansible, puppet and Chef best in class support for VMware, extending to the edge via bare metal as a service or AWS outpost, or abstracting multicloud storage resources to any industry Kubernetes offering via OpenShift, Rancher, Tanzu, EKS, GKE, or AKS. The Kubernetes
13:00
space is especially critical for automation, the agility and change rate of modern Kubernetes based applications is so fast and so fluid. There is no choice but automating the data management tasks. This is what a Portworx solution is all about. And in addition to a container native provisioning, migration
13:23
and disaster recovery, a key data management task that Portworx or Portworx automates is container native data protection. I would like to welcome engineering Puritan Aditya Donny our lead for Portworx PX backup. Aditya is also closely involved with PX DR. And PX Enterprise and has
13:47
helped Portworx customers deploy cloud native architectures at massive scales. Aditya, thanks for joining us. Aditya Dani: Thanks, Ajay. I'm excited to be here. Ajay Singh: So Aditya, you're on the forefront of the move to containers and how does to modernize automate data
14:04
management for modern applications. In considering what DevOps teams are trying to accomplish, what excites you about Portworx in the entire Pure stack? Aditya Dani: Your right there has been a huge uptick in container adoption, especially with Kubernetes. You see,
14:19
Kubernetes provides this very fine grained role based access controller our back that allows DevOps to deploy these modern applications in a multi tenant cluster. Keeping this in mind, we have been building our back support in PX backup one of our products. Essentially, with our back, we are allowing different
14:38
user personas to access the same product from a different point of view. So let me give you an example. And infrastructure admin creates a backup target and shares it with their users. An application owner on the other hand now doesn't have to worry where their applications are getting backed up. They can
14:55
focus more on writing code and delivering these business critical applications. You see What I like most about my work here is this customer engagement. We have worked with a lot of customers and in understanding their pain points in deploying the stateful applications in Kubernetes. This
15:11
customer first approached has helped us driver design decisions and deliver quality features. Ajay Singh: It's great to see the Portworx platform continue to expand. And I know you were really excited to announce this last week at coop con. Thank you that the automation of formerly
15:31
manual infrastructure management processes is a great first step in modernization to drive efficiency. But, but it's only the beginning. The next step to leverage the is the power of AI and machine learning to find efficiencies that are beyond the obvious. At Pure, we've been collecting telemetric storage
15:54
workloads data for a decade now. And our met AI engine is constantly getting smarter and delivering AI power predictions, automation experiences for our customers. One particular area we worked a lot on is workload centric predictions. How workloads will change over time, what performance and capacity
16:17
they will need? how to best optimize many workloads across your fleet of storage resources. To talk to more about our innovation in automated predictions, I would like to welcome another engineering Puritan Ajay D'Souza. Ajay worked for our digital experience team, and was
16:38
instrumental in building the recommendation engine in Pure one workload planner. Ajay. Great to see you. And by the way, great name. Ajay D'Souza: Thank you, RJ. You may have noticed that we both share a very exciting name, which loosely in South Asia
16:55
means invincible, I'm thinking why not make our infrastructure admins invincible by empowering them with machine learning and AI. Ajay Singh: That would be great wouldn't it? Accurately predicting how many many applications behave together on
17:12
a storage system is challenging. But it's key in making infrastructure admins invincible, so that they can make the right choices to manage and grow the environment. Ideally, we can help them make the choice for them. With a project like that, RJ, where do you start? And when you're
17:34
imagining users interacting with the planner, what did you want them to experience? Ajay D'Souza: It all starts with the focus of the team, the focus of the team is more power to the customers. And the way we do that is we use machine learning to simplify the interaction and
17:52
to efficiently improve the utilization of storage and storage services. The machine learning that we do on the field telemetry data gives us very unique insights into the profile of the workload patterns that the customers are running. So these insights help us to build very simple and efficient
18:10
applications that enable the customers to plan for the growth of these workloads both in capacity as well as in performance, and also help them to place these workloads optimally on the storage arrays, the machine learning that we use to build simulation tools help the customer to run simulation
18:31
on various planning scenarios, when getting suggestions and recommendations from the tool. What is really unique to the machine learning that we have here is that our machine learning is not just limited to statistical learning on the field elementary data built into it is the engineering knowledge
18:47
base that we have here at Pure in Flash and storage. So our machine learning models are built ground up for storage. We are excited to be on this journey with our customers, where we are partners in machine learning and where we help them towards storage automation.
19:02
Ajay Singh: Thank you invincible, Ajay. Okay, we've talked about modernizing infrastructure. We've talked modernizing data management processes, through automation, and AI driven predictions. So let's talk about the third goal, delivering the data services that modern applications need. A
19:26
new stack of data centric applications is changing how we process customer and machine data. And they are now dramatically different from a need standpoint. These applications are scale out in nature, often deployed in containers and need fast access to data across very, very large
19:45
data sets. And since they deal largely with unstructured data, they are right in the middle of the transformation from file to object data formats. You've already heard a big piece of history torey in the space boatworks. But another piece is flash blade, a unique fast file an object platform for modern
20:10
applications. To talk more about delivering the data services that modern applications need, I would like to welcome engineering Puritan Nikita sirohi. Nikita joined the flash played control plane team one year ago, and established herself as a senior engineer. She was quick to run the code
20:32
base, and is an example of highly collaborative teamwork. Hi, Nikita. Nikita Sirnhi: Hey, it's awesome to be here. Ajay Singh: Great to hear you Nikita, one thing I've heard about you is you have a strong sense of ownership. What gets
20:51
you excited about the breakthrough fast file and object services your team is providing to customers. And you'll work on the control plane to connect these with modern applications. Nikita Sirnhi: Well, let's start with the technical side. So on
21:05
the technical side, I'm excited about the new frameworks, we're building the new challenges the new ways we have to think about concurrency, and the opportunity to explore so many new tools and languages on FlashBlade. And then tying that back to the long term for our customers. As we've said, unstructured data is the
21:24
new currency for modern applications. As a developer, on FlashBlade, it's always been about stability, robustness, just building a really good product. But now it's also about platform independence, growing these data sets supporting customer agility. And when thinking about control planes in
21:43
general, not just on FlashBlade, it's empowering to realize how many control planes in the world are powered by FlashBlade. And what that will mean for the growth of unstructured data be more manageable, more sustainable than the legacy of traditional storage, day to day, a lot of the same problems I get
22:03
to solve in like when writing code when doing designs are the same ones that our customers are solving, actually with FlashBlade, the product that I'm building. So tying that back to ownership. That's exciting as a developer, seeing the whole puzzle like that, from the nitty gritty of technical exploration,
22:22
to the wide view of what we're building for customers. And that's really what motivates. Ajay Singh: Thanks again for joining us, Nikita. So in closing, we've walked you through a digital transformation journey of how a modern data experience can help us
22:43
accelerate your path to modernization. These three steps, modernizing infrastructure, modernizing data management processes, and building data services for modern applications are universal. And I hope you all found key points within that, that you can take back to your
23:04
teams and start working on today. As with any journey, the most important part is to start. So let's get started. We would love the opportunity to partner with you in this exciting journey. And the engineers you met today are just a few example of the talented team at Pure that is excited to engage with
23:28
you to accelerate your journey. And now it's time to announce the winner of our breakthrough award in the visionary category. Visionaries think big about solutions. But they also think big about how to make those things happen. This particular customer had a huge vision for their customers. And they also
23:54
knew they needed a visionary technology solution to make it happen. Check it out. Options Technology: The market never closes in global finance. It moves from one exchange to another circling the planet and a rapid flurry of intricate transactions, information
24:21
processing and complex instruments that keep the world economy afloat. There are no margins for error, not a moment to lose no opportunities for do overs. nanosecond speed is the norm, relying on the speed and stability of Pure Storage's wide range of services options technology stands tall in this
24:43
space as a visionary provider committed to making sure markets run right. More than 200 Global firms depend on their high performance trading infrastructure and cloud enabled services to provide an agile cyber secure, scalable platform their clients success is the testament to their success. That
25:04
Options Technology, a powerful vision of the future places it at the cutting edge of Finance. Unknown: Congratulations to the Options Technology team. They use virtually every part of the Pure portfolio to make it all happen. And it's a pleasure to see it all come together so
25:32
seamlessly. And congratulations to all the breakthrough customer award winners. Hats off to each of you and your incredible accomplishments.
