What’s in store for memory and storage: Micron’s Raj Hazra talks about digital transformation in 2021
An interview with Raj Hazra
As digital transformation accelerates around the world, impacting businesses, health care, supply chains, customer service, entertainment and more, it brings with it an explosion of data and growing demands on computation, memory, storage and infrastructure.
For Raj Hazra, who is senior vice president of corporate strategy and communications at Micron, there has never been a more thrilling time than this golden age of data. "We live in a world where the rate and pace of the possibilities for innovation are unbounded. We are now at the doorstep of taking things that we thought were science fiction and making them real, and it's only going to be exponentially faster going forward," says Hazra. "The next 20 years are going to be nothing like the last 20, and that is exciting."
We live in a world where the rate and pace of the possibilities for innovation are unbounded. We are now at the doorstep of taking things that we thought were science fiction and making them real, and it's only going to be exponentially faster going forwardIn Hazra's role at Micron, which began earlier this year, he's working to ensure that data—which he describes as "the oil that drives the economy forward"—is collected, moves and is used as effectively and efficiently as possible. Data should be an asset, he says, and not a liability, and the way that it's stored and accessed makes all the difference in whether it's applied or waste.
Protocol sat down with Hazra to talk about the exponential growth of data and its accompanying challenges; the ways that new Micron technologies are quickly transforming immense amounts of data into valuable insights; and how businesses should think about digital transformation as they look to the future.
Can you briefly take me through the history of data and why it's grown so rapidly?
Raj Hazra: If you think about it, data has always been around. In ancient history, people wrote things on walls, on caves and on papyrus. Data, at the end of the day, is a recording of something. What has really made this explosion has been going digital. Think of the electronic health records that were written on scribbled sheets by people who, alone, could understand their own handwriting; data around how machines work; data around customers and their behavior. All of that going digital now means two big things: No. 1, data can be shared easily, and mechanisms like the Internet have democratized digital data. No. 2, it can be computed on—not just seen and read. As Moore's law has made computing more affordable, equitable and almost omnipresent, the confluence of digital data and technologies to compute them have driven this to where data is now a real asset.
As data has grown exponentially, what challenges has it revealed?
The key question now driving this forward is what do you do with this data, in terms of storing and moving it? In the last 15, 20 years, the storage capacity of data was fairly well understood. We were more constrained by algorithms and methods to use data. Whereas now, data is so explosive, coming so fast and there's so much of it. And with technological advances such as artificial intelligence, there really is no bad data. There's no data that you can say oh, let me throw it away, I don't need it anymore. Every piece of data—even in history—is important for learning, training and doing AI.
So now the storing of data and making it available to the right compute at the right time is the bottleneck problem. And that's what Micron is addressing. How do you facilitate the storage, retrieval and movement of data so the right insights can be taken? Innovations in how you store that mass of data are what's going to unleash the future of the golden age of data. Otherwise, data looks like a deluge we could sink under.
In what ways is Micron working to address these storage challenges?
We're getting better at memory and storage. We're able to store higher capacities at lower costs; we're able to store and move data with lower levels of energy spent; and most importantly, we're able to come up with architectures for storing masses of data that make it easier for someone to compute on that data.
To expand on that last point, we're coming up with new technologies that make something called 'persistent data' happen, so you don't have to power up the memory like you do in a laptop today, rather, like your hard disk, where your data stays whether the power is on or not. We're coming up with architecture that says, 'Here's the data, here's intelligent organization of the data,' so you know where your data is—kind of like your chapter index in a book. That facilitates someone being better able to use the data. And then of course there's raw performance. When you want the data, how quickly can you get it? This one of the significant gains to making a decision, or, as they call it, 'time to result.' We are working on the forefront of all of those.
Can you highlight any of your products that are specifically helping to solve for these demands?
3D XPoint is one memory technology that's part of the quiver. We obviously have DRAM—you can't build a computer without DRAM, which is the memory that stores all your data and software when the system's processing unit needs to execute. 3D XPoint is a remarkable innovation that provides phenomenal capacity at a very reasonable cost and then provides persistence. Think of what would happen if, when you close your laptop or powered it off, the data that is "cached" in your system's memory still stayed. Not your hard disk, or SSD, but all data stayed, and was always available, versus having to be brought back from other persistent storage. That's what 3D XPoint does: it is the combination of a new memory technology that has tremendous performance and cost benefits—particularly over higher-performance DRAM—but it also has persistence, so it lets you build things.
What's a real-world example of how that could be useful to a business, for instance?
When supercomputers run, that's hundreds and thousands of servers running at one time on one big application. If one of those servers goes down, where do you store the state of where it was when it went down? If you start all hundred-thousand servers all over again on the job, that could take several weeks. What this does is once the server dies, there's a way to go in and look at memory like 3D XPoint on the server, and it holds the state of the server before it died. You can actually restart the computation at that point versus wasting weeks of computation to return to that point. That's just one of many examples of what 3D XPoint allows you to do, delivering on capacity, performance and persistence, as well as cost.
How does the innovation of computing infrastructure relate to data insights, and how does Micron fit into the equation?
In the past, it was really computing architectures, like CPUs, and then GPUs that drove requirements for innovation in memory technology. It was very simple: you've got to feed the beast. So as the people that made the computing beasts put in more cores or more compute power, it was like come on, you've got to give me faster memory. As we were lowering the power on the processor we had to lower memory. So in some ways it was difficult, but a well-defined world of working to be faster, lower power and lower cost.
But now, look at the big cloud service providers. They construct systems, and systems are a large number of servers working together, and the construct is now no longer about just giving me faster, better memory for a server, but it's about how you distribute the data across the system. Memory has to become way more intelligent than just serving it up faster with lower power and lower cost. The role of memory in creating user value is now higher. So memory is now no longer the thing that you design last. It is what you start with in order to define what the system looks like, and what the specifications are.
What should business leaders be thinking about for 2021?
Right now, "digital transformation" is the big buzz word, and business leaders should think about what it really means. For most companies, whether you're in manufacturing, retail or service, IT is becoming the business, and you need to embrace the cloud. The cloud isn't just running to a public cloud service provider. Cloud is an architecture that will be in hybrid form. That is, you will have some things in public cloud services, some things on premise, built in a cloud-like architecture to manage, and you'll be using multiple cloud service providers in your own enterprise as well.
This is a massive change. It's a massive change from not just the technologies, but your key partnerships you've built over the years, from vendors to the internal talent that must understand and work with these technologies. This digital transformation is a change that I would say is irreversible, and, long-term, extremely beneficial. Going to digital and going to the cloud will be sometimes rough, because all new technology and disruptions are rough. But business leaders must embrace the fact this is the future, and they must invest in it and learn it.
When you think about a future where data insights are virtually limitless, what excites you the most?
What makes me passionate about what I do is the ability to take these technologies and not confine them to the needs of a few, but to touch the lives of all in the world. These technologies and innovations are saving lives through personal medicine and through the more accurate prediction of natural calamities, like earthquakes and hurricanes. They're predicting supply chains so that vital supplies can get to people in time. They're making education accessible to so many.
This ability to enrich life, not just lives, and create a new normal for the way we expect to live in this world, interact with each other and create value, is amazing. Being in the business to serve others is a most gratifying and humbling privilege.