NetApp Unveils Unified Data Storage Built for the AI Era | Sandeep Singh | NetApp

theCUBE.net

Clips
News
More from NetApp Unveils Unified Data Storage Built for the AI Era

Sandeep Singh

SVP & GM, Enterprise Storage

NetApp

Unified data storage gets new look for the AI era with NetApp’s latest releases

On Tuesday, NetApp Inc. introduced new hardware designed to run demanding generative AI workloads in on-premises data centers.The data storage and management company announced the NetApp AFF A-Series systems, all-flash hardware that will support acceleration of advanced workloads while optimizing storage costs.Rob Strechay and Sandeep Singh on theCUBE set at NetApp Converge in Las Vegas.“The major part of our announcement was that next major advance to unified data storage,” said Sandeep Singh (pictured), senior vice president and general manager of enterprise storage at NetApp

play_circle_outline Unlocking Efficiency and Simplifying IT: Embracing a Unified Storage Approach to Overcome Common Challenges and Reap Benefits

play_circle_outline Privacy and security considerations in AI deployments

play_circle_outline Introduction of new NetApp AFF A-Series systems for performance and data protection

play_circle_outline Collaboration with Lenovo and Nvidia in building AIPod

Info

Sandeep Singh | NetApp

Sandeep Singh

SVP & GM, Enterprise Storage NetApp

search

>> Hello and welcome back to NetApp's Converge 2024. We're here, really just so exciting to understand what's going on within the entire ecosystem that NetApp has brought together with their sellers and their partners. And I got to say I'm really excited because I get to talk to somebody I've actually known for quite some time and we've worked together and built product together. And it's just exciting to see what he's been doing at NetApp. So, I really want to invite Sandeep Singh on, who's the SVP and general manager for enterprise storage here at NetApp.

Sandeep Singh

>> Rob, thank you. It is a great pleasure for us to be reconnecting and having an opportunity to have a conversation.

>> Yeah, I mean, I think again, it's been a while. We've both done a lot of other things and it's been a lot of fun to really dive into what you guys have been up to, and now you get to see it from just a whole different level. I mean, George was talking about it earlier about the high-level vision. But why don't you kind of give us a high level vision of the unified storage and data approach that you guys are really taking? Because it's not just about block file and object, it's about everything else that goes along with that.

Sandeep Singh

>> Yeah. I'll perhaps start at the beginning in terms of... I've been here about 18 months. I visited hundreds of customers. Often, the common thread is data, cloud and AI are the three forces that are really shaping the IT landscape and the plans for customers. And data has become so much more about having to go not only store and serve the data, but protect it from ransomware attacks and cybersecurity threats. And clearly cloud and AI customers are figuring out how to harness the power of cloud and AI across the board. But in the customer conversations, the one challenge that everybody articulates is complexity. And as soon as I double-click on complexity, it's complexity that gets compounded through bespoke infrastructure silos. And the reality for a lot of the enterprise organizations is they have seven or 10 different bespoke infrastructure silos when they have talent shortages and skills gaps, and they're having to go retrain and re-automate. When you think about data security, if you have seven or 10 different data security models, there's no way to have confidence in that data security. And when you think about protecting from disasters and recovery workflows, how do you have confidence in your recovery workflows there? This is where unified data storage is the game changer for customers. And only NetApp is delivering truly unified data storage. And what we mean by that is, basically, a single storage OS ONTAP that can underpin any application data, workload, any data type, whether it's structured or unstructured, across on-prem and cloud. And what that is bringing to customers is truly simplicity at scale, not just within silos. So, they can train once and automate once. They can have one data security model they can be confident in. They can have one data recovery workflow, so that they're confident about their recovery instances as well. This is what's exciting about unified data storage.

>> Yeah, and I think again, that cognitive load that's put on it people today where they have to learn all of these bespoke technologies and stuff like that. And like you said ONTAP, I mean I was at NetApp even before you were at NetApp a long time ago, and I look at it and go, it's always been about ONTAP and how you really operate wherever it is on-prem, in co-lo, in the cloud, it's really ONTAP. You're learning that one thing. Right? I mean that's really helps these organizations, especially when they're under attack from things like ransomware. And other security and cyber resiliency they need to bake in. Tell us about that and what you're hearing from customers.

Sandeep Singh

>> Yeah. Ransomware is just a top-of-mind C-suite priority across every customer. Every customer that I talk to nowadays, most of the questions are going to be around ransomware. It's not a matter of if, it's a matter of when. People recognize that and they also recognize that storage often becomes that last line of defense. And this is where it becomes incumbent on the IT leaders to get the most secure storage in their infrastructure. And that is where NetApp comes into the conversation, where we are delivering the most secure storage on the planet. What we mean by that in the context of ransomware is we've got a rich suite of capabilities for ransomware protection, but where we're really changing the game is in the ransomware detection side of it. Detection so far has been an offline post-process via backups, that can be hours to days later. We are graduating detection to be a first-class citizen in primary storage where it's real-time, it's built in, and it's designed for 99% plus accuracy. What that means is now customers can be confident about their storage infrastructure, being able to detect a ransomware attack, immediately, have a recovery point created and alert the admins. And now, they can enable rapid recovery before their data really gets impacted. So, that's where we're really changing the game. We're super excited about that. We're seeing a dramatic uptake in the overall capability. That's through the autonomous ransomware protection that's built into ONTAP. We're just seeing a dramatic adoption across the board. And with today's announcement, we've gone a step further. We've had ransomware recovery guarantee. We've now added in a cyber vault solution available to customers, so they can now add an additional site that is a cyber vault. It's built on low-cost infrastructure, whether on FAS or in NetApp C-series, with the logical air-gapped storage infrastructure that's been hardened from a security standpoint. And it provides these indelible, immutable snapshots, the full copies of the data that are available for recovery points for customers.

>> Yeah, I think that is the key. It's built off of those snapshots that have really been a defining, I would say, technology since the beginning of NetApp about having those and being able to make them immutable. And then, going that next step and making them air-gapped is critical. And I think that has to be one of those things that cross when we were talking about federal and CISA and some of what's going on with SLED, that has to be top of mind for everybody, not just in the government.

Sandeep Singh

>> Absolutely. And talking about federal, this is where NetApp is the only storage validated for storing the top-secret data. That's where customers are trusting NetApp with that level of security to make sure that their data is going to be secure. And then, the other thing I'll just touch upon briefly is the major part of our announcement was that next major advance to unified data storage. That's with the introduction of the new NetApp AFF A-Series systems. And in the context of ransomware, it is built in. All of these capabilities that we just talked about, they are built into the new NetApp AFF A-Series.

>> That was the thing is performance and data protection. That was what I took away from it, and it really helps everybody that it's built in. And in fact, we were talking to Gabie earlier on tonight, today because it isn't tonight yet, that around how basically it's built in. And I think everybody likes that. It goes back to people understand how to work with the NetApp array, and it's how that works all over the place as well.

Sandeep Singh

>> Yeah, absolutely. I'll touch briefly on the major portion of our announcement with the new AFF A-Series. What we are talking about is it's unified data storage built for the AI era. What that means is now basically the new AFF A-Series is powerful, intelligent, and secure. Powerful in the sense that it can help customers turbocharge every workload. It's better performance. It is up to 40 million IOPS. That's an astounding number. You and I have lived the storage .

>> I remember when we got to a million IOPS, it's like 40 million though. Yeah.

Sandeep Singh

>> Yeah, up to one terabyte per second of throughput across a cluster. All of that, when you couple that with six-nines measure data availability, always on data reduction there. Now, what that means is customers can turbocharge every workload without trade-offs, whether it's optimizing their VMware environment, accelerating their SAP, SQL, Oracle apps, or seizing the AI advantage and building the capabilities for AI, gen AI workloads. So, that's a powerful side of it from an intelligence standpoint, top of mind for everybody right now is lowering costs. Customers can easily protect, manage, and mobilize their data, and they can do it at lowest costs across the data lifecycle. They can simplify hybrid operations across the board, and intelligence is also part of building that foundation for enterprise AI. And then, secure part of it, you and I just talked about all of the ransomware side of it.

>> Yeah. No, I think that's great. But you also had another announcement, and I love this because I was around at NetApp when FlexPod launched. So, you have the AIPod, and you're doing it this time with Nvidia and Lenovo, and we're going to welcome them up in a short little while. But give a little bit of the high-pitch on why. Why AIPod?

Sandeep Singh

>> Absolutely. Look, overall, AI, gen AI, every organization is looking at how to harness the power of AI and gen AI, whether it's to drive productivity gains, to deliver that next level of innovation, or deliver that next level of customer experiences. So far, we've seen a lot about larger-scale AI as a service providers for large foundational model training. That's an expensive investment across the board. As enterprises are graduating and looking of graduating from experimentation to production deployments. There, the use cases are shifting a lot more to beyond model training into fine-tuning, retrieval-augmented generation and inferencing. And really the power for enterprises is to be able to take their enterprise data and combine that with the pre-trained gen AI models to make it more contextual for them. And this is why we teamed up with Lenovo and Nvidia to build a new conversion infrastructure. This is the new AIPod with Lenovo ThinkSystem servers. Alongside with Nvidia OVX certification. It's designed to enable enterprises to take their enterprise data, use RAG with it, with this conversion infrastructure. It's super simple, so that it's pre-validated. It can be fast time to value for customers. It's a great value proposition for partners to help deliver it to their customers along the way. It's powerful because it brings along the ThinkSystem servers, it brings along the Nvidia L40S GPUs with it, and it brings along the NetApp the AFF C-Series across the board. You've got data management built in. You've got the enterprise AI, NeMo Retrievers overall software stacks built in from Nvidia. So, think of it as a gen AI, converge gen AI in a box. And it's affordable for customers. So, as organizations are looking at deploying and production deployments, affordability absolutely matters.

>> Yeah. No, we see this. Most organizations aren't going to go against and try to compete against the large LLMs. What they're going to do is bring a model in like an open source model into some of the Nvidia systems or partner with Nvidia, with the models that they have, and really ride that curve. And what we've seen is actually in our power law distribution for gen AI is that, yeah, the big models are here, they're huge, but as you start to go out where the vast majority of the customers are buying for, it's for verticalized instances such as finance or telco. And when you get to telco, hey, you're going to put something out at the edge at a base station or something like that to do inference and to be able to be responsive. I think that's the kind of thing where you need these types of AIPods that really go and fit.

Sandeep Singh

>> Yeah, absolutely. We are super excited about this AIPod with Lenovo and Nvidia. It augments our overall portfolio of solutions that we're building and enabling customers with around AI and gen AI. Look, NetApp hosts a lot of the customer's unstructured data that AI and gen AI are built on. And we see it as being that good steward and shepherd of enabling customers to be able to truly harness the power of AI and gen AI.

>> And like you said, you're out talking to customers all the time. We talk to them in our research, and one of the things they always talk about is they don't necessarily want to bring all their data to the AI. They want to bring the AI to the data. Is that a common theme that you're hearing as well?

Sandeep Singh

>> Absolutely. I would say first, gen AI, it absolutely builds upon unstructured data, so that becomes important. If you already have the unstructured data stored on NetApp, for example, it becomes the opportunity to go, how do we enable AI, gen AI with it? Bringing data or AI to data is critically important. Lots and lots of that data has gravity. At the same time, it's archetypical workload for hybrid workflows. This is where NetApp has invested over a decade in really integrating at this cloud-native level with the hyperscalers. And we have a variety of technologies to enable customers to securely and efficiently be able to mobilize their data. And that's part of enabling those hybrid workflows. And what's becoming essential, especially for the production workloads in enterprises, is all about how do I have data management tied for enterprise AI? Model versioning, the data sets associated with those models, the data lineage, the data security and governance and privacy of data just becomes so critical. It's just a top of mind. I was in a conversation with a CIO of a financial services provider, and that was just the top of mind, critical part of it. So, these are all elements that are feeding into important needs for customers across the board.

>> Yeah, especially when they're looking at things like the European AI Act and other stuff coming down the pipe. But why don't we park it there for a moment, because we have some special guests that are going to come up on stage with us in a minute, and we'll talk through that in a second. So, why don't you stay right there and why don't you stay right there while we bring up those special guests. We're going to bring up Nvidia and Lenovo, and we're going to dig a little bit deeper into the AIPod that was just announced today. So, you stay right there with theCUBE, the leader in high-tech news and analysis. Thanks.

Sandeep Singh | NetApp

search

>> Hello and welcome back to NetApp's Converge 2024. We're here, really just so exciting to understand what's going on within the entire ecosystem that NetApp has brought together with their sellers and their partners. And I got to say I'm really excited because I get to talk to somebody I've actually known for quite some time and we've worked together and built product together. And it's just exciting to see what he's been doing at NetApp. So, I really want to invite Sandeep Singh on, who's the SVP and general manager for enterprise storage here at NetApp.

Sandeep Singh

>> Rob, thank you. It is a great pleasure for us to be reconnecting and having an opportunity to have a conversation.

>> Yeah, I mean, I think again, it's been a while. We've both done a lot of other things and it's been a lot of fun to really dive into what you guys have been up to, and now you get to see it from just a whole different level. I mean, George was talking about it earlier about the high-level vision. But why don't you kind of give us a high level vision of the unified storage and data approach that you guys are really taking? Because it's not just about block file and object, it's about everything else that goes along with that.

Sandeep Singh

>> Yeah. I'll perhaps start at the beginning in terms of... I've been here about 18 months. I visited hundreds of customers. Often, the common thread is data, cloud and AI are the three forces that are really shaping the IT landscape and the plans for customers. And data has become so much more about having to go not only store and serve the data, but protect it from ransomware attacks and cybersecurity threats. And clearly cloud and AI customers are figuring out how to harness the power of cloud and AI across the board. But in the customer conversations, the one challenge that everybody articulates is complexity. And as soon as I double-click on complexity, it's complexity that gets compounded through bespoke infrastructure silos. And the reality for a lot of the enterprise organizations is they have seven or 10 different bespoke infrastructure silos when they have talent shortages and skills gaps, and they're having to go retrain and re-automate. When you think about data security, if you have seven or 10 different data security models, there's no way to have confidence in that data security. And when you think about protecting from disasters and recovery workflows, how do you have confidence in your recovery workflows there? This is where unified data storage is the game changer for customers. And only NetApp is delivering truly unified data storage. And what we mean by that is, basically, a single storage OS ONTAP that can underpin any application data, workload, any data type, whether it's structured or unstructured, across on-prem and cloud. And what that is bringing to customers is truly simplicity at scale, not just within silos. So, they can train once and automate once. They can have one data security model they can be confident in. They can have one data recovery workflow, so that they're confident about their recovery instances as well. This is what's exciting about unified data storage.

>> Yeah, and I think again, that cognitive load that's put on it people today where they have to learn all of these bespoke technologies and stuff like that. And like you said ONTAP, I mean I was at NetApp even before you were at NetApp a long time ago, and I look at it and go, it's always been about ONTAP and how you really operate wherever it is on-prem, in co-lo, in the cloud, it's really ONTAP. You're learning that one thing. Right? I mean that's really helps these organizations, especially when they're under attack from things like ransomware. And other security and cyber resiliency they need to bake in. Tell us about that and what you're hearing from customers.

Sandeep Singh

>> Yeah. Ransomware is just a top-of-mind C-suite priority across every customer. Every customer that I talk to nowadays, most of the questions are going to be around ransomware. It's not a matter of if, it's a matter of when. People recognize that and they also recognize that storage often becomes that last line of defense. And this is where it becomes incumbent on the IT leaders to get the most secure storage in their infrastructure. And that is where NetApp comes into the conversation, where we are delivering the most secure storage on the planet. What we mean by that in the context of ransomware is we've got a rich suite of capabilities for ransomware protection, but where we're really changing the game is in the ransomware detection side of it. Detection so far has been an offline post-process via backups, that can be hours to days later. We are graduating detection to be a first-class citizen in primary storage where it's real-time, it's built in, and it's designed for 99% plus accuracy. What that means is now customers can be confident about their storage infrastructure, being able to detect a ransomware attack, immediately, have a recovery point created and alert the admins. And now, they can enable rapid recovery before their data really gets impacted. So, that's where we're really changing the game. We're super excited about that. We're seeing a dramatic uptake in the overall capability. That's through the autonomous ransomware protection that's built into ONTAP. We're just seeing a dramatic adoption across the board. And with today's announcement, we've gone a step further. We've had ransomware recovery guarantee. We've now added in a cyber vault solution available to customers, so they can now add an additional site that is a cyber vault. It's built on low-cost infrastructure, whether on FAS or in NetApp C-series, with the logical air-gapped storage infrastructure that's been hardened from a security standpoint. And it provides these indelible, immutable snapshots, the full copies of the data that are available for recovery points for customers.

>> Yeah, I think that is the key. It's built off of those snapshots that have really been a defining, I would say, technology since the beginning of NetApp about having those and being able to make them immutable. And then, going that next step and making them air-gapped is critical. And I think that has to be one of those things that cross when we were talking about federal and CISA and some of what's going on with SLED, that has to be top of mind for everybody, not just in the government.

Sandeep Singh

>> Absolutely. And talking about federal, this is where NetApp is the only storage validated for storing the top-secret data. That's where customers are trusting NetApp with that level of security to make sure that their data is going to be secure. And then, the other thing I'll just touch upon briefly is the major part of our announcement was that next major advance to unified data storage. That's with the introduction of the new NetApp AFF A-Series systems. And in the context of ransomware, it is built in. All of these capabilities that we just talked about, they are built into the new NetApp AFF A-Series.

>> That was the thing is performance and data protection. That was what I took away from it, and it really helps everybody that it's built in. And in fact, we were talking to Gabie earlier on tonight, today because it isn't tonight yet, that around how basically it's built in. And I think everybody likes that. It goes back to people understand how to work with the NetApp array, and it's how that works all over the place as well.

Sandeep Singh

>> Yeah, absolutely. I'll touch briefly on the major portion of our announcement with the new AFF A-Series. What we are talking about is it's unified data storage built for the AI era. What that means is now basically the new AFF A-Series is powerful, intelligent, and secure. Powerful in the sense that it can help customers turbocharge every workload. It's better performance. It is up to 40 million IOPS. That's an astounding number. You and I have lived the storage .

>> I remember when we got to a million IOPS, it's like 40 million though. Yeah.

Sandeep Singh

>> Yeah, up to one terabyte per second of throughput across a cluster. All of that, when you couple that with six-nines measure data availability, always on data reduction there. Now, what that means is customers can turbocharge every workload without trade-offs, whether it's optimizing their VMware environment, accelerating their SAP, SQL, Oracle apps, or seizing the AI advantage and building the capabilities for AI, gen AI workloads. So, that's a powerful side of it from an intelligence standpoint, top of mind for everybody right now is lowering costs. Customers can easily protect, manage, and mobilize their data, and they can do it at lowest costs across the data lifecycle. They can simplify hybrid operations across the board, and intelligence is also part of building that foundation for enterprise AI. And then, secure part of it, you and I just talked about all of the ransomware side of it.

>> Yeah. No, I think that's great. But you also had another announcement, and I love this because I was around at NetApp when FlexPod launched. So, you have the AIPod, and you're doing it this time with Nvidia and Lenovo, and we're going to welcome them up in a short little while. But give a little bit of the high-pitch on why. Why AIPod?

Sandeep Singh

>> Absolutely. Look, overall, AI, gen AI, every organization is looking at how to harness the power of AI and gen AI, whether it's to drive productivity gains, to deliver that next level of innovation, or deliver that next level of customer experiences. So far, we've seen a lot about larger-scale AI as a service providers for large foundational model training. That's an expensive investment across the board. As enterprises are graduating and looking of graduating from experimentation to production deployments. There, the use cases are shifting a lot more to beyond model training into fine-tuning, retrieval-augmented generation and inferencing. And really the power for enterprises is to be able to take their enterprise data and combine that with the pre-trained gen AI models to make it more contextual for them. And this is why we teamed up with Lenovo and Nvidia to build a new conversion infrastructure. This is the new AIPod with Lenovo ThinkSystem servers. Alongside with Nvidia OVX certification. It's designed to enable enterprises to take their enterprise data, use RAG with it, with this conversion infrastructure. It's super simple, so that it's pre-validated. It can be fast time to value for customers. It's a great value proposition for partners to help deliver it to their customers along the way. It's powerful because it brings along the ThinkSystem servers, it brings along the Nvidia L40S GPUs with it, and it brings along the NetApp the AFF C-Series across the board. You've got data management built in. You've got the enterprise AI, NeMo Retrievers overall software stacks built in from Nvidia. So, think of it as a gen AI, converge gen AI in a box. And it's affordable for customers. So, as organizations are looking at deploying and production deployments, affordability absolutely matters.

>> Yeah. No, we see this. Most organizations aren't going to go against and try to compete against the large LLMs. What they're going to do is bring a model in like an open source model into some of the Nvidia systems or partner with Nvidia, with the models that they have, and really ride that curve. And what we've seen is actually in our power law distribution for gen AI is that, yeah, the big models are here, they're huge, but as you start to go out where the vast majority of the customers are buying for, it's for verticalized instances such as finance or telco. And when you get to telco, hey, you're going to put something out at the edge at a base station or something like that to do inference and to be able to be responsive. I think that's the kind of thing where you need these types of AIPods that really go and fit.

Sandeep Singh

>> Yeah, absolutely. We are super excited about this AIPod with Lenovo and Nvidia. It augments our overall portfolio of solutions that we're building and enabling customers with around AI and gen AI. Look, NetApp hosts a lot of the customer's unstructured data that AI and gen AI are built on. And we see it as being that good steward and shepherd of enabling customers to be able to truly harness the power of AI and gen AI.

>> And like you said, you're out talking to customers all the time. We talk to them in our research, and one of the things they always talk about is they don't necessarily want to bring all their data to the AI. They want to bring the AI to the data. Is that a common theme that you're hearing as well?

Sandeep Singh

>> Absolutely. I would say first, gen AI, it absolutely builds upon unstructured data, so that becomes important. If you already have the unstructured data stored on NetApp, for example, it becomes the opportunity to go, how do we enable AI, gen AI with it? Bringing data or AI to data is critically important. Lots and lots of that data has gravity. At the same time, it's archetypical workload for hybrid workflows. This is where NetApp has invested over a decade in really integrating at this cloud-native level with the hyperscalers. And we have a variety of technologies to enable customers to securely and efficiently be able to mobilize their data. And that's part of enabling those hybrid workflows. And what's becoming essential, especially for the production workloads in enterprises, is all about how do I have data management tied for enterprise AI? Model versioning, the data sets associated with those models, the data lineage, the data security and governance and privacy of data just becomes so critical. It's just a top of mind. I was in a conversation with a CIO of a financial services provider, and that was just the top of mind, critical part of it. So, these are all elements that are feeding into important needs for customers across the board.

>> Yeah, especially when they're looking at things like the European AI Act and other stuff coming down the pipe. But why don't we park it there for a moment, because we have some special guests that are going to come up on stage with us in a minute, and we'll talk through that in a second. So, why don't you stay right there and why don't you stay right there while we bring up those special guests. We're going to bring up Nvidia and Lenovo, and we're going to dig a little bit deeper into the AIPod that was just announced today. So, you stay right there with theCUBE, the leader in high-tech news and analysis. Thanks.