NetApp Unveils Unified Data Storage Built for the AI Era | Sandeep Singh | NetApp, Bob Pette | NVIDIA & Kamran Amini | Lenovo

theCUBE.net

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

Sandeep Singh

SVP & GM, Enterprise Storage

NetApp

Kamran Amini

VP & GM, Lenovo Servers & Storage Solutions

Lenovo ISG

Bob Pette

Vice President and General Manager, Enterprise Platforms

NVIDIA

Three insights you might have missed from theCUBE’s coverage of the ‘Unified Data Storage Built for the AI Era’ event

AI is only as good as the data that fuels it, and for intelligent data infrastructure company NetApp Inc., this new era calls for a unified data approach.“Now more than ever, companies, businesses are feeling this pressure to accelerate innovation and improve employee productivity, unlock new customer experiences,” said theCUBE host Rebecca Knight.For NetApp, intelligent data infrastructure and unified data storage play an important role in empowering companies to do what they’re under pressure to do. What comes next for NetApp’s ecosystem was a central focus during last week’s NetApp Unveils Unified Data Storage Built for the AI Era event

From blueprint to action: The AI Pod initiative by NetApp, Nvidia and Lenovo

Many have likened the rise of generative artificial intelligence in 2022 to the iPhone’s release. As the next evolution frontier in enterprise computing, gen AI is changing how businesses operate and innovate.Enterprises are moving from AI blueprinting to action, employing models on-premises. To facilitate those operations, NetApp Inc. has partnered with Nvidia Corp. and Lenovo Group Ltd. on the AI Pod initiative

play_circle_outline The Crucial Role of AI and Its Deployment in Revolutionizing the Market, Cloud, Governance, and Enterprises

play_circle_outline Partnership between NetApp, NVIDIA, and Lenovo to simplify the AI journey

play_circle_outline Maximizing Efficiency: The Convergence of NVIDIA, Lenovo, and NetApp Technologies for Seamless AI Integration

play_circle_outline Inclusive AI Solutions for Enterprises: Empowering Business with Easy Deployment, Compelling Attributes, and Ethical Compliance

Info

Sandeep Singh | NetApp, Bob Pette | NVIDIA & Kamran Amini | Lenovo

Sandeep Singh

SVP & GM, Enterprise Storage NetApp

Kamran Amini

VP & GM, Lenovo Servers & Storage Solutions Lenovo ISG

Bob Pette

Vice President and General Manager, Enterprise Platforms NVIDIA

search

>> Hello and welcome back to NetApp Converge 2024. Glad to have you on board with us. So excited about this. As I kind of teased a little bit, we have a panel. I'm joined back by Sandeep Singh, SVP and GM for NetApp. I'm also joined by Bob Petty, who's the VP and general manager for enterprise platforms at NVIDIA. And Carmen , the vice president and GM of Lenovo Server Storage and software. So welcome on board guys. This is, I mean, really huge announcement today. Fantastic. You can't get through, I think we'd be thrown off the internet if we didn't talk about AI today, and especially the fact that there's an AI pod. I think, again, we were talking and I'm like, hey, listen, I was here when FlexPod started back in the days. That's how old I am I guess from that perspective and how far I go back, but I think one of the things that has changed is we're a year and a half out from kind of that ChatGPT moment. People are looking at Gen.AI and saying, okay, great. I wanted to test the waters. Maybe I got into some pilots and proof of concepts, but now what? What are you guys seeing from a market perspective with GenAI and how it's really being deployed?

>> Yeah, so I'll take a stab at that first. We've seen a tremendous boost in the cloud in terms of building large language models, as you mentioned, ChatGPT moment, the iPhone moment for AI. But we've established this enterprise platform groups at NVIDIA specifically to address what we see as the biggest opportunity and more than half 50, 75% of the AI in the world we see to be done on-prem. From a government regulation standpoint, a sovereign AI standpoint, IP standpoint, you want the AI running where your data is, and that's why something like the AI pod is so significant. So people have done the POCs. They might be a little concerned about do they need the same infrastructure as Google? And the answer to that, as you'll see, is no, but we definitely see the enterprise explosion beginning to happen now, and this effort I think is going to make it easy for people to get started and continue to grow in a graceful way.

Sandeep Singh

>> And I would just add to it, customers enterprises are absolutely looking at how do they harness the power of gen AI and make it work for them in the context of their enterprise. It becomes critically important for them of how do they take enterprise data, combine it with the pre-trained LLMs, and this is just that opportunity of providing a AI pod that basically packages the best of breed across NVIDIA, across Lenovo, and NetApp and provides that in a pre-validated, pre-integrated form factor to customers so that they can effectively just do GenAI or RAG in a box in that way.

Carmen Crincoli

>> Absolutely. Absolutely. And I think this is where customers that have tons of data, they want to get a business outcome out of that data, and this is where AI comes into play, and I think the partnership between the three of us here is about how do we make it easy for them and be able to get outcome-driven value with the solution we could deliver.

>> I think that is the outcome and what we see, and we partner up with a company called ETR. They do quarterly tech spending intentions, and one of the things they do is look at AI use cases, be it, "Hey, we want to do for software development. Well, I don't want my code going out into the public cloud and things of that nature. I want to do customer support or customer success, help with the people be more responsive to customers." And then there's also kind of the marketing, or I guess you could say messaging generation type of stuff that they're using. Those are kind of the three things. What are kind of the attributes that end user organizations will really find compelling out of the three of you coming together to build this AI pod? Why don't we start with Carmen?

Carmen Crincoli

>> I think it's about how we could bring the best of the technologies. As Bob implied, the early adopters of AI was all the cloud service providers that drove massive scale of large language model systems. Most of the enterprise customers that did not have the skills or the knowledge of what do we do around AI, it's about the partnership we could bring in to really simplify that journey. AI, at the old days of thinking cloud, it was very scary initially when you thought about cloud, what should I do at an enterprise customer? AI is the same concept, and I think we're trying to bring the value to simplify that experience and the journey between the partnerships we have here.

Carmen Crincoli

>> Yeah, no, absolutely. The work to do training and large language model training is very different from what we believe enterprises want to do, are doing, and will need to do, and that's taking those models, fine-tuning them, and then adding additional data to them, call it retrieval, augmented generation, or RAG workflows so that it represents their semantics, their language, their industry semantics, their company metrics and their data. Our NVIDIA AI enterprise has been founded to really help the enterprise get to the data with things like Nemo Retriever that NetApp has integrated into their 800 storage platform to set up guardrails because care about privacy and security, they want to take advantage of all the latest and greatest features, running on the Lenovo platform and accessing data on that app. So we have NIMS microservices that are updated on a regular basis so they can get the finely tuned AI for their respective use cases. And I think just the culmination of a fantastic platform that Lenovo has in the management stack, NVIDIA GPUs and our OVX certified reference platform with L40S GPUs, AI enterprise, and the coup de grace, the piece that I've been salivating over is the integration with NetApp, with Nemo Retriever. You own half the enterprise data in the world? It's close.

Sandeep Singh

>> Yeah. Majority of the enterprise that are storing their unstructured data sets on NetApp.

Sandeep Lahane

>> So if you're building a chatbot for customer service, that chatbot's got to access your customer service database, it's got to access your CRM, your sales database, you're not going to upload that to the cloud.

>> Yeah. I guess what you're really getting at is kind of what the secret sauce is that you're each bringing to this, and I think why don't we kind of jump into that because you started going down that.

Sandeep Singh

>> Yeah, and maybe I'll add a higher level first. I'll just also articulate that we have had a long rich history of partnering with NVIDIA. At the same time, Lenovo and NetApp have had a five-year amazing partnership, and we're all together taking this to the next level. It's bringing that best of breed across compute and GPUs and the right storage infrastructure and then combining it with the necessary software stacks and overall gen AI capabilities that organizations need. That's what this ultimately AI pod represents, making it easy to use, bringing the power along to go and do retrieval, augmented generation, fine-tuning, or inferencing and make it affordable for enterprise customers

Sandeep Lahane

>> And easy to build upon. What I love about the AI pod is people may think, like I said, they have to go build this massive infrastructure, and at maybe some point they will, but they can start with the right size, continue to add use cases, and as data grows and use cases grows and value grows, they've got these very easy deployable chunks of these AI pods that they can deploy throughout the enterprise.

Carmen Crincoli

>> Yeah, I think the key thing we're trying to do as Lenovo and the partnership we've established with both NetApp and NVIDIA is bringing AI to all, because again, AI is a journey and it's a different size and deployment for every size of customers from a small medium to all the way large enterprise customers, and it's how do you make it easy? How do you take the best of breed of technologies, put it together for the application, generate RAG and inferencing, and actually be able to consume and deploy and get the outcome that you're looking for? And that's the value we're bringing together three of us.

Carmen Crincoli

>> Yeah, no, I think it's key because we actually talk about it back in September, we came out with this power law of GenAI where all the way on the left here... On the left here? Left, that would be my left. Their right. You really have all of those large language models that are getting built, the ChatGPTs, all of the ones, the Titans and every other one that the cloud guys have, but that's those guys. They're building those. Most companies are not building huge LLMs. What they're doing is taking an open model, taking a Llama to a Llama 3, fine-tuning it, and they're bringing it on-prem and then fine-tuning it for financial services because I want to help my customer support people understand how to answer a question when it's complex, still keeping human in the loop to your point. And that's got to be a piece of it with the guardrails and how you bring it all together with the open models.

>> Yeah, no, absolutely. We want ethical AI first and foremost, and we want to protect data privacy, and then we have a slew of regulations that we're going to have to adhere to. That's why the data and the access to the data and these REG workflows is so important. Who can access it? When can they access, what information can they get to? And then what guardrails do you place on how far you let AI imagine so that you can avoid hallucinations and any unnecessary outcomes. I want to touch on one point in terms of bringing that infrastructure down is we collectively designed this to ensure that it could easily fit into the enterprise footprint. So it's got our SpectrumX Ethernet, it's not InfiniBand like you typically see in large scale. It's got SpectrumX and InfiniBand. We're putting it into 10K, 12K racks, not 100K powered racks that you see in the cloud. So it's really meant to be an easy drop into your existing infrastructure and get going.

Sandeep Singh

>> Those are fantastic points too, just reinforcing that the design center here is for enterprises to be able to easily consume it and bring their data and do RAG with it and leverage AI with it.

Sandeep Lahane

>> Yeah, no, I think that's a great place. Kind of leave it from that perspective. I think it is about the use cases. And I will ask one last question. If we're sitting here a year from now after this guy is done talking and start talking about this and say, hey, listen, here's what success has looked like for the last year. What are you hoping to see for next year?

Sandeep Singh

>> Look, I think it starts first and foremost with helping customers, being that strategic partner to customers and enabling them to truly go and harness the power of GenAI and AI. That's the beginning point for the success, what success looks like for the AI pod. Secondly, I will add is basically for our overall partners, this is a fantastic opportunity for partners because partners are the go-to-market motion here. They are the ones that now also get to take their customers and help them harness the power of GenAI while delivering all of the pre-integration services at the same time. So there's an absolute metric here of making sure that this is a successful outcome for our overall go-to-market partners. And then for the three of us, this just continues to represent, in my view, of that collaboration that customers are expecting as vendors where we are the stewards of enabling this enterprise and responsible AI for them.

Sandeep Lahane

>> Yeah, no, I think it's a great point. Certainly I would like a year from now, I'd like no one to be fearful or afraid to implement. People should be starting now and a year from now, they should have demonstrable proof across a number of different use cases within, and it's not just about making more money and making more profit. It's about improving lives, improving productivity. To go all visionary, it's curing cancer, it's curing world hunger. It's enabling underserved communities by bringing this technology down and automating some of the things that allow people to focus on improving the quality of their lives. But first and foremost, I'd like people to accept this as the way of computing because it is. It is going to be the way of computing from here on out and it's fear no more.

Carmen Crincoli

>> I think it's echoing both. It's about being considered the trusted partner for our customer's journey in AI. And as Sandeep said, it's bringing our channel partner community as they're the other arms that are engaged with our end customers mutually to really enable delivering the power AI to drive different business outcomes for our customers. So we're very excited.

>> Yeah, I love leaving things on a very positive note, I'm very positive about GenAI and AI in general, and I think, again, this is, I think really exciting news that you guys are launching today. And I think just walking around last night and talking to partners, they're excited about all the news over this week here and are really leaning in on this, and I think it's great. So thank you for having us here, and thank you for being on with us. And again, hope to have you back next year when we're talking about the successes.

Carmen Crincoli

>> Sounds great.

>> Thank you.

Sandeep Singh

>> Yeah, thank you, Bob. Thank you, Carmen.

Carmen Crincoli

>> Thank you. Appreciate it, Sandeep.

Sandeep Lahane

>> And thank you for watching The Cube here. We're talking about tech, analysis, and news, and we're the best at it, so stay tuned. And we've got more to come from Converge 2024 with NetApp.

Sandeep Singh | NetApp, Bob Pette | NVIDIA & Kamran Amini | Lenovo

search

>> Hello and welcome back to NetApp Converge 2024. Glad to have you on board with us. So excited about this. As I kind of teased a little bit, we have a panel. I'm joined back by Sandeep Singh, SVP and GM for NetApp. I'm also joined by Bob Petty, who's the VP and general manager for enterprise platforms at NVIDIA. And Carmen , the vice president and GM of Lenovo Server Storage and software. So welcome on board guys. This is, I mean, really huge announcement today. Fantastic. You can't get through, I think we'd be thrown off the internet if we didn't talk about AI today, and especially the fact that there's an AI pod. I think, again, we were talking and I'm like, hey, listen, I was here when FlexPod started back in the days. That's how old I am I guess from that perspective and how far I go back, but I think one of the things that has changed is we're a year and a half out from kind of that ChatGPT moment. People are looking at Gen.AI and saying, okay, great. I wanted to test the waters. Maybe I got into some pilots and proof of concepts, but now what? What are you guys seeing from a market perspective with GenAI and how it's really being deployed?

>> Yeah, so I'll take a stab at that first. We've seen a tremendous boost in the cloud in terms of building large language models, as you mentioned, ChatGPT moment, the iPhone moment for AI. But we've established this enterprise platform groups at NVIDIA specifically to address what we see as the biggest opportunity and more than half 50, 75% of the AI in the world we see to be done on-prem. From a government regulation standpoint, a sovereign AI standpoint, IP standpoint, you want the AI running where your data is, and that's why something like the AI pod is so significant. So people have done the POCs. They might be a little concerned about do they need the same infrastructure as Google? And the answer to that, as you'll see, is no, but we definitely see the enterprise explosion beginning to happen now, and this effort I think is going to make it easy for people to get started and continue to grow in a graceful way.

Sandeep Singh

>> And I would just add to it, customers enterprises are absolutely looking at how do they harness the power of gen AI and make it work for them in the context of their enterprise. It becomes critically important for them of how do they take enterprise data, combine it with the pre-trained LLMs, and this is just that opportunity of providing a AI pod that basically packages the best of breed across NVIDIA, across Lenovo, and NetApp and provides that in a pre-validated, pre-integrated form factor to customers so that they can effectively just do GenAI or RAG in a box in that way.

Carmen Crincoli

>> Absolutely. Absolutely. And I think this is where customers that have tons of data, they want to get a business outcome out of that data, and this is where AI comes into play, and I think the partnership between the three of us here is about how do we make it easy for them and be able to get outcome-driven value with the solution we could deliver.

>> I think that is the outcome and what we see, and we partner up with a company called ETR. They do quarterly tech spending intentions, and one of the things they do is look at AI use cases, be it, "Hey, we want to do for software development. Well, I don't want my code going out into the public cloud and things of that nature. I want to do customer support or customer success, help with the people be more responsive to customers." And then there's also kind of the marketing, or I guess you could say messaging generation type of stuff that they're using. Those are kind of the three things. What are kind of the attributes that end user organizations will really find compelling out of the three of you coming together to build this AI pod? Why don't we start with Carmen?

Carmen Crincoli

>> I think it's about how we could bring the best of the technologies. As Bob implied, the early adopters of AI was all the cloud service providers that drove massive scale of large language model systems. Most of the enterprise customers that did not have the skills or the knowledge of what do we do around AI, it's about the partnership we could bring in to really simplify that journey. AI, at the old days of thinking cloud, it was very scary initially when you thought about cloud, what should I do at an enterprise customer? AI is the same concept, and I think we're trying to bring the value to simplify that experience and the journey between the partnerships we have here.

Carmen Crincoli

>> Yeah, no, absolutely. The work to do training and large language model training is very different from what we believe enterprises want to do, are doing, and will need to do, and that's taking those models, fine-tuning them, and then adding additional data to them, call it retrieval, augmented generation, or RAG workflows so that it represents their semantics, their language, their industry semantics, their company metrics and their data. Our NVIDIA AI enterprise has been founded to really help the enterprise get to the data with things like Nemo Retriever that NetApp has integrated into their 800 storage platform to set up guardrails because care about privacy and security, they want to take advantage of all the latest and greatest features, running on the Lenovo platform and accessing data on that app. So we have NIMS microservices that are updated on a regular basis so they can get the finely tuned AI for their respective use cases. And I think just the culmination of a fantastic platform that Lenovo has in the management stack, NVIDIA GPUs and our OVX certified reference platform with L40S GPUs, AI enterprise, and the coup de grace, the piece that I've been salivating over is the integration with NetApp, with Nemo Retriever. You own half the enterprise data in the world? It's close.

Sandeep Singh

>> Yeah. Majority of the enterprise that are storing their unstructured data sets on NetApp.

Sandeep Lahane

>> So if you're building a chatbot for customer service, that chatbot's got to access your customer service database, it's got to access your CRM, your sales database, you're not going to upload that to the cloud.

>> Yeah. I guess what you're really getting at is kind of what the secret sauce is that you're each bringing to this, and I think why don't we kind of jump into that because you started going down that.

Sandeep Singh

>> Yeah, and maybe I'll add a higher level first. I'll just also articulate that we have had a long rich history of partnering with NVIDIA. At the same time, Lenovo and NetApp have had a five-year amazing partnership, and we're all together taking this to the next level. It's bringing that best of breed across compute and GPUs and the right storage infrastructure and then combining it with the necessary software stacks and overall gen AI capabilities that organizations need. That's what this ultimately AI pod represents, making it easy to use, bringing the power along to go and do retrieval, augmented generation, fine-tuning, or inferencing and make it affordable for enterprise customers

Sandeep Lahane

>> And easy to build upon. What I love about the AI pod is people may think, like I said, they have to go build this massive infrastructure, and at maybe some point they will, but they can start with the right size, continue to add use cases, and as data grows and use cases grows and value grows, they've got these very easy deployable chunks of these AI pods that they can deploy throughout the enterprise.

Carmen Crincoli

>> Yeah, I think the key thing we're trying to do as Lenovo and the partnership we've established with both NetApp and NVIDIA is bringing AI to all, because again, AI is a journey and it's a different size and deployment for every size of customers from a small medium to all the way large enterprise customers, and it's how do you make it easy? How do you take the best of breed of technologies, put it together for the application, generate RAG and inferencing, and actually be able to consume and deploy and get the outcome that you're looking for? And that's the value we're bringing together three of us.

Carmen Crincoli

>> Yeah, no, I think it's key because we actually talk about it back in September, we came out with this power law of GenAI where all the way on the left here... On the left here? Left, that would be my left. Their right. You really have all of those large language models that are getting built, the ChatGPTs, all of the ones, the Titans and every other one that the cloud guys have, but that's those guys. They're building those. Most companies are not building huge LLMs. What they're doing is taking an open model, taking a Llama to a Llama 3, fine-tuning it, and they're bringing it on-prem and then fine-tuning it for financial services because I want to help my customer support people understand how to answer a question when it's complex, still keeping human in the loop to your point. And that's got to be a piece of it with the guardrails and how you bring it all together with the open models.

>> Yeah, no, absolutely. We want ethical AI first and foremost, and we want to protect data privacy, and then we have a slew of regulations that we're going to have to adhere to. That's why the data and the access to the data and these REG workflows is so important. Who can access it? When can they access, what information can they get to? And then what guardrails do you place on how far you let AI imagine so that you can avoid hallucinations and any unnecessary outcomes. I want to touch on one point in terms of bringing that infrastructure down is we collectively designed this to ensure that it could easily fit into the enterprise footprint. So it's got our SpectrumX Ethernet, it's not InfiniBand like you typically see in large scale. It's got SpectrumX and InfiniBand. We're putting it into 10K, 12K racks, not 100K powered racks that you see in the cloud. So it's really meant to be an easy drop into your existing infrastructure and get going.

Sandeep Singh

>> Those are fantastic points too, just reinforcing that the design center here is for enterprises to be able to easily consume it and bring their data and do RAG with it and leverage AI with it.

Sandeep Lahane

>> Yeah, no, I think that's a great place. Kind of leave it from that perspective. I think it is about the use cases. And I will ask one last question. If we're sitting here a year from now after this guy is done talking and start talking about this and say, hey, listen, here's what success has looked like for the last year. What are you hoping to see for next year?

Sandeep Singh

>> Look, I think it starts first and foremost with helping customers, being that strategic partner to customers and enabling them to truly go and harness the power of GenAI and AI. That's the beginning point for the success, what success looks like for the AI pod. Secondly, I will add is basically for our overall partners, this is a fantastic opportunity for partners because partners are the go-to-market motion here. They are the ones that now also get to take their customers and help them harness the power of GenAI while delivering all of the pre-integration services at the same time. So there's an absolute metric here of making sure that this is a successful outcome for our overall go-to-market partners. And then for the three of us, this just continues to represent, in my view, of that collaboration that customers are expecting as vendors where we are the stewards of enabling this enterprise and responsible AI for them.

Sandeep Lahane

>> Yeah, no, I think it's a great point. Certainly I would like a year from now, I'd like no one to be fearful or afraid to implement. People should be starting now and a year from now, they should have demonstrable proof across a number of different use cases within, and it's not just about making more money and making more profit. It's about improving lives, improving productivity. To go all visionary, it's curing cancer, it's curing world hunger. It's enabling underserved communities by bringing this technology down and automating some of the things that allow people to focus on improving the quality of their lives. But first and foremost, I'd like people to accept this as the way of computing because it is. It is going to be the way of computing from here on out and it's fear no more.

Carmen Crincoli

>> I think it's echoing both. It's about being considered the trusted partner for our customer's journey in AI. And as Sandeep said, it's bringing our channel partner community as they're the other arms that are engaged with our end customers mutually to really enable delivering the power AI to drive different business outcomes for our customers. So we're very excited.

>> Yeah, I love leaving things on a very positive note, I'm very positive about GenAI and AI in general, and I think, again, this is, I think really exciting news that you guys are launching today. And I think just walking around last night and talking to partners, they're excited about all the news over this week here and are really leaning in on this, and I think it's great. So thank you for having us here, and thank you for being on with us. And again, hope to have you back next year when we're talking about the successes.

Carmen Crincoli

>> Sounds great.

>> Thank you.

Sandeep Singh

>> Yeah, thank you, Bob. Thank you, Carmen.

Carmen Crincoli

>> Thank you. Appreciate it, Sandeep.

Sandeep Lahane

>> And thank you for watching The Cube here. We're talking about tech, analysis, and news, and we're the best at it, so stay tuned. And we've got more to come from Converge 2024 with NetApp.