NetApp INSIGHT 2024 | Hoseb Dermanilian & Krish Vitaldevara, NetApp & Monica Jain, J&J

Clips
News
More from NetApp INSIGHT 2024

Hoseb Dermanilian

Sr. Director, Global Head of AI GTM & Sales

NetApp

Krish Vitaldevara

Senior Vice President, Platform

NetApp

Monica Jain

Director, R&D Data Science

Johnson & Johnson Innovative Medicine

Three insights you might have missed from NetApp Insight 2024

As the demand for more efficient, secure and flexible infrastructure grows, organizations are increasingly turning to intelligent data solutions that simplify multicloud and hybrid environments. Highlighting these trends, NetApp Insight 2024 showcased innovations that unify systems into a single platform to meet diverse operational needs.NetApp’s Cesar Cernuda talks with theCUBE about the role of unified storage solutions in driving innovation and efficiency for businesses.“There’s plenty of data that lives on NetApp, and we have a big responsibility toward our customers,” Cesar Cernuda (pictured), president of NetApp Inc

How AI Centers of Excellence are transforming global operations at Johnson & Johnson

As AI infrastructure evolves rapidly, enterprises increasingly rely on AI Centers of Excellence to scale operations while maintaining governance and compliance. These dedicated hubs empower organizations to streamline AI model training and data management, enabling compliance across global operations.For example, Johnson & Johnson Innovative Medicine, in collaboration with NetApp Inc., is successfully using its AI Center of Excellence to balance scalability, security and innovation in healthcare, setting a strong precedent for using AI to enhance global operations

play_circle_outline Preparing Data Infrastructure for AI Adoption and Exploring Applications in Healthcare, Finance, and Entertainment

play_circle_outline Navigating Data Tsunamis and AI Challenges: A Focus on Governance, Security, and Efficiency in Organizations

play_circle_outline Transition from large language models to small language models for problem solving.

Info
Transcript

Hoseb Dermanilian & Krish Vitaldevara, NetApp & Monica Jain, J&J

Hoseb Dermanilian

Sr. Director, Global Head of AI GTM & Sales NetApp

Krish Vitaldevara

Senior Vice President, Platform NetApp

Monica Jain

Director, R&D Data Science Johnson & Johnson Innovative Medicine

On Day 2 of NetApp Insight 2024, a panel discusses the importance of AI and data infrastructure in achieving business outcomes. Krish discusses how customers across various sectors are using AI to leverage their data. Josep talks about the importance of preparing data infrastructure for AI adoption. Monica emphasizes the need for governance and security in handling proprietary data in the healthcare industry. Krish highlights NetApp's focus on providing the best infrastructure for AI, including the NetApp AI data engine and integrations with hyperscalers. Mon... Read more

explore Keep Exploring

What are some examples of the business outcomes that customers are aiming for with AI? add

What is the significance of governance in regards to security and efficiency when organizations consider implementing AI technology? add

What are enterprises currently focusing on in terms of AI strategy and data processing? add

bolt Powered by CUBE AI

Hoseb Dermanilian & Krish Vitaldevara, NetApp & Monica Jain, J&J

search

>> Hello and welcome back to NetApp Insight 2024 live from Las Vegas. We're kicking off day two here with some really incredible discussion around AI, data infrastructure, and what you really need to do to bring it together. I have a wonderful panel with me. I have Josep, who's the senior director, global head of AI sales and go-to-market at NetApp. We're also joined by Krish, who's the senior vice president of platforms at NetApp, and also they brought a wonderful guest and customer, Monica Jane, who's the director of R&D Data Science at J&J. This is, I think, again, last night you were on stage. It was a fantastic discussion you had there. People haven't seen it, they should go back and check it out. Great discussion. I think let's kick it off with Krish and really let's talk about some of the business outcomes that you're seeing customers are trying to aim for with their AI, because right now people are trying to get from POC to production with AI.

>> Oh yeah, so we talked to customers across our verticals, right? Media and entertainment, healthcare, EDA, financials, and they're all using AI to get full value of their data. This is about unleashing the power of their enterprise data using AI. When you talk about J&J and other customers in the healthcare industry, they are rapidly changing the pace at which the drugs come out. This kind of innovation is game-changing, right? It changes everything we do as humans, and some of the innovation is mind-boggling for us that we see every day.

>> And I think that to me is one of the big pieces. I mean, again, Josep like 75% of CEOs view AI strategic. I'm surprised it's that low. To tell you the truth, we see it being higher in the 80% in some of the numbers we're seeing, but only 37% say that their data infrastructure is really ready to support the needs. What are some of the discussions you're having with those organizations about getting ready?

>> Well, before we go to the question, I think it's important to lay out the landscape of our customers that we have today. So we have some customers who have built AI Center of Excellence like Johnson & Johnson, who are enabling their internal users, whether they are developers, data scientists across the globe to do different types of workloads, whether it's model training, and we've been doing this for six years. It's not that during the Gen.AI boom, NetApp suddenly decided to get into the AI business. So our long-standing relationship with Johnson & Johnson goes more than two, three years. And that's where we enable these customers to build different workload types of environment. And then you've got the newcomers, and those are the top 50, I would say, in the SMP where they are building their own infrastructure to train this large language model. So they're building super pods, super clusters, now they're calling it AI factories to generate tokens and all that. And then on the side you have the majority of the enterprise who is doing just basically using the large language models and available tools in the cloud. And then they need to bring their own data to make these models more appropriate to their businesses. So if you look at these three areas, they have separate or different requirements. Those who are building the large language models and building the factories, they need a super performing cluster. Their mission is to create an infrastructure that is dedicated to this work. And for that, we just announced that we will be enabling ONTAP for Super Pod very soon. We already have it today with our E-Series and BGFS. But for those who have been standing customers of ONTAP, they're going to love what we will be announcing. For those who will be doing cloud and RAG and fine tuning, Krish talked today at the keynote that we will be adding the AI data engine within ONTAP. That's going to bring enterprise. I think the key over these three days is enterprise. Enterprise needs access and data. It's not all of them will be building this large language models. It's not all of them will be building these factories. A lot of them will need us, our help to get them into where they need to be with AI. And then on the side, you've got the AI center of excellence, which we've been engaged for a very long time. For those certain of customers, you'll hear Monica talk about data privacy, data governance, data management, and then hybrid cloud connectivity is the key. That's where they need to prepare their infrastructures to get ready for this.

>> Yeah, I think you hit on a good point because I look at it and go, when I talk to organizations, they're really concerned about things like efficiency, security, how do they get to scale, right data, right place for the right AI? Because AI, funny enough was around when we used to call it ML and other stuff prior to ChatGPT in 2022 when everything became Gen AI. But a good place, Monica, from your perspective and what you see, how do you think organizations should really think about governance and security and efficiency when they look at AI?

>> Yeah, for sure. I mean, as what Krish and Josep has been discussing about how data infrastructure all look like, right? From AI perspective, governance is the key essential piece to take care of because something which we are exposing very proprietary data needs to stay secure, needs to stay within the compliance. So any industry, and especially in the healthcare industry like Johnson & Johnson, we are also very, very, I would say, putting our all efforts to make it very good, compliant, secure so that whatever we are exposing, whatever we are trying to analyze through AI, it shouldn't go exposed anywhere so that we can have our patients or any other information which should stay in the private environments, it should stay there. It is when we are working on the data management as a whole, data governance is an essential piece to it. Our leaders also have a focus, especially on the governance. We have a separate committee like ethics and compliance to take care of the data as well. So I think it is a very important piece here to take care what we are doing with our data. So this is why any such infrastructure which is helping you is to make your data safe and secure, compliant as per the HIPAA rules we in the life sciences follow. It is very essential piece.

>> And I think you can't get away from it because data is the lifeblood of AI. You can't do anything without it. And like you said, I was talking to a company that was different part of the healthcare industry and they do a lot with actual X-rays and CAT scans, and it depends on the angle of the CAT scan and all the metadata that's wrapped around it. Krish, coming back to you, data is really at the key of the intelligent data infrastructure. What are organizations, what are your advice to them to get ready for AI?

>> Yeah, so let me actually walk you how we think about AI and how we think about intelligent data infrastructure is going to help organizations achieve the full potential of AI. So when we talk about it, what we are doing is we are getting their infrastructure and data. They both need to be ready for AI. And we do that by giving them the best infrastructure for AI. Part of it is price performance. Part of it is power efficiency. Part of it is sustainability. Part of it is scale, but you want to be able to trust your infrastructure to give you that performance, but give you that performance on standard protocols. Because a lot of organizations also don't want to get into non-standard protocols or vendor lock-ins or create data silos. So it's incredibly important for us that as ONTAP, we give that best infrastructure, best performance over standard protocols like S-III and PNFS over standard ethernet. So our organizations love that. The other thing we are doing is we are building this NetApp AI data engine. This is game-changing. I think we're going to let organizations find a way to create structure over their unstructured data. This combined with the change detection and all the ONTAPisms that the organizations allow is going to redefine how the data mobility and the data mobility happening with security, privacy, and governance that Monica was talking about. We are also integrating vector databases. So this is how you bring your AI to the data in place, wherever they are. And Josep talked about our hybrid cloud integrations. There is no one who is 1P native in all the three hyperscalers. So when we are 1P native in all the three hyperscalers, we can take your data, make it available to the entire AI toolkit of these hyperscalers, and we're not stopping there. We are also doing end-to-end ecosystem integrations and bringing intelligent data services. So when we look at what we do as an intelligent data infrastructure company, it's a complete package. It's end to end, no silos, no lock ins. I cannot think of anyone who can actually do what we are doing as NetApp when it comes to enterprise AI.

>> And I think my favorite part of yesterday's keynote was the fact that you brought up Snapdiff because I'm like as a geek from back when I was here at NetApp, I look at it and go, that was one of the things. I looked at that and I said, hey, you're basically fingerprinting the data. You understand when it's changing. Monica, I think we all know it's a data tsunami that you have to deal with to find the right data, the right data sets and people practitioners like yourself. How are you really handling that data tsunami and what are kind of the efforts that you've gone through to in your team to hone that that people can learn from?

>> Yeah, for sure. I mean, if you see my role, I'm leading the data strategy and GenAI initiatives at J&J. So it is very strategically planned role because before going into GenAI space or rushing into the GenAI use cases, we have to solve some very stringent problems we have overall. But what we need to do is to have our strategy very robust build to make our data infrastructure ready for AI, as what Krish mentioned, that we really need to have over large data sets, findable, all those fair principles needs to be perfectly followed. It should be findable, it should be reusable, it should be accessible, and we would be able to then scale it. So how are we managing the large data and the tsunami of the large data is to centralize it. We are trying our best efforts to have one stop solution so that it's not multiple silo efforts or silo places our data scientist has to go and to find their data. So there should be one place. On top of it, it should be findable, accessible with all governance in place as well. Every day we get new requests to get the new data, and we have the legacy system. So it should also be well-connected. How the legacy systems with maybe the cloud systems with the on-prem data, because life sciences industry like us, they certainly have a lot of private data sitting under on-prem. So all should be well-connected. So data mobility shouldn't be an issue. It should be very flexible to have those data accessible to our data scientist as well. So what we are focusing with the tsunami of the data is to follow our fair principle to make our data infrastructure very robust enough. We are there yet, maybe no., we are working on it for sure because it just keep coming and we are focusing. Before rushing into the GenAI or AI as a whole, we are focusing on our data management to be perfect. It cannot be perfect for sure, but we are trying to do that with all governance control.

>> It's absolutely a journey. I mean, I don't think any... Show me a company who's perfect on it. I don't care, knowing a lot of them, even the ones who are building the LLMs and things of that nature. But it would seem like, again, with the focus on GenAI has been about these large language models, but I would generally... I'm just going to fathom the guess that you see small language models or SLMs as being more important to the work that you're doing and the types of what practitioners like yourself would be doing. Would that be accurate? That small language models, you would use an LLM, but you're going to ground it and fine tune it on your stuff and focus it.

>> Yeah, that's correct. That's correct. Those are kind of phase approach we do. All those large language models actually are breaking down into small language models which are combining and solving the problem because one large language model cannot solve the complexity we have, especially in the R&D space. So those needs to break down into small pieces in the small language models and then to combine with the bigger picture to solve the bigger problem. So we are breaking them into pieces to solve small problems, combining them together, and then we are solving the problem.

>> If we can add to that one, the one thing we are definitely seeing is this transition from LLM and the focus on the fine-tuning and model generation to actually enterprise AI. So there are enterprises like Monica was talking about, that are doing SLMs, but majority of the enterprises are also thinking about how do I bring that enterprise data so that which solution can let me bring my LLM, my toolkit to my data so that I can actually get to the insights faster? So when you talk about our strategy, that's why we really focus on the part of, hey, you get to bring your LLM, it could be your SLM, but bring it with your data, with your ecosystem and we'll put it all together. And the data tsunami one resonates with me because as we talk to organizations like Josep can add a lot to this one. The organizations that are successfully doing AI are the ones that are better prepared for this data tsunami. So this is where on ONTAPisms, like you talked about SnapDiff and being able to bring the data management technologies that we built over last two decades together becomes incredibly important in how you cut through this complexity because it's a pretty complex picture out there.

>> Oh, I would agree. I think that to me, this whole thing is coming around and I think that's a great place to leave it because next thing I would want to go down because you're hitting on it is when you break down those silos and things of that nature as agents and agentic technology, but we'll park that for another day. I appreciate the time that I've had with you today because this has been really great. So thanks for coming on board.

>> Thanks for having us.

>> Thank you for having us.

>> Thanks Monica.

>> Thanks Monica. Thank you.

>> Thank you for having me.

>> It's always better with a customer. I tell that to everybody. It brings it right and grounds it right down.

>> And tomorrow I'll be talking about cloning and snapshots too.

>> There you go. And I love that. I mean, always key and space efficiency as well, which I know.

>> Exactly. Space efficiency.

>> Yeah. Well, hey, thank you for watching this episode. We're going to be right back from NetApp Insight 2024, live from Las Vegas. We got so much more to unpack today. Stay tuned. We'll be right back.