Karissa Breen [00:00:15]:
Welcome to KB On the Go. This week, I’m on the ground at NetApp Insight 20 24 Conference of the MGM Grand in the heart of Las Vegas. For this bonus series, we’ve been lucky enough to have lined up conversations with the selection of NetApp executives and other guests exploring the future of data and AI. Stay tuned for the insight track from some of the world’s leading authorities presenting at insight 2024 as KBI Media brings you all of the highlights. Joining me now in person is Haiyan Song, executive vice president, intelligence services at NetApp. And today, we’re discussing CloudOps and the role and trends and the way forward. So, Haiyan, thanks for joining.

Haiyan Song [00:00:57]:
Thank you for having me.

Karissa Breen [00:00:58]:
Now I know you had a bit of a job title change, so I really want to perhaps let’s explore a little bit more about what is that role of intelligent services. And then how does that play in intelligent data infrastructure then?

Haiyan Song [00:01:13]:
You know, the title is really just a reflection of the evolution of NetApp’s intelligent data infrastructure vision coming to life. You know, I was super excited to be able to come in front of this audience this week and talk about the evolution. And when we had CloudOps, and that was part of our evolution as a company going from hardware to expanding to cloud. And in order to get to really build a cloud business, you’ve got to understand how cloud operates, and you got to talk to the right personas and build things so you can understand how that operates. And now I think we’re at the phase that we understand customers look at their enterprise and their cloud as their intelligence data infrastructure that to really serve what they need to do, whether they’re in the AI journey, where where whether they’re in a digitization journey. It’s the food data infrastructure that would help them, would modernize, you know, their processes and things. So we evolved to be the intelligence data infrastructure company. And what constitute that is the unified data storage, which is the very foundation.

Haiyan Song [00:02:25]:
But you can’t operationalize that infrastructure without all those intelligent services who manage data to get the best value of infrastructure. So the title change, I think, is really just a indication of the company’s evolution.

Karissa Breen [00:02:39]:
So what I’ve seen as an undertone throughout this conference is how the customers and sort of the the outside market views NetApp in terms of evolved. Yep. Would you say as well that that people’s version of that is changing around from your just a storage company to so much more. We’re cyber speed podcast, as you know, and there’s been a very big narrative around cybersecurity. So was there anything you’d like to share in terms of that evolution and where sort of NetApp’s headed from your perspective?

Haiyan Song [00:03:09]:
Yeah. I I think this is the conference I met many customers. And in the feedback I’m hearing is a while. This is the time that we see how all the different pieces come together. And we love the fact that you’re using intelligence data infrastructure to articulate how each part of your technology is helping us in a holistic way. And cybersecurity was one certainly top of mind for many people, and it’s a passion of mine given, you know, my background in that area. And I think we are exemplifying to the field and to the customers that security doesn’t have to be an afterthought. And we have protection for ransomware built into our secure storage.

Haiyan Song [00:03:51]:
Right? People may have heard my conversation with the CTO from DreamWorks. And one of the things is we built ransomware protection, and customer can benefit from built in security so is peace of mind. And that’s a evolution as we’re not just storage. We’re actually bringing intelligence to data. We’re bringing intelligence to operations.

Karissa Breen [00:04:11]:
So talk to me a little bit more about intelligence to data. What what’s that sort of mean? And I asked that because people had different versions of what that means. So I was gonna start off on this, like, we’re walking, and singing from the same hymn book as they say.

Haiyan Song [00:04:25]:
Yeah. I think there’s several aspect I can talk about the intelligence for data. Right? One of the conversations that people are having is, well, we live in the world that data is exploding, and we live in in in the era of data and AI and intelligence. So use AI as an example. Intelligence of data means you’ve got to understand what data you have. Right? Because there’s a lot of conversations on data are the fuel for AI. Data is gonna help you with training, you know, whatever models you want. But if you don’t understand what data you’re using to train the models, then you would not have a good sense and the result, the outcome of your models and how useful, how biased, and how probably partial it can be.

Haiyan Song [00:05:12]:
So one thing that’s why we announced the data explorer to give people a view of, hey. Get a full view of all the data you have in your enterprise, and you can make better decisions on how you apply those data for your AI 20 as a use case. And the other part of the intelligence for data is the classification part. Right? Do we know if we have certain PIIs in the system? We got to provide different type of protection for the customers. So that’s the other, value of bringing that intelligent data to help with privacy, to help with AI training. Two examples. Okay. So going back to what data you were using to train the model.

Karissa Breen [00:05:52]:
Now I’ve spoken a lot in the past, but also over the last few days around, like, hallucinations. Yeah. What concerns you about data in terms of, you know, the the hallucinations and people sort of deriving perhaps wrong insights or incorrect insights from perhaps not the right data.

Haiyan Song [00:06:09]:
Yeah. I think this has been some good conversations I’m having with customers, others on the quality of data. Do we have a understanding of the quality of the data that how they contribute to trainings? I think that’s probably gonna be the frontier of, you know, what people are thinking, what more intelligent data you can bring into your training versus what this one really helps, and that one is really noise. The other angle, I think, on the hallucination part, I don’t claim I’m an expert in that area, but AI is still in this stage. The development of AI in this stage that the human need to get sort of involved and validating and providing the correct sort of hey. This is hallucination and we what’s causing it? How do we reduce the the amount of sort of noise that’s causing this? I think there’s still that whole process. That’s why we talk about RAG and other things. The more you can make it more specific to a particular use a use case or domain, the better you can spot those hallucination and fix them.

Karissa Breen [00:07:12]:
So I made you

Karissa Breen [00:07:12]:
a comment before, Haiyan, around quality of the the data or data, as you say. Sorry. Yeah. How does someone understand if it’s high quality or not? What would that look like?

Haiyan Song [00:07:23]:
I think that is gonna be a new frontier that, we in the data’s, you know, industry have to really work on is you got to trace the lineage of this data was used to train this model, and this model is used in those cases. And there is some hallucination happens, and you go back and say, hey. This is causing some of the problems. And with us fixing some of the you know, or augmenting the data, now we got much better accuracy. And how do we really evaluate that process and give, let’s say, scores or metrics to start thinking about how do we measure the quality of a data in the context of AI. I think that’s a frontier that we need to go explore. I am not aware of any particular solutions to say, I I can give you the readings of the data, but you can see the need that you know, because you don’t have the luxury to say, I just gonna put every day every bits of data into the system is super expensive and could create a lot of noise. So I think that’s a frontier that, I’m excited that we can continue to help internally.

Karissa Breen [00:08:27]:
So, obviously, we’re still early days as we’ve discussed in terms of identifying the quality of the data. Yes. What about sort of the next 12 months? Yeah. Perhaps when I come back and I sit down with you again for Yeah. Now I’ve been inside 2025. Do you think that’s sort of evolved in terms of the approach and methodology to that?

Haiyan Song [00:08:44]:
I think so. I I think, you know, our data score is just the very beginning. Right? We’re starting with all the metadata and all the other things. I’m sure it was, you know, working with design partners and customers, and especially one of the other dimensions could be how do we help with the data mobility between enterprise and and, the cloud. I think you’ll hear us talk about the progression in multiple dimensions. And maybe let’s touch on a little

Karissa Breen [00:09:10]:
bit more about NetApp’s approach to cloud operations or cloud ops Yeah. And intelligence services. What are they gonna start to see now moving forward?

Haiyan Song [00:09:18]:
Yeah. One of the things I announced is how we’re, you know, releasing capabilities of intelligence services in tandem with our storage products. Right? That’s kind of the new beginning because we do believe that in order for you to operate your data infrastructure, you need more and more of the solutions and services to make it simpler, make it faster, make it more automated. And, so I think you’ll see us doing more of that releases in tandem, and you’ll see us to take more solution. Pay. You know, we go into the cloud and customer migrating their workloads to the cloud. How can we build end to end? You know, I talked about we’re there for every step of the way. How do we connect that? So people who are thinking of going into the cloud don’t think of that as a major, major undertaking, and they can have tools and and and partners along the way to make it simple or easier.

Haiyan Song [00:10:13]:
And and also sometimes people want to repatriate. Right? We want to get that flexibility. So you’ll see us really bringing more automation, bringing better solutioning sort of tandem releases. Those probably are the 2 things I I’ll call out.

Karissa Breen [00:10:29]:
So in terms of automation, is that what you mean by how to connect things, make it easier and simple? Is that what you mean by that?

Haiyan Song [00:10:34]:
Like, automation, like, for example, right now, you know, when people go into the cloud, they discover a lot of new things. Sure. They deploy their workloads. They realize, oh my god. This is, like, really expensive because we’re not able to utilize the best resources that’s there. And they have the data, but they don’t have the automation. So automation means, it will give you observability. We can give you corrective actions or auto scaling.

Haiyan Song [00:10:58]:
So some of those things will be interesting to to just continue to improve.

Karissa Breen [00:11:04]:
So in terms of, trends in intelligence services and cloud operations, what are some of the things that are impacting customers? Obviously, you just said you you spent last 3 days, you know, side by side with customers. You’ve interviewed one of them on the main stage. So what what is impacting some of your customers you can share with us today? Yeah. On a

Haiyan Song [00:11:22]:
verified level, right, you know, data gonna continue to grow. Data gonna just have more ways of impacting our day to day life through AI or through, you know, different ways of being leveraged. So I think the trend is how do we really turning data and into opportunities and turning the challenges of customers having to manage and and harness the value of data in a way that is working positively for them versus becoming something that keep them up at night. Right? So I think the data explosion, the data quality thing we touched upon a little bit. I think the other thing is really around. Right? So intelligence services is two aspects, the data and AI services, and then the infrastructure and workload services. I think the trend on infrastructure and workload is there will be more workload coming into the cloud. There’ll be more demanding workload coming to the cloud versus, you know, the initial simpler ones, lift and shift.

Haiyan Song [00:12:23]:
When you have more demanding workload, it’s the entire data infrastructure that supports workloads needs to be managed operated in the best possible way. So I think those are the demands that, we want to step up to and continue to help customers.

Karissa Breen [00:12:40]:
So in terms of more demanding workload, what what what are sort of the, you know what’s that doing to support customers with that? Sorry. For example.

Haiyan Song [00:12:50]:
Yeah. You know, we talk about some of the most demanding workloads. Think of SAP. Right? We have already have, SAP application running in our Azure in the cloud, for, ANF, as we call it, Azure NetApp Files. Mhmm. And we had customers talk about how because right? The this is the Japan Coca Cola, example. How because we’re natively integrated with the hyperscalers that they were able to do this in a few months. So that’s an example.

Haiyan Song [00:13:21]:
Another example around, you know, EBAs, you can imagine, those are gigantic, you know, files, and they, requires a lot of compute. So you really get to work the system that your infrastructure for compute, your infrastructures for storage, and all the things. And, also, they’re under a lot of pressure. You gotta do real time. So those are two examples I can give you. And, of course, AI is probably trimming rack, and I think that’s the most topical workloads.

Karissa Breen [00:13:52]:
I wanna sort of close our interview now, Haiyan, and asking one question around, what’s your definition of cloud ops or cloud operations? Yeah. Do you think that perhaps people use these turn interchangeably a little bit? What what what does that mean to you?

Haiyan Song [00:14:07]:
A 100%. So cloud operations, you know, rooted from when people started moving a lot of things to cloud. They said, hey. What is the cloud operating model? Because that’s different than what you’re doing on prem. And it’s more agile. It’s it’s, you know, blending the lines between developer and SRE. And it’s the rules, you know, and the tools for CICD, the continuous integration, continuous delivery. So cloud operations and and sometimes people accept that to say, hey, security is an important part of it.

Haiyan Song [00:14:40]:
It’s not as, you know, afterthought. So some people say, hey, it’s dev sec you know, dev sec ops. Yeah. So cloud ops, I think, is a general term to really think about operating your cloud infrastructure, cloud workload, cloud processes in the most agile, mostly integrated, and most secure way. That’s how I think about it.

Karissa Breen [00:15:06]:
Joining me now in person is Sandeep Singh, senior vice president, general manager, enterprise storage at NetApp. And today, we’re discussing enterprise storage ring front of mine. So, Sandeep, thanks for joining and welcome.

Sandeep Singh [00:15:15]:
Thank you for having me, KB. It’s great to be here.

Karissa Breen [00:15:18]:
Well, I wanna start with storage. Now it’s one of those things that perhaps gets relegated. What are your thoughts then on that?

Sandeep Singh [00:15:28]:
Well, look, in terms of storage, it’s one of those things that can be behind the scenes, but fundamentally and crucially important to customers, especially in today’s day and age where data has become the crown jewel. It used to be about, you know, where storage was all about just storing it and serving it in data to your applications. But now when you think about the role of storage, it has expanded dramatically, shifting from not only storing your data and serving it, to the applications, but how do you protect that data from ransomware and cybersecurity attacks? How do you protect the privacy and, you know, make sure your data is compliant to governance practices? And how do you enable customers, to be able to, benefit from the agility of having your data on prem as well as in the cloud? And finally, which everyone is realizing right now, how can you truly unleash the power of Gen AI and AI, which can only be done with your enterprise data? So fundamentally, even though it’s behind the scenes, it’s become absolutely crucial to every organization. But there’s, the other side of it, which is in terms of relegating behind the scenes from an operational standpoint Mhmm. That’s what you want. You want it to be simple and just behind the scenes and just, you know, it just works. It can be simply managed and protected and upgraded, where you don’t have to pay attention to it and don’t have to worry about it.

Karissa Breen [00:16:56]:
The reason why I wanted to start there is because it’s one of those things that perhaps out of sight, out of mind. And what I mean by that is it’s sort of like electricity. You go home, you turn the light, and it works. Although when it doesn’t work, you really notice it, and it’s going back to data and storage and things like that. It’s something that’s important. But perhaps, again, can’t really see it, and it gets sort of pushed to the back. Maybe not as such a sexy conversation as some of the other things in technology. So are you seeing that a lot as well in terms of we can’t see it, we forget about it?

Sandeep Singh [00:17:27]:
Actually, storage has become the front and center conversation for many, many customers now because the focus goes back to data. Data is the most fundamental important asset for every organization now, and it begins with having to store that data. But the whole security aspect of data is bringing storage at the forefront again because storage often becomes that last line of defense for customers, from ransomware cybersecurity attacks. And if you can protect and detect, right there in real time and do that with high degree of accuracy, you can enable a rapid recovery for customers. And, again, coming back to the AI conversation, that conversation is fundamentally, you know, tied to AI is really it’s a data problem, that customers need to be able to bridge the gap between AI and data. So storage is absolutely becoming front and center for customers now.

Karissa Breen [00:18:24]:
Makes sense. So then let’s go perhaps on your what your role and your background and your experience. What’s your view then on enterprise storage?

Sandeep Singh [00:18:33]:
Yeah. Enterprise storage is, you know, become the foundation for customers.

Karissa Breen [00:18:39]:
Okay.

Sandeep Singh [00:18:40]:
Right? Enterprise storage needs to always have the, you know, per high performance for customers, high high and consistent performance. It has to have that mission critical reliability. These are the high end capabilities that customers look for. It absolutely needs to ensure that it has this advanced data management, capabilities to be able to manage your data that is stored there. So that’s the foundation of it. But customers tell us continuously that they don’t wanna have to sacrifice getting the high end capabilities. They’re one of their other bigger challenges is complexity, and they need to be able to solve that complexity challenge both within the system and at scale. So what that means is they need simplified operations

Haiyan Song [00:19:27]:
Mhmm.

Sandeep Singh [00:19:27]:
And they need those simplified operations and and that they can bring the consistency of those operations across, their different application, workloads end to end. Those are some of the basic, elements of enterprise storage there.

Haiyan Song [00:19:44]:
Yep. Yep.

Sandeep Singh [00:19:45]:
And increasingly where customers need it is it’s gotta have the build built in security and designed for security from the beginning. It has to be designed to enable, their, you know, hybrid cloud for their data.

Karissa Breen [00:19:58]:
Okay. So you said before, simplifying operations. Now if I look at a bank I worked in a bank before, you’ve got very legacy systems, technical debt. How do you simplify operations then? That’s not an easy thing to do. It’s said than done.

Sandeep Singh [00:20:11]:
Easier said than done. You know, customers are struggling with complexity. When you double click on what is leading to that complexity, very quickly, it becomes complexity that’s compounded through bespoke infrastructure silos. What I mean by that is customers will have, file environments, their, VMware environments, their database application workloads, their AI analytics workloads, their secondary or tertiary workloads, and then they’ll also have cloud storage environments. And if every one of those is a bespoke infrastructure silo

Karissa Breen [00:20:46]:
Run.

Sandeep Singh [00:20:47]:
What it means for those banking customers or any large customer is they have inconsistent management while they’re facing talent and, you know, skills gaps. They have inconsistent automation. They have inconsistent data security model that they can’t trust in. They have inconsistent operational recovery workflows in the event of a disaster, and they have an inconsistent vendor experience. When you change that equation and say, what if I can bring a consistent management, consistent automation, consistent data security model, consistent, disaster recovery workflows? All of a sudden, that’s a game changer for customers, and only NetApp is, you know, enabling and empowering customers to do that, with the power of ONTAP.

Karissa Breen [00:21:35]:
So you said before around we talked about NIS by storage. Would you say that’s your definition of high performance?

Sandeep Singh [00:21:42]:
High performance will vary, in enterprise storage absolutely needs to deliver high performance capability to customers.

Karissa Breen [00:21:50]:
Of course.

Sandeep Singh [00:21:50]:
And at the end of the day, that high performance ends up becoming from the application perspective. Is the application able to go, at the pace that the application is consuming? And then, yeah, enabling customers to consolidate multiple workloads on shared infrastructure and still being able to, you know, get consistency. This is where our ability of where NetApp is helping customers is to be able to get high levels of performance in a single system. And through our technology, customers have the flexibility to scale out across a cluster, and they’re able to continue to scale the performance levels and dynamically low balance their workloads.

Karissa Breen [00:22:33]:
So I think we sort of just discussed. We’ve looked at, you know, the value, of enterprise storage. We’ve spoken about how things are forgotten. But more importantly, what do you think things are probably missed? Or what are some of the, myths that people have around enterprise storage from your experience?

Sandeep Singh [00:22:50]:
Yeah. Well, where I would, take that is enterprise storage, you know, the I would say that the changes that are taking place, the when you look at the security and ransomware, this is where I get the vast majority of questions from customers. And You’re

Karissa Breen [00:23:08]:
in cybersecurity podcast, by the way.

Sandeep Singh [00:23:11]:
The what’s happening is essentially that there’s a game changing, you know, proposition taking place, where essentially the detection in terms of ransomware is, you know, typically happening today and post process through backups, and that can be hours to days later, versus where we’re leading that change is to real time detection of ransomware attacks. Now if you can change that to detect in real time in seconds to minutes, and do that with high degree of accuracy where we’ve designed it for 99% plus accuracy, all of a sudden, you’re you’re minimizing the impact, and you’re minimizing the false positives, and it becomes the enabler then for rapid recovery for customers. That’s a game changer for customers, and that’s where we’re seeing just a fantastic adoption right now across the board.

Karissa Breen [00:24:07]:
So I wanna follow this a little bit more. So, obviously, you’re saying now customers are very focused on ransomware, which makes sense. Right? What I mean, it’s happening, probably why, see at NetApp, to discuss some of the cybersecurity perspective. But maybe, like, if your lens on okay. The reigns were itself, but then the other thing that’s coming up a lot from a media perspective and in the some of the interviews that I’m conducting is, business continuity as well as downtime. So if you’re running a manufacturing company, you had an issue Yes. Can’t deploy things, that’s one thing. But then in terms of your downtime, how much of an impact that has in terms of long term long tail impact?

Sandeep Singh [00:24:39]:
Yeah. Do

Karissa Breen [00:24:39]:
you think it’s something that customers are really starting to understand more, or you need something they’re considering perhaps?

Sandeep Singh [00:24:46]:
100%. Business continuity and high availability, is incredibly important for customers. If you think about it in terms of the mission critical workloads that typically run on block storage systems, what we just announced, for example, the new ASA a series. One of the core pillars of it in addition to simple, powerful, affordable, the powerful pillar is all about giving you the scale as well as the 6 nines availability guarantees alongside with all of the business continuity capabilities to be able to simultaneously have active active business continuity, across two sites, to be able to protect your application. So incredibly important, an absolute must have for customers.

Karissa Breen [00:25:34]:
And in terms of moving forward, I know, you’ve released a lot of new product announcements. They’re all over our site. But do you have any sort of high level thoughts you can share with us today?

Sandeep Singh [00:25:43]:
Yeah. First of all, you one of the core areas is, you know, we are bringing intelligent data infrastructure for customers, to empower them to solve the challenges, both across, their mission critical environments, both across their security and protection mechanisms, so as well as for AI Gen AIs. In terms of the specific areas of announcements, the way I would recap it, in enterprise and unified data storage. We, have rapidly expanded our portfolio, and what we’ve introduced is the new ASIA series. This is for helping customers with their VMware database application workloads and giving them this, you know, fundamental proposition. They no longer have to choose between simple and powerful. They can get the all of the simplicity, that they need, including simplicity at scale, alongside with all of the high end capabilities and, you know, giving them what is affordable, making it available for them to modernize to all flash and be able to do that within their budget. That’s one core part of the announcement.

Sandeep Singh [00:26:52]:
Second core part, talking about security, ransomware, with our, next version of ONTAP, just the real time detection with 6 nines or sorry, 99% plus accuracy. That becomes generally available. And in addition to that, we’re orchestrating the entire workflow for customers to protect, not only detect and then recover from an application workflow perspective. And then thirdly, the big one in terms of helping customers truly unleash the power of AI and Gen AI and be able to do that with their enterprise data, by helping customers get their data AI ready, as well as bring AI to their data to fundamentally help them saw, you know, solve this chasm between AI and data chasm. And that’s the promise that, George made on the day one keynote for customers. We’re doing that both for our on prem customers through just a series of announcements that we laid out, as well as from a cloud storage perspective working with, each of our hyperscaler partners to ensure that all of the 1,000,000,000 of dollars that the hyperscalers are putting into AI and Gen AI tools, customers can take advantage of that with their enterprise data.

Karissa Breen [00:28:11]:
And there you have it. This is KB on the go. Stay tuned for more.