Data Gravity and IT Evolution with Nutanix’s Lee Caswell

This is Textron tv. Hey guys, thanks for the throw. We're here with Lee Caswell, who is Senior Vice President for Product and Solutions marketing for Nutanix.

And we're talking about how cloud computing's gonna be involving because, well, we're seeing all this cloud native stuff, and there's a lot of it, and things are getting a little more complex than we possibly thought was gonna be the case this time last year, or even the year before that. Lee, welcome to show. Yeah, thanks so much, Mike.

Happy to be here. What is going on from your perspective? At least I feel that, you know, the first generation of the cloud was lift and shift.

I'm gonna take my stuff that runs on a VM and a monolithic environment and make it run somewhere else, or, you know, you're a mess somewhere else, as they say. Right? Right.

But now it feels like we're moving towards these cloud native technologies. There's Kubernetes out there, there's serverless computing frameworks, there's all kinds of interesting stuff happening. What's your assessment of where are we in this adventure?

You know, I think over time, Mike, and you've seen this in over times, right? You know, we centralize and then decentralize over time, and there are forces that press you one way or the other. Um, certainly the cloud was a really interesting time, particularly around Covid, for example, right?

And it was difficult to collaborate in things, and it was kind of a fresh beginning, right? From the sense of a lot of new technologies, right, were initiated in the cloud, right? And so what we're now seeing though, is as customers start reevaluating where they would optimally place not just applications, right, but work, you know, but the, the data associated with those applications.

Now I've got a distributing manner, you know, part of it is ai, but you know, that's a little bit future, right? But the idea is that now all of a sudden I'm gonna start looking for performance cost, data sovereignty, privacy even. Where should my applications and data be?

Where should they exist? And so now what customers are looking at is, hey, some of those cloud native activities, right? You know, Kubernetes containers, for example, right?

And even several, some things are now moving back into the on-prem environment where initially customers thought it was gonna be a cloud only. It's my view In some ways, are we coming full circle on that? Because back in the day when you and I were probably much younger, they taught us that, you know, move the compute to the data because moving data around was a career threatening issue, right?

And we, nevertheless, we moved all kinds of data into the cloud, but now we seem to be figuring out that maybe we wanna actually, uh, cons process and analyze the data at the point where it is being created and consumed. So have we come full circle? You know, I think data, people always realize, right?

That a distributed model is the more likely case, and particularly as you see the growth in the edge, right? And it turns out like sensors can generate data a lot faster than people. And so as you look at the amount of data, right?

Up to 50% of data is gonna be created at the edge, you know, by some analyst, uh, calculations within the next two years. Now all of a sudden you do start looking at, well, where, you know, how can we move the applications to the datas 'cause it's impractical, right? And probably not very, not very secure either to, to centralize all your data in one place.

So a distributed model certainly looks more likely. And by the way, right, you know, people were looking at their on-premises data centers, maybe not this center of data anymore, but they were looking at that data center and thinking, you know, that's where I'm keeping all the systems of record still. And so for things that are, you know, uh, let's call 'em, you know, more easily planned, um, you know, you start looking at the cost and the ability to manage things.

You, you know, we're certainly seeing that you're gonna have access of data distributed now in a way that yeah, probably looks like, um, you know, how, how we thought about it, uh, you know, 10 to 15 years ago, Can I have my cake and eat it too? And by that I mean, can I, uh, decentralize the data but centrally manage the whole environment so that I can get the level of cost and scale that I need? 'cause I cannot have management consoles everywhere.

Yeah. Yeah. Well, then you start looking at, well, what are the common things that would allow you to have data and applications move across this, what we call hybrid multi-cloud, by the way?

So hybrid being to one cloud, multi-beam to multiple clouds. And so it turns out, right, there's two standards now, right? If you think about a data standard across hybrid multi-cloud, we believe it's S3, right?

So S3 is a, you know, it's a publicly available standard, right? That you can go and write to and now move data and move it very cost effectively also, right? As an, you know, from in an object store, um, um, protocol.

Similarly, by the way, on the applications front, right? There's Kubernetes, right? And so for an application standpoint, Kubernetes is a portability standard really to make applications.

Both of these, by the way, interestingly enough, right? We're open source largely, or, you know, close enough, um, by, you know, Amazon and Google respectively as ways to bring data and applications into their hyperscaler properties, if you will, right? But what we're finding is now these are now standards, and you'll see Nutanix leaning in because once you move to a server base, modern infrastructure, um, context or construct, now you're able to run identically with a software layer that allows you to grow and run the same way you would run on servers that you own or are now on servers that you rent in AWS and Azure, for example, in our case today.

So that's a really interesting model where you started thinking, oh, I could go and have data and applications be portable across this new multi-hybrid cloud, and I only have to have one team that can manage it. Is the term cloud computing somewhat obsolete? Because we aren't going back to this distributed architecture, and I have to wonder if I scratch my head, we have cloud and cloud native, and yet I see Kubernetes clusters out at the edge as well.

So what the heck does cloud mean these days? Yeah, I think cloud was, you know, it is an original cluster. It was something that you ephemeral, right?

You know, amorphous that you looked at and said, Hey, I can't really see any individual part. Now what you're starting to think is, you know, instead of having a cloud right now, the cloud is actually now an expansive multiple clouds. And you're starting to look at, hey, the edge should be integrated with the data center proper and then into the public cloud.

And, you know, the, you know, the new, uh, AI workflows actually show this in a really interesting way, right? Where you'll probably develop large language models, right? Of, you know, a billion or a trillion parameters in the public cloud, but then you're gonna wanna basically start retraining those with your own private data in a data center environment, and then run them in an inferencing manner, or even do, you know, rag on these now, right?

So, um, retrieval augmented generation as a way to go and basically get your model pointed to new data. And so these workflows right now starts, you start thinking, Hey, how do I make sure that I've got the same operating model across that? And we at Nutanix, since we're a data center company, really by heart, we also think that the common way to basically manage data services, because it'd be like snapshots, replication, disaster recovery, actually has a really interesting thing.

If you can have a common way to do that across all these different endpoints, now you've got a really interesting way to go and get some standardization in how you go and protect the data across this, uh, hybrid multi-cloud environment. Is the environment getting too complex for us to manage? It's just too challenging.

We don't have enough people, we don't have enough skills, or what do we need to do to kind of lay the foundation to achieve that goal? Yeah. Well, I, I, I think it's true.

You know, we had a long history of trying to basically make the amount of storage expertise that you have kind of minimal, if you will, right? You know, we brought storage, you know, by eliminating proprietary hardware storage systems, right? Where you had to go and, you know, know each one individually and have different architectures by the way, for files, blocks, and objects, by, by consolidating all that right into a server-based system, that was a really, you know, an interesting way to go and kind of eliminate some of the complexity that you used to have.

And, you know, I'll give you an example, Mike. I mean, you remember when you had an SLR camera and you had a, you know, an F stop, right? You know, lenses and you know, what kind of film were you gonna use?

I mean, there's a lot of expertise around that. And, and, you know, everyone's a, you know, a professional photographer without any of that knowledge these days. So that was one degree of simplification.

Another degree of simplification was, as you pointed out, right? It's not all VMs anymore, right? You got VMs plus containers going to serverless, right?

And so how do you get the same operating model across those different constructs? It's one of the things we do here as well, right? So you can run containers and, um, VMs with not just a con, you know, as con a constant runtime, but also with d data services that are, um, identical across those.

And then finally, the ability to do that, right? In an on-premises environment, which could be the edge, right? Or on-premises data center, and then do it similarly in the public cloud.

These are simplifications that allow you to say, all right, I'm not sure exactly what I'm gonna need, but I'm gonna have a architecture that allows me to get the degrees of freedom I'm most likely to exercise over time. Nutanix just picked up D two IQ, or at least the assets thereof. Um, is this gonna be the year where we finally manage Kubernetes at that level of scale?

Because it seems like, at least from my perspective, we've been talking about it now for more than half a decade. And, you know, I'm still running into people who are like, this is their first day on it, and they're like, what do we do with this thing? And how does it work?

But, um, are we gonna be managing fleets of these Kubernetes clusters soon? Yeah. Well, we're thrilled to have the D two IQ team on board, uh, you know, a great set of, uh, Kubernetes and container, uh, expertise right?

To, to join the company. Um, I was just pretty early, uh, in the timeframe to go and, you know, show how we're gonna use that, uh, directly, but certainly like the management of Kubernetes, right? Is something that's, um, you know, complex and we see that team, um, helping, particularly when, you know, early on, Mike, a lot of customers thought their containers were only gonna be in the public cloud stateless containers, right?

And that was gonna be, you know, no, I, I don't need any containers on prem. Thank you very much. Right Now what's happening, right, is you're watching this distributed, you know, the, the, the accordion kind of, uh, you'll go back out again, is people are looking and saying the value of containers is really high.

You know, it's very high from a development standpoint to speed the pace of development. It's also important in a hybrid multi-cloud environment because you don't know precisely where your application will run. So eliminating OSS dependencies becomes a easy way or a faster way to deploy in a distributed manner, right?

So what we're seeing is customers are now waking up and saying, Hey, I am gonna have containers on-prem, and to the extent that they containerize any existing applications they have state. And so what you're gonna watch, right, is the data services that we have are exceptionally important. If you take a database today, put it on containers for development reasons, and now you want to go and make sure you've got snapshots and replication, and Dr.

Applied to that stateful container, you know, we've got great assets around that, right? And, you know, at the same time, we've got great, um, relationships with our development partners too, like Red Hat for example. Of course, that's not the only disruption going on in the world these days.

One of your rivals is being acquired by a very large company out there. And, um, what are you hearing from, who could You be talking about, Mike, What are you hearing? What are you seeing from folks out there in terms of that transition?

Because, well, it's been a lot of folks who've been long-term customers of VMware. So are they calling or are they basically standard pat? Well, um, you know, I don't think they're just ducking and covering.

I mean, I think, you know, what, what's happening really is, uh, customers are going through a three phase, um, you know, um, AC acceptance process, if you will, right? So one is, you know, they're nervous and they're nervous because Broadcom has a history of taking companies, whether they're hardware or software by the way, and, you know, basically taking a lot of money out of the existing, or at least the top tier of customers and not actually caring very much about the others, right? And so the idea of, um, you know, what happens to those customers who have l little choice or, you know, it's not easy for them to move.

And certainly that's the case with, you know, vSphere for example. You know, one of the, I think the best products, you know, in, in IT history, right? Um, and so customers who are using vSphere, for example, to connect to a san, you know, listen, Microsoft tried for years, right?

With Hyper-V you got single digit, um, market share Citrix tried with Zen server, right? You don't recall that, right? So, you know, it's really a very sticky project, project or product in part because of the, you know, just a huge complexity of the San compatibility matrix of making that of that work.

It's kind of a long way of saying that customers realize that they're a little bit over a barrel near term in what they could do in the vSphere environment. And so for those customers, a lot of them have basically tucked in a one year or three year, or even a five year ELA, right? An agreement to contain the first thing they cared about, which was price risk.

But now what's gonna happen very quickly, right? Is now they're gonna be worried about support risk. It's a very different support model, Mike, for supporting, you know, ongoing or, you know, real time mission, critical infrastructure than it is doing a backup product or a security product, or a chip design.

I mean, this is where, you know, companies lifeblood, particularly when you think about applications as the new competitive currency for any company, right? Your loyalty is based on your ability to, you know, interact in, in, engage with a company on, on their apps today. So, you know, we think the support element is really gonna be, um, the one that where, um, you know, a lot of customers get, um, um, get hung up to your question, what are they gonna do?

Well, I think what customers are gonna do is they're gonna ring fence what they have today, just like they did with Unix systems, Mike, they didn't ever throw 'em away, but all the new applications, everything new goes on, new infrastructure. In that case, it was X 86. In this case, we think a lot of it's gonna come to Nutanix with some going to the cloud as well.

So we're at the end of the year looking into 2024. What's your crystal ball telling you? A crystal ball?

Well, you know, I think we're really bullish on not just techs, you know, tech spending overall. We think, you know, people are gonna continue to invest in tech. And to the extent that these, you know, they invest, we think they're gonna invest in software ahead of even hardware, right?

And so why is that, right? And the reason the way you can do that is that you can differentiate more quickly on the software elements of your business. And so we think the idea of a server based architecture, right, is the modern infrastructure path.

We're super excited about being able to take our software and help customers deploy applications, new applications, right? Whether it's, you know, from databases to analytics into, uh, a generative AI and deploy those wherever they optimally should reside. And that could be, you know, from the edge to the core to the cloud.

So one thing we see is software-defined server-based architectures are gonna expand dramatically. And you know that VMware's gonna help with that, uh, in some, some sense. Maybe that's number one.

Does that, does that make sense to you? Makes sense to me. All right.

Yeah. I had one other by the way, right? So this, go ahead.

You know, accordion reaching out, right? Or the pendulum swinging back the other way around on distributed architectures, we see that ai, you know, has got the interest of every, every industry right across every industry of verticals, across any of the horizontal industries that we have, right? Every company is looking at how to have AI be a differentiator.

And we've launched something called this GPT in A Box, which is a way for customers to take GPU enabled infrastructure and quickly bring that, you know, a new application, a new AI application. So we think 2024 is gonna be the area of getting your first AI application up and running with an idea of then running in volume in 2025. All right, folks, you heard it here.

The great IT paradox is still with us. On the one hand, it is changing faster than ever. On the other hand, the more things change, the more they stay the same.

Hey, Lee, thanks for being on the show, Mike. Best to see you always. All right, back to you guys in the studio.