Why “Cloud-First” Is Failing—and What Comes Next

Hey everyone. Welcome back here to Tech Drunk tv. I want to introduce you to my next guest.

It's the first time he's been on Tech Drunk tv, so we'll, interested to hear his story and, and get to know him. His name is Indu p Indu as his friends call him Indu, uh, Keri, KERI, uh, indu as his friends call him is the SVP and GM for hybrid Multi-cloud at Nutanix, one of our companies we cover a lot. Indu, welcome to Textron tv.

It's great to have you on here. Thank you, Alan. It's great to be here.

So, indu, we were talking a little bit in the green room. I, I don't even know why they call it a green room, but we do, you know, we do it with a green screen, but we in the green room, we were talking a little bit and what we're gonna talk about. And I, I mentioned that, you know, with a title of SVP and GM for the hybrid multi-cloud business over at Nutanix, one would think this, this gentleman's pretty technical, but maybe not give, give, give our listeners a sense of your journey, of your career, how you came to be the SVP and GM here.

Yeah, let me answer that in a couple parts. Um, I'm trained as an engineer. I have a degree in computer science.

That's what my undergrad was. I also have a, uh, PhD computer science where I did my work in compilers. And maybe Alan, I'll leave you this, uh, funny anecdote.

Um, before I finished my PhD, I worked on the compiler at, uh, Silicon Rapids Really? And, and in 2011 I actually received a royalty check for $3,500 for third that I had written in 1998 as part of the, uh, SGI high-end paralyzed compiler. Really?

Very cool. Yeah. Yeah.

It's, it's one of those things I've just decided to frame and I'll give it to my son someday. Absolute. You know, I was talking to someone last week, this gentleman, he was one of the original three program product managers for Windows, for Windows nt, not the original Windows, windows nt, but still Windows NT is 96, 97, something like that.

Yep. And, um, and we were talking, so back then I had started my very, that was the first company I founded. It became, it was before we called it hosting, but eventually it became called web hosting.

And you know what great choices we had then? We had the Sun Ultra Spark machines, you had the silicon graft, the SGI workstations and, you know, and, and then Windows came along. And quite frankly, we laughed at it, didn't we?

Right. It was like, eh, it's a toy compared to these things. Well, you know, here we are 30 years later and uh, we don't have SGI as great as those machines were, and I I'm still a sun person at heart.

Right. But Windows and Linux took over the world, didn't it? Yeah.

I think it's really hard to argue with, uh, Moore's Law and mm-hmm. I think the relentless execution that Intel had over so many years, I think led to the, um, the victory of cyst over risk. I think it's in the Yeah.

In between 2000 and 2015. But now with the argue of that, it's the revenge of the wrist, right? Yeah.

Yeah. If you live long enough into the worm turns right? Absolutely.

And the wheel goes round, round and round as they say. I agree with you, man. It's, uh, it's interesting.

It really is. But as you say, I mean, look, we're, we're at the end of the year here. We're, we recorded this right before the end of the year.

Most of other audience will see it right after it's certainly been hardware is, hasn't been this sexy, I think since you were working on SGI. Right. Or hardware now is, I mean the, the, the semis and, you know, GPUs and, and inference chips and everything else.

It it's, it's crazy. Um, I mentioned you run SVP GM hybrid Multi-cloud at Nutanix. Yep.

Let's first talk about Nutanix a little in, do me, 'cause I I think most of our audience certainly has heard the name Nutanix. They may or may not know the whole story. Let, let's, you know, let's assume someone said, oh, I've never heard of Nutanix.

And do what, what do you guys do? I would say we are one of the more interesting companies that nobody, or maybe not as many people have heard of as they should. Our vision, which is all around enabling the intelligent work placement of workloads anywhere, um, is another way to think about it is, you know, we were one of the first companies to come up with a hybrid multi-cloud vision, you know, in, um, for the first, I would say 15 years of the cloud from 2007 to 2020 or so, I think the cloud went from being interesting for dev test to eventually it became a bit of a religion where everybody had a cloud first agenda, regardless of whether it was a right one or not.

But even as late as 2019, we had come to the realization that you are going to have a pretty broad set of deployment infrastructures. Cloud will be one of them, but not the only one. And so in 20 19, 20 20, we came up with this notion of hybrid multi-cloud, where we basically said, regardless of your workloads, whether they're VMs or containers, regardless of where you want to run them on your data center at the edge or in the public cloud, we are going to be the middleware provider that's gonna make that possible.

And I think that vision has really helped us, um, draw our business. And it has helped us solve a problem that in many ways today has become a bit of religious debate. On the one hand, you have Broadcom saying that you the only need the private cloud and nothing else, and you clearly have the public cloud providers that, um, drive primarily adoption of the public cloud.

But the reality is, customer environments are not just one or the other. They are myths of both private and public clouds, and in many cases multiple public clouds. And so the problem they're really trying to solve is how do I place my workload, whatever the best place might be, whether it's virtualized workloads that, you know, we have all been familiar with for the last 20 years, or whether it's modern workloads like containers.

Um, what is the best way to position them in the right place? And that's what we do for them. I love it.

You know, I, I've got a confession to make just between you and I, I always knew that hybrid cloud was going to be a dominant form factor, right. Because I, I never thought people would move everything to the public cloud, right. With few exceptions, right.

Moving everything to the public cloud. Now, if you were a greenfield and starting from scratch, maybe, but, you know, those are the exceptions. Yep.

But I never saw multi-cloud coming until it ran me over. Right. The idea of, you know, being both on Amazon and Azure or Azure and Google or, or whatever combination you want and throw Oracle or whatever else you want in there, you know, I never thought it would be as popular as it is, but when you think about it, it makes sense because each of the, of the hyperscalers, each of the public clouds, they all have their little specialties.

You know, this one does cloud native code better, this one does, uh, DevOps or something better. That one is better, it's serverless or, or what have you. And um, so it does make sense because the lesson there is people don't want to be locked up.

Mm-hmm. They don't want to be locked up into one platform, into one form factor. And, and so they liked the idea of being able to pick the right tool or the right cloud for the right job.

Mm-hmm. And, and, and so that's the world we we live in today. Now you mentioned a few companies here.

Look, to be fair, the, the, the folks at Broadcom, VMware, whatever, whatever how they pronounce it, and I don't know if you call it Broadcom or VMware or Broadcom, VMware or whatever, they have sort of tried to make an overture for people who wanna take the Broadcom platform to public Right. And try to run it there as, and then of course the public cloud provider said, if you're gonna run it here, we we've got better tools for you all that. I think what the market really wants though, du, is I don't want five different providers from my platform.

I'd like a platform, I'd like that platform to run anywhere. Right. And everywhere, whether it's the edge, right?

The, the, the, the hyperscaler data center, my own data center, multiple data centers, multiple hyperscalers. I, I can't afford to have five different teams. Right.

That's why your, in, in some ways, your title makes perfect sense. You run one unit that's for hybrid multi-cloud, you don't have a unit for a WSA unit for Google, a unit for private a unit. You know what I mean?

I I I don't think the market really wants that other than Nutanix, though. There aren't a lot of solutions that, what's the right word, that are Switzerland enough, right? Neutral enough to, to play everywhere you want to be, You know, Alan, um, I don't think I sort of said our value proposition better than you.

What, what you just did. Maybe I think it's luck. I may need a job.

Holy kidding. Um, and I think it's actually really worth reinforcing the top of things that you said, right? One of the foundational principles that we use to operate is we really believe in giving our customers choice.

Mm-hmm. And that has been true throughout our history. So for example, um, at the storage layer, we offer our own version of software defined storage, but we also support external storage from, for example, Dell Power Flats or Pure Storage.

Mm-hmm. As a compute layer, we support our own hypervisor, but obviously we also support ess i similarly from a cloud perspective on a, on a really run your workload perspective, we let you run your workload on-prem. Obviously we've been doing that for years, but we also support AWS Azure.

And as of this month gcp. Really Totally, And you actually said something that's very insightful, right? Each cloud does have its own sort of special niche.

You know, Alan, I don't know if you remember, but like, I think this was 80 years ago or something where Thomas Watson said, the world needs five computers. And the thought of true, right? I, let me just be very clear, I wasn't alive then.

Okay. But, but we, the quote I've heard, I've heard he said that We've read the quote, right? Right.

We read it. And so the five are like called AWS and Azure and GCP and OCI and private cloud, right? Mm-hmm.

And what is absolutely true is every one of them is sort of optimized for a particular sort of experiences, right? Yeah. If you start with Linux workloads, a WSI think is by far the most natural choice.

And then serverless, if you start in sort of a Windows world, I think especially things like server windows, those sorts of things, Azure becomes an interesting sort of, um, default choice. If you have data intensive applications, GCP becomes a very, very interesting choice, right? Um, and so what customers really want is they're very workload driven and they want to be able to use the public clouds for what the public clouds are ideal for, which is for very, very elastic applications for experimentation.

And for that phase where you're not quite sure what your usage patterns are gonna look like, right? Where there are a lot of unknowns, either known unknowns or unknown unknowns, where you're trying to figure out what does this do when this workload blows up, right? And there is a lot of that that needs to be done.

And once you fiddle that out, um, quite often it turns out that you wanna move your workload somewhere else. And the really hard part of doing that, honestly, is not compute. It's the data.

Yeah. The data has gravity, right? I agree.

You know, when something goes down, you can restart a VM or a container somewhere else, but if it doesn't have the data that it needs to operate, um, then you're dead. And that's where we really shine because of the origin story of the company. We are able to handle data availability and data migration and data availability wherever you want, almost in a, um, sort of under the surface kind of a way where as a user, you establish policies that around data availability.

And we made data available wherever it needs to be. And that's an incredibly important part of how we deliver the seamless hybrid multi-cloud experience. Um, and that's, and you're absolutely right.

I mean, I would say that really is no other player out there that can deliver this sort of an experience for both VMs and containers running their workloads wherever is the best place for them. Absolutely. And I leave you with one other thought, Alan, which is, um, our exceptional NPS.

So we really are customer obsessed. And the net promoter store for the company throughout its history has been north of 90. And that is, that's great.

Incredibly, uh, uncommon for an interface. World class. World class.

Absolutely. No Doubt. I mean, I, that's the definition of it.

You know, a concept that the Nutanix people talk about and that kind of resonated with me is this idea of cloud first to cloud fit, right? And, and, and it, and, and it's, there's a nuance there, right? And a distinction between you want to be cloud first 'cause you want to be one of the cool kids have had it, but there's more to just being cloud first.

It, it's, it's being cloud, right? Let's call it right? By having the right fit there.

And, and this, this can change, right? Because your, your needs, your uses change in today's world, it changes every day it seems, right? So it's gotta be sort of a living, breathing sort of fit, right?

You may, you know, take a little in here, put a little out there. Um, talk to us about this notion of cloud first to cloud fit. That's a great observation, Alan.

And in fact, um, I'll, I'll tell you, in a prior life before I came to Nutanix, I was at Intuit mm-hmm. Where I helped move turbo turbos onto the, onto the public cloud. Now, turbo tats is a very interesting application.

Um, as you might imagine, five days of the year account for 80% of the usage. In fact, the joke that used to be that the gear that ran turbo tats used to be idle 95% of the time, it used to be 95% idle, 95% of the year. And it turns out that's a wonderful application for the public cloud.

The elasticity of that just makes it wasteful for you to have dedicated hardware for that application just to take care of the workload four or five times a year. But most applications are not like that. And so what cloud, right, a cloud fit really does, is to look at an application and a workload and it's life cycle.

There are typically three phases when you're building the new workload. The first step is you're trying to establish product market fit. You're doing a lot of, um, experimentation and speed really matters.

And quite often if you go to go to internal IT and say, I need five servers, they'll come back to you in three months. And that's just not the pace that you want for that sort of rapid experimentation. So the public cloud is a fantastic place for that.

Now, once you have found product market fit, there is a phase that you go through for a workload, which is typically scaling the usage. You found 10 users, you found 50 users, now you need 5,000 users. And that also is a phase where you need access to new capacity really, really fast.

And public clouds tend to be really good for that sort of a, uh, stage in the acquisition life cycle. And then there is sort of the third stage, right? You have not, you, you know, if you, if you compare the first one to sort of the, the right after we were born, then maybe the teenage years, and then you become an adult, right?

And once you become an adult as an application, sometimes you're incredibly elastic like TurboTax, but sometimes you're just like, you know, your usage is pretty much the same. Maybe there's a 5% up and down, 10% up and down. And at that stage, what you realize is that maybe the public cloud is where I started, but it may not be the best destination for my workload.

And at that point, you start to look at alternatives. And I'll give you an analogy. Imagine you wanted to drive for the next seven years, and what you do is you could do one of two things.

You could rent a new car every day, or you could buy a car tomorrow. Obviously a rent in the new car every day will be much more fun, but it'll also be much more expensive. But there is also a downside with buying the car tomorrow, which is you might buy something that you may not like, but here is where you let the best of both worlds.

You rent a new car every day for 30 days, find out what you really like, and at the end of the 30 days pitch the one that you love. Right? And so that's sort of a mobility between, of workloads, between sort of experimentation in the public cloud to a place where you can manage the steady state is one of the big drivers for why you need application mobility, right?

You also made the point earlier on about multi-cloud, you know, every cloud has, you know, a different performance. It has a different footprint in terms of data centers. It has different exposure to, you know, natural disasters and events that you cannot control.

And one of the ways to hedge your beds is in fact to say that my failure, my failover for one public cloud might be another public cloud. And so this is where you think of the public cloud and the on-prem private cloud as different tools that you have to make your overall application portfolio more resilient and more available. And that's when you get away from religion and really start to be very nuts and bolts about what is the best way from me to get the most return on my investment?

And that's what we really advocate for. I love, I love that du, that that's really, I think that's a great way of looking at it. Um, because these, as I said, these are not static.

Our applications today are far from anything but static. Yep. Um, and you know, I, I've been in technology 30, 35 years.

One of the things, the lessons I've learned, the hard way I probably have a couple of t-shirts to prove it, is you, you can't build portability in when you realize you need to move or be portable. You've gotta build this in, right? I early in the process, right?

You've gotta kind of design this to say, Hey, what happens if I need to move? What happens if we want to go this direction or that direction? And you know, people, you and I, we've been around the block, we, we've seen this all too often though.

We see people who, you know, they die on that hill, right? Because that's the only hill they knew. Yep.

How does, how does Nutanix or how can Nutanix help them with that? That's it. And Alan, I think this is, it, it it almost feels like you and I have been sort of doing these battles together, right?

Um, there are two incredibly important ways where we help here. The first one we already talked about, which is making sure that data protection and data availability is so built into the platform that as a user, you can make sure that in the event of something bad happened, your data is available wherever you need it. And you do that upfront in terms of poli establishing the right policies, and the platform gets clear of that for you, right?

And that's incredibly important, right? Because if you've lost your data and if you have never done anything to make it available somewhere else, there is nothing you can really do till the primary systems come back up, right? So you really, data availability is one of the most critical foundational aspects of maintenance happen.

The other one is that our platform is fully software defined and driven by APIs. So really one of the big changes that has happened in infrastructure software over the last 15 years is that we have really separated out the data layer from the metadata layer, and all the control and management operations are separated from the data path. What that allows you to do is to really separate out the control operations from how the data flow rotates place and have APIs that drive the behavior of the underlying system.

So for example, um, compared to, I mean, as a contrast to some of the other alternatives out there, when we support our stat on AWS or Azure or GCP, it's literally the same software stat that runs, that runs on-prem. It's not a managed service, it's not a service that different set of APIs, it's not a different control plane, it's not a different management plane, it's not a different interface. The experience that you have is literally the same as if you're running in what we're used to.

And so that consistency, that underlying architecture that separates out, you know, the data operations from the control operations and the management operations, I think is intrinsic to how we have been able to make this happen. And that's really the reason we have been able to do this so effectively, I got one other little thing I wanna mention and then one other topic, uh, we could wrap up with. So Indu, you mentioned elasticity earlier.

Look, let's be honest, for most of our users out here, cloud elasticity only goes one way. It's like a timeline, right? It's like the balloon you blow up, you may let the air out of the balloon, but it's never as tight as it was before you blew it up.

How is, how's Nutanix helping with elasticity issues? And, you know, you may never get that balloon is tight, but you know, the, the idea that workloads do go up and down in real life, The, the dium is I think the cloud providers have all actually built incredibly effective tools that allow you to manage and really, uh, make an assessment of the utilization of the underlying resources. So again, I think this is where the visibility and transparency to the underlying infrastructure makes a huge difference.

And there, there is a whole finops, uh, discipline where you go look at your spend and you can go optimize that spend. For example, if you don't have enough, uh, CPU utilization or if your data, for example, if you have too much of data sprawl and if you're spending too much on storage, then you can go optimize that. Um, I think the cloud providers actually understand that it's in their interest to make this easy, because the more that you drive sort of, um, efficiency in consumption, uh, paradoxically drives more consumption.

Um, and so I think because again, much like, uh, our platform, most of the cloud providers, all the cloud provider platforms are primarily software defined and provide a lot of visibility in terms of, um, in terms of usage and cost. Uh, if you are thoughtful about it, you can actually get the most out of your cloud real estate. Excellent.

So I can't believe we made it this far into our discussion, and we really haven't talked about ai, it's a new year, people are watching this in the new year, everybody's ai. First, how is AI impacting all of this, and how is Nutanix kind of, you know, internalizing it and, and it's reflected in the product offering? You know, Alan, if I may, I would like to give a short answer now and maybe have a deeper follow on conversation, uh, in the new year.

Anytime you'd like. I'll give you my 30 seconds. Okay.

AI is the single most interesting hybrid multi-cloud workload to have come along in a long, long time ever. And it's going to completely change the nature of infrastructure. That sounds more like a tease than an answer, but I'll take it and I'm gonna hold you to it.

You can come back on and let's discuss this. I would love to, Alan. Um, it's been an absolute pleasure.

Thank you so much for your time. And, uh, thank you. I look forward to speaking again in the new year and wishing to you Happy holidays.

Yep. And a happy holidays to you and all the folks at Nutanix. And welcome back everyone to Tech Drunk tv.

It's gonna be a great year. It sounds like with interviews like this. Indu, thank you for coming on here.

We, we are gonna have you back on. You owe it to us. Now.

You teased us about it. We're going to talk AI in hybrid and multicloud, but for now, this Alan Shimmel for Tech Drunk tv. Thanks for joining us.

We'll be back. We've got more text Drunk TV coming at you.