Nutanix .NEXT 2026 | Ty Peavey, Dartmouth College & Deepak Goel, Nutanix

Clips
More from Nutanix .NEXT 2026

Ty Peavey

Director, Infrastructure

Dartmouth College

Deepak Goel

CTO, Cloud Native

Nutanix

play_circle_outline Dartmouth Migrates from VMware to Nutanix AHV After RFP and POCs Amid Contract Expirations

play_circle_outline From Three-Tier to 2U Hyperconverged: Simplifying the Data Center and Accelerating AHV Adoption

play_circle_outline Containers and VMs complement each other for isolation, scheduling, efficient GPU use

play_circle_outline Unified Platform for VMs and Containers: Simplifying Security, Storage, and Operations Across Hybrid Cloud and Edge

Info
Transcript

Ty Peavey, Dartmouth College & Deepak Goel, Nutanix

Ty Peavey

Director, Infrastructure Dartmouth College

Deepak Goel

CTO, Cloud Native Nutanix

In this interview from Nutanix .NEXT 2026, Ty Peavey, director of infrastructure services at Dartmouth College, joins Deepak Goel, chief technology officer of cloud native at Nutanix, to talk with theCUBE's John Furrier and co-host Alison Kosik about how a proactive VMware migration laid the foundation for running AI and cloud-native workloads at scale. Peavey recounts how Dartmouth seized a rare window — expiring VMware and hardware contracts converging simultaneously — to run a competitive evaluation that led the team to choose Nutanix with AHV over VxRail ... Read more

explore Keep Exploring

How did Dartmouth decide which next-generation hypervisor to deploy, and why did it choose Nutanix AHV? add

How did moving to Nutanix hyperconverged infrastructure change your data center hardware and operations, and how did your staff adapt to the new hypervisor? add

What is the role of virtual machines today compared with containers and Kubernetes, and how would you explain their respective benefits (isolation, security, hardware efficiency, lifecycle management) to C-suite executives? add

How can an organization simplify and unify operations and policies across containers, VMs, AI workloads, virtualization, security, and storage across on‑premises, hybrid‑cloud, and edge deployments? add

bolt Powered by CUBE AI

Ty Peavey, Dartmouth College & Deepak Goel, Nutanix

search

Alison Kosik

>> Welcome back to theCUBE. We're coming to you from Chicago and we are streaming live at Nutanix .NEXT. I'm Alison Kosik. I'm hosting the show today. My big co-pilot here is John Furrier. He's the co-founder and co-CEO of SiliconANGLE, theCUBE. And throughout the day, I think one of the biggest buzzwords that we've heard is migration. And we're going to put migration together and partnership together and all that matters in this situation.

John Furrier

>> Yeah. We have Dartmouth College coming back on theCUBE, Brian War was on before, pioneering with Mist. Dartmouth College is known for pioneering, taking risks that have good rewards and the good pragmatic strategy. And they're a leader in IT transformation projects. So this case study really highlighted the pioneering of really the early mover on migrating off VMware to Nutanix, which people thought was crazy two years ago. Or, what? Four years ago. I'm looking forward to this conversation-

Alison Kosik

>> Already....

John Furrier

>> because this has been a hot topic.

Alison Kosik

>> Yeah, let's bring in our guests. We have Ty Peavy, the director of infrastructure services at Dartmouth College, and Deepak Goel, CTO of cloud native at Nutanix. Thanks so much for joining us, for stopping by theCUBE. I want to start with you, Ty, because I think we can first tell a story about the process that you went through at Dartmouth to go through that migration and maybe the pushback that you got, and then maybe the outcome of that.

Ty Peavey

>> Yeah. Dartmouth, we were in a really unique situation where both our contracts with VMware and with our hardware vendors were coming to term at the same time, and so we had a really good opportunity to think about what we wanted to deploy as our next-generation hypervisor. We already knew we wanted to go into hyperconverged. We made some assumptions that we would just stick with VMware. It was prior to the Broadcom acquisition. We weren't 100% happy with VMware, heavy costs, not great support, not a great partner at the time. So we wanted to be very open-minded. Did the typical RFP, led to POCs. We ultimately got to a place where it was boiled down to between VxRails and Nutanix, and actually two flavors of Nutanix, Nutanix with AHV or Nutanix with ESXi. I pulled my team together after the bake-off and we unanimously said, "Let's go all in. Let's go AHV." It's very exciting. Nutanix seems to want to partner with Dartmouth. And from there, it was off and we moved towards migration.

Alison Kosik

>> That's great. And... Go ahead.

John Furrier

>> Go ahead.

Alison Kosik

>> No, go ahead.

John Furrier

>> I'm going to ask about this because this has been a big discussion. A lot of people think, "Well, it's not worth it." They're indifferent because they go, "Is the juice worth the squeeze?"

Ty Peavey

>> That's right.

John Furrier

>> Is it going to be too expensive?

Ty Peavey

>> Right.

John Furrier

>> What worked for you? What was the key? And now, compared to now with the AI tooling, take us through what worked and why, and then what's it like today?

Ty Peavey

>> Well, first of all, let's just recognize hyperconverged as one of the core components of Nutanix. We were able to get rid of three-tier hardware in our data center, greatly reducing the amount of technology we needed to support. No longer were there large sand cabinets with lots of fiber channel configurations. We really went down to very comfortable 2U Intel servers that we've been supporting for years. So we simplified our data center in that fact. The other thing was that we were really concerned about what the staff translation into a new hypervisor would be. How much would we have to retool our group? How much training would be involved? And quite honestly, we over-worried about it. We found once we were in, our staff was very comfortable with taking VM snapshots and cloning machines and building machines. We felt very at home very quickly. The fear was really all in our heads.

John Furrier

>> Talk about the operational efficiency piece, because one case study that you have highlights a post I made two days ago, a research note around how we're in a great re-platforming stage. You did it out of necessity, and necessity is the mother of all inventions. You reduce your footprint, check. What else efficiencies did you see? Engineer percentages? How many engineers? Did that maximize the 10X engineer?

Ty Peavey

>> We wanted the infrastructure team, which is a relatively small team. We're 11 engineers. I really strongly discourage having specialized job descriptions. I wanted to move away from storage admins, Windows admins, Linux admins. Those all felt like silos to me. I felt like those are all four competencies that really define what an infrastructure engineer. And when I simplified the data center, or we simplified the data center, we were really able to move towards more of a group that was very comfortable working in all layers of the infrastructure platforms.

John Furrier

>> And you're happy with the number of VMs, the number of containers you'd got producing?

Ty Peavey

>> We're around a thousand VMs split between Windows and Linux. Very automated in our management of that. We have about 600 containers running in two clusters and NKP. Yeah, I would say we're doing fantastic. We're firing all cylinders right now.

John Furrier

>> Deepak, I want to get you in here because we're dominating with Dartmouth because he's sharing a lot of good data. You're the CTO on the cloud-native piece at Nutanix.

Deepak Goel

>> That's right.

John Furrier

>> Cloud native has been a gift to what we're seeing on the AI-native side. He's talking about some of the basics, foundational things around VMs and containers, obviously his platform. That's footprint. Now, the cloud-native piece is kicking in. The AI action with agents sits on top of essentially hybrid cloud. You've got containers are going to play a much more, almost a renaissance of containers. Again, what's your view on this as cloud native intersects the AI-native world?

Deepak Goel

>> Yeah. Actually, like LLM found their home in GPUs, agentic workloads or AI workloads found their home in containers and Kubernetes. And the reason for that is it's a very natural fit. For the AI workloads, the properties that they look for is readily available in Kubernetes. In fact, Kubernetes was popularizing all these architecture constructs that were present in containers that was so natural fit for agentic AI. For example, let's say these agents needs to be lightweight. They need to be scaled up based on demand, and scaled down, which is what Kubernetes can do very efficiently with some of the Horizontal Pod Autoscaler and Vertical Pod Autoscaler construct. Agent needs security. We all talk about how these agents are referencing sensitive data and how they're operating on that. Kubernetes comes with baked-in security. It has ingress controllers. It has AI gateway. So there are tons and tons of properties, which was just, if you match the two natives, as you said, cloud native and AI native, it feels like we're talking of one and the same thing.

John Furrier

>> And it's coming together because the Linux foundation with CNCF really spent decade-plus grinding out Kubernetes. That's my word. They would probably be offended by that word, but they did the hard work.

Deepak Goel

>> Right.

John Furrier

>> And that's key.

Deepak Goel

>> Correct.

John Furrier

>> Now, it's one.

Deepak Goel

>> Yes.

John Furrier

>> The role of virtualization was pre-Kubernetes.

Deepak Goel

>> Yes.

John Furrier

>> Kubernetes is like, I call it the TCP/IP of cloud natives. It causes a lot of great things to happen around it. So what is the role of VMs? Virtualization used to be the thing-

Deepak Goel

>> Correct....

John Furrier

>> for everything.

Deepak Goel

>> Correct.

John Furrier

>> Now, it's having nice, I won't say niche, but purpose-built role, isolation, security. So the containers and the VMs together with Kubernetes, how would you describe that to someone in a C-suite without saying, "Hey, we have the API gateways." What would you say to the CXO and the CFO and the chief human resource officer? "I'm giving out tokens to everybody and I got Kubernetes and containers and VMs." How would you translate that?

Deepak Goel

>> Yeah, so I'll give you a story, and which will span out. Any new technology that comes in, it is adopted by a rush, by a, "Oh, let's adopt it as quickly as possible." But then standardization and consistency follows, and when things get stabilized. So right now what's happening in the industry is the hardware is churning faster than software. NVIDIA releasing their new GPU performance every six months. And so in order to get to those hardware, the fastest ways to run containers on bare metal and to utilize the underlining hardware properties through containers directly. But virtualization will have its role to play, which is when they come in, they help in isolation, they help in security, they help in efficiency because hardware management is not the piece of everybody's cake. It's a very hard thing to do, which is what virtualization helps the IT administrator to do. Upgrades, installations, lifecycle management of these systems is what virtualization is always support-

John Furrier

>> So, effective operations.

Deepak Goel

>> Yes.

John Furrier

>> More money making.

Deepak Goel

>> Yes. Also, there is one thing, is how to use these expensive GPU unit efficiently. And virtualization has this nice scheduling property that understands the underlining GPU constructs interconnect on the socket and the chip set, which container doesn't understand. So all these...

John Furrier

>> By the way, Jensen Huang, when he presented his Pareto graph at all three generations, the Vera Rubin being the latest, he showed basically tiered services. He basically said, "Don't use Vera Rubin to ask what's the weather like outside." Hello, that's high, expensive gear.

Deepak Goel

>> Correct.

John Furrier

>> You know what I'm talking about.

Deepak Goel

>> I do.

John Furrier

>> It's like, so software, scheduling-

Deepak Goel

>> Correct.

John Furrier

>> Sounds like an operating system to me.

Deepak Goel

>> Yes. So the best combination is, and it'll be bound to happen, it's like containers utilizing the properties of virtualized layer, which will virtualize the infrastructure, virtualize the scheduling aspect, virtualize the isolations, and then the security-

John Furrier

>> It's a nice compliment....

Deepak Goel

>> and then container.

John Furrier

>> It's a nice comment. And the AI stack's emerging. Every enterprise might have a different AI stack. We'll see. Ty, I want to come back to you because, and I asked the question about the C-suite because you're progressive, you guys are early adopters. You're not afraid-

Ty Peavey

>> We were....

John Furrier

>> to take risks.

Ty Peavey

>> No.

John Furrier

>> Mist. All the examples that you guys have, the migration, you're first on a lot of things. At some point, do people go, "Hey, you running a science fair over there? What's going on with your department?" When do you decide it's worked or not working? When do you say, "You know what? We took a chance."?

Ty Peavey

>> Yeah. It's not all fun and games sooner or later. Every institution has to figure it out for itself, but at Dartmouth, at the end of the day, it's about teaching and learning, the students. That's what we're there for, is making sure that the experience for the students and for the faculty teaching them are stable, are supportive of their mission. And so we have to be particularly cautious around things like admission decision days and class selection days. And so we heavily factor those factors into our decision-making around change control and what we're willing to do.

John Furrier

>> I think this is a great opportunity to zoom out. As you look at going forward in your mission, teaching, learning and testing and teaching. You got faculty too, you got to serve them.

Ty Peavey

>> Absolutely.

John Furrier

>> You have a student body. It's like you're a sports team. You got to run the team, fan experience, and the players.

Ty Peavey

>> Yeah, absolutely.

John Furrier

>> You got a mission, then you got the enterprises. They're waiting for agentic because they don't want to take on a burden, a lot of risk. AI stacks aren't developed, but they have all that data to unlock.

Deepak Goel

>> Right.

John Furrier

>> Hybrid cloud's the answer. What do you guys see the next steps here? What are the key dots to connect here in this world today?

Deepak Goel

>> You want to go first?

Ty Peavey

>> No, you go ahead.

Deepak Goel

>> Okay. So it's platform consolidation. So you earlier raised a problem to our current organization, is we have containers, we have agents, we have AI workloads, we have virtualization, we have security, we have storage. How do anybody make sense out of it? And then we have on-prem, hybrid-cloud, edge deployments. I think what will ease this complexity is consolidating and unifying this on a platform. And that's what we are working towards here in Nutanix, is we are providing a platform that can ease the operational burden for the organization as they are embarking towards these newer workloads, which are, frankly, stretching the very fabric of infrastructure in all direction. So how can you consistently deploy on a VM and containers? How can you consistently have security policies? How can you consistently have, as you're scaling your infrastructure, how's your operational burden is not scaling twice of that? How do you bring down the cost? There is an expectation that with adopting modernization, there is an ROI that we will return. But if it is not done right with taking a platform approach, you would have this straw of infrastructure all over the place.

John Furrier

>> There's an old expression. Don't play chess unless you know what checkmate looks like. ROI is elusive. You don't know what you're measuring.

Deepak Goel

>> Yes.

John Furrier

>> What are you actually doing?

Deepak Goel

>> Right.

John Furrier

>> What's your thoughts on the dots to connect?

Ty Peavey

>> Yeah, so as a consumer, as a user of AI, I think there are three real questions. One is, how do we leverage it securely? How do we have the proper agreements in place to protect the data? How do we make sure that AI is affecting change in our infrastructure securely? Two is funding. It's very expensive right now, and getting more expensive. So where do we find those funding sources? And three is, how do you keep up with the functionality? It changes so, so quickly, so fast. How do you stay on top of what is available? I think people consume it faster than we understand how it works.

Deepak Goel

>> And as we saw this morning, a simple application consumes some 11 million tokens, and token has become a new oil.

Ty Peavey

>> Yeah, exactly.

Deepak Goel

>> How do-

John Furrier

>> People are going to have tokens in their compensation plan.

Deepak Goel

>> Correct.

John Furrier

>> It's like the engineers. "Here's your token budget." I know some sales guys were telling me from the old cell phone days, if your cell phone bill wasn't $1,000, you weren't calling customers. In a way, if you burn your-

Deepak Goel

>> Exactly.

John Furrier

>> How come you're not spending your tokens? Well, I'm writing code manually. What are you doing?

Deepak Goel

>> Exactly.

John Furrier

>> Jensen's argument is if you're going to pay the big salaries, here's your budget for tokens. Go be productive. Paint a Picasso. Build something.

Deepak Goel

>> Correct. Next big challenge is how to bring down this token cost, how do not only the number of tokens to be less, but each token cost has to come down.

John Furrier

>> Like the migration story, consolidate, get a nice, solid, stable footprint, baseline it, foundationalize it, build on top.

Ty Peavey

>> Create an abstraction layer, remove the complexity, simplify it, bring functionality to the IT teams and the users.

John Furrier

>> And make good bets.

Deepak Goel

>> Yeah.

Ty Peavey

>> And make good bets.

John Furrier

>> And then depending on what your risk profile is, the bigger risk you take, the better the reward, how you manage that.

Ty Peavey

>> And recognize the failures and learn from them, and be willing to accept that there will be failures on the way, and use them to make better decisions. It's okay.

John Furrier

>> All right.

Alison Kosik

>> All right, Ty, Deepak, thanks so much. Great conversation. Enjoyed your insights.

Ty Peavey

>> Thank you.

Deepak Goel

>> Thank you for having us.

Ty Peavey

>> Yeah, thank you very much.

Alison Kosik

>> Thanks. More to come. You're watching theCUBE, the leader in live technology coverage. We'll be right back.