娇色导航

Keith Shaw: Hi everybody, welcome to DEMO, the show where companies come in and show us their latest products and platforms. Today, I'm joined by Kurt Madel, a Senior Solutions Architect at Loft Labs. Welcome to the show, Kurt. Kurt: Thanks for having me.

Keith: All right, so tell us a little bit about Loft Labs and what you're going to be showing us today. Kurt: Sure!

Loft Labs is a Series A startup dedicated to simplifying Kubernetes. We believe the next big leap in technology will come from virtualizing Kubernetes and making it the standard for building and operating cloud-native platforms and applications.

Keith: And I see “vCluster” on your shirt—that’s the name of the product, right?

Kurt: Yes, vCluster is our flagship product, and it's what I'll be demoing today. We believe virtualizing the cloud-native stack will have as big an impact on the cloud era as virtualizing physical servers did over 30 years ago. Keith: Wow, okay.

So who is this designed for within a company? Is it something a 娇色导航would be looking at, or is it really for developers working directly with Kubernetes and clusters?

Kurt: It’s really for both. But our main target—especially from a sales perspective—is platform engineering teams. These are the folks building Kubernetes-based solutions for developers, supporting multi-tenancy, and so on.

Keith: All right, take us through the current challenges companies face, and how vCluster helps solve them. Basically—why should anyone be watching this video? Kurt: Of course.

Kubernetes usage is exploding across companies worldwide, and managing it at scale is becoming increasingly expensive. vCluster and the vCluster Platform help platform engineering teams more easily manage Kubernetes while also reducing costs. As Kubernetes adoption grows, vCluster lets teams create isolated, multi-tenant clusters within a host cluster.

Features like sleep mode and the ability to share the host cluster’s platform stack help reduce overall costs.

Keith: Does this problem get worse as companies scale? Is it linear or exponential? When do companies typically notice they have an issue? Kurt: It varies.

Sometimes companies notice early—even with a small Kubernetes deployment—because developers can’t get the isolated environments they need. Sharing clusters means they lack autonomy. vCluster solves this by letting each team or individual have their own dedicated cluster, even in smaller organizations. As Kubernetes usage scales, the cost savings scale exponentially.

Keith: So what do companies do if they don’t have something like vCluster? Are they wasting time with slow deployments or just overspending?

Kurt: A bit of both—great question. Without vCluster, companies either spin up more physical clusters—which are harder and more expensive to manage—or they use Kubernetes namespaces for limited multi-tenancy. Namespaces don’t give developers the freedom or isolation they need to be productive. Keith: Got it.

Let’s jump into the demo and check out some of the cool features. Kurt: Sure.

Today, I’ll walk through vCluster and the vCluster Platform. We'll start with a common use case: giving every GitHub pull request a dedicated Kubernetes cluster. I’ve built a workflow using Argo CD, Crossplane, and vCluster.

When a developer creates a pull request, they just add a label—say, “create PR vCluster”—and that kicks off a workflow that deploys the application into a vCluster. Thanks to sleep mode and auto-delete, that vCluster doesn’t run continuously.

If no one interacts with it, it goes to sleep, spinning down all cost-incurring resources. When needed, a developer clicks a link, interacts with the API server, and the vCluster spins back up—fast, like a container.

Keith: So, just to make sure I understand: without sleep mode, a company would be leaving clusters “on,” like leaving the lights on at home?

Kurt: Exactly—or they wouldn’t provide a dedicated cluster at all. That can lead to bugs surfacing later, costing more money. Spinning up an EKS cluster takes 20–30 minutes, while a vCluster spins up in seconds, so developers aren’t waiting around. Here’s the vCluster for the pull request we just discussed.

It's running a demo app. We even deployed Argo CD and the app itself inside the vCluster. If I show the namespace where this is running, you’ll see pods consuming memory and CPU.

To demonstrate sleep mode, I’ll manually put it to sleep—though it's configured to sleep after two hours of inactivity. Now, the same namespace shows zero pods running. Just as quickly, we could wake it up and all resources come back online.

Now, let’s take a look at what a developer's experience might be. Say the platform team hasn’t set up automation for pull requests, and a dev just needs a one-off cluster. I'll impersonate a user—Anna—using the impersonation feature. The UI updates to reflect her limited permissions.

When she clicks on a vCluster she has access to, like “API Framework,” she can’t access the host cluster. No need to give developers full access—they only see and use what they need.

Within this project, Anna can only create vClusters using templates. That way, she doesn’t need to know the internals—she just picks a template like “Sleepy,” names it, and sets options like the Kubernetes version.

She can even run a different version in the vCluster than the host cluster—great for testing compatibility.

As that vCluster spins up, let’s switch to a VS Code IDE. We’re logged in as Anna, and using the vCluster CLI to create a new vCluster called “Anna’s Dev,” with a specific template and Kubernetes version. The CLI integrates with the platform, showing the projects Anna can access.

She selects the API Framework project, and the vCluster starts creating. Now, back on the platform, we see “Anna’s Dev” and “Sleepy” clusters starting up.

Anna can also use the UI to view pod logs and interact with deployments. Her CLI-created vCluster is now up. Here’s a great use case: deploying something like Crossplane, which can’t run in a namespace.

The host cluster runs Crossplane 1.16, but Anna wants to test 1.18, so she installs it in her vCluster. Templates can also include Helm dashboards for visualizing deployments. Anna deployed Crossplane via Helm, and she can view it in the dashboard.

When she’s done, she deletes the vCluster—or relies on auto-delete to clean up. And now, we see Anna’s vCluster is being deleted.

And “Sleepy” is gone too. Keith: So, everything we just did took 6–7 minutes. How long would that take without vCluster? Kurt: Depends.

On-prem, it can take a month to get a cluster—crazy. In the cloud, it’s faster, but still 20–30 minutes and far more expensive. Now, let me show you the dashboard, especially the cost-savings view. Besides the overview of running vClusters, the cost dashboard highlights savings from features like sleep mode.

Just today, we saved $27 in this small demo. Over the past month, we’ve saved around $2,000.

Keith: All right, you showed a ton of features, but I’m sure there’s more to explore. Where can people go to learn more? Do you offer a free trial? Kurt: Yes!

We offer a 14-day free trial. Just go to vcluster.com, which links to our documentation. If you want to learn about our other products, head to loft.sh.

Keith: Kurt Madel, thanks again for being on the show. Kurt: Thanks for having me. Keith: That’s all the time we have for today’s DEMO. Be sure to like the video, subscribe to the channel, and drop your thoughts in the comments. Join us every week for new episodes.

I’m Keith Shaw—thanks for watching! ?