Tech Blog by vCluster

Why AI workloads, GPU infrastructure, and platform scale are forcing Kubernetes teams to rethink multi-tenancy

AI workloads and GPU infrastructure are exposing the limits of traditional Kubernetes multi-tenancy. In 2026, platform teams need a new layer.

A 5-stage maturity model for evolving your GPU infrastructure from basic clusters to production-ready AI factories with secure multi-tenancy

Most enterprises start with basic GPU clusters and hit scaling walls fast. This guide walks through the 5-stage journey from manual provisioning to a fully automated AI factory, covering multi-tenancy, cost optimization, and how to achieve cloud-like operations on-prem without vendor lock-in.

Lessons in Intentional Tenancy and Security at Scale from a Cloud Platform Director

If a workload needs isolation, give it its own cluster. It sounds safe, but at scale, this logic breaks down. Learn why consistency, not separation, is the real security challenge in modern Kubernetes environments.

Streamline MLOps and Multi-tenancy by Running Isolated GPU Workloads with Virtual Clusters

Scale your AI infrastructure without the overhead. In this hands-on tutorial, we demonstrate how to use vCluster technology to virtualize GPU resources, ensuring secure multi-tenancy and efficient resource sharing for production-grade MLOps.

Why MPS and MIG fall short—and how virtual clusters deliver isolation without hardware lock-in

GPUs are expensive, but most organizations only achieve 30-50% utilization. The problem? GPUs weren't designed for sharing. Software solutions like MPS lack isolation. Hardware solutions like MIG lock you into specific vendors. vCluster takes a different approach—solving GPU multitenancy at the Kubernetes orchestration layer.

Introducing the first vCluster Ambassadors shaping the future of Kubernetes multi-tenancy and platform engineering

Meet the first vCluster Ambassadors - community leaders and practitioners advancing Kubernetes multi-tenancy, platform engineering, and real-world developer platforms.

Explore time-slicing, MIG, and custom workarounds for sharing GPUs across Kubernetes workloads—plus their trade-offs for isolation and management.

GPUs are expensive and rarely used to full capacity. Learn the standard methods for sharing GPUs in Kubernetes—time-slicing and MIG—along with DIY alternatives, and discover why combining these techniques with vCluster delivers the isolation multi-tenant AI/ML workloads demand.

How to design, build, and operate a cost-effective private cloud infrastructure for enterprise AI at scale

Public clouds are convenient for AI experimentation, but production workloads often hit walls. For enterprises running continuous training and inference, a private cloud can deliver better ROI, data sovereignty, and performance. This comprehensive guide walks through architecting a private cloud for AI workloads from the ground up.

How to deploy Karpenter on Google Cloud Platform (and why you might want vCluster Auto Nodes instead)

Karpenter has become the gold standard for Kubernetes autoscaling on AWS, but what about GCP? This guide shows you how to set it up and introduces a better way.

How to safely share expensive GPU infrastructure across teams without sacrificing performance or security

GPUs don't support native sharing between isolated processes. Learn four approaches for running multitenant GPU workloads at scale without performance hits.

What the AI Infrastructure 2025 Survey Reveals, And How Platform Teams Can Respond

The latest AI Infrastructure 2025 survey shows that most organizations are struggling not due to GPU scarcity, but because of poor GPU utilization caused by limited multi-tenancy capabilities. Learn how virtual clusters and virtual nodes help platform teams solve high costs, sharing issues, and low operational maturity in Kubernetes environments.

Announcing the Infrastructure Tenancy Platform for NVIDIA DGX—plus what we learned from 100+ conversations at KubeCon about GPU efficiency, isolation, and the future of AI on Kubernetes.

KubeCon Atlanta 2025 was packed with energy, launches, and conversations that shaped the future of AI infrastructure. At Booth #421, we officially launched the Infrastructure Tenancy Platform for NVIDIA DGX—a Kubernetes-native platform designed to maximize GPU efficiency across private AI supercomputers, hyperscalers, and neoclouds. Here's what happened, what we announced, and why it matters for teams scaling AI workloads.

New platform combines virtual clusters, dynamic GPU autoscaling, and hybrid networking to maximize NVIDIA infrastructure efficiency—from DGX systems to bare metal and cloud.

vCluster Labs announced its Infrastructure Tenancy Platform for AI at KubeCon North America 2025, delivering a Kubernetes-native foundation for running AI workloads on NVIDIA GPU infrastructure. The platform introduces vCluster Private Nodes, Auto Nodes with Karpenter-based autoscaling, vCluster VPN for hybrid networking, and direct integrations with NVIDIA Base Command Manager, KubeVirt, and Netris—helping organizations maximize GPU utilization while maintaining full workload isolation and security across cloud, on-prem, and bare metal environments.

Why virtual clusters aren’t “fake” Kubernetes clusters — they’re the foundation of real multi-tenant platforms built for cloud, on-prem, and bare metal.

Many engineers think “virtual” means “not real.” In Kubernetes, the opposite is true. Virtual clusters run full control planes inside namespaces, giving teams real isolation, faster CI/CD, and multi-tenant efficiency across cloud and on-prem infrastructure with vCluster from Loft Labs.

A practical guide to implementing cloud bursting using vCluster VPN, Private Nodes, and Auto Nodes for secure, elastic, multi-cloud scalability.

Cloud bursting lets you expand compute capacity on demand without overprovisioning or re-architecting your systems. In this guide, we break down how vCluster VPN connects Private and Auto Nodes securely across environments—so you can scale beyond limits while keeping costs and complexity in check.

vCluster Labs and Netris team up to bring cloud-grade Kubernetes automation and network-level multi-tenancy to AI factories and GPU-powered infrastructure.

vCluster Labs has partnered with Netris to revolutionize how AI operators run Kubernetes on GPU infrastructure. By combining vCluster’s Kubernetes-level isolation with Netris’s network automation, the integration delivers a full-stack multi-tenancy solution, simplifying GPU cloud operations, maximizing utilization, and enabling cloud-grade performance anywhere AI runs.

How vCluster’s Private Nodes, Auto Nodes, and Standalone releases redefine multi-tenancy for modern Kubernetes platforms.

From hardware-isolated clusters to dynamic autoscaling and fully standalone control planes, vCluster’s latest launch series completes the future of Kubernetes multi-tenancy. Discover how Private Nodes, Auto Nodes, and Standalone unlock new levels of performance, security, and flexibility for platform teams worldwide.

Bootstrapping Kubernetes from scratch, no host cluster, no external dependencies.

Kubernetes multi-tenancy just got simpler. With vCluster Standalone, you can bootstrap a full Kubernetes control plane directly on bare metal or VMs, no host cluster required. This walkthrough shows how to install, join worker nodes, and run virtual clusters on a single lightweight foundation, reducing vendor dependencies and setup complexity for platform and infrastructure teams.

How vCluster and NVIDIA’s KAI Scheduler reshape GPU workload management in Kubernetes - enabling isolation, safety, and maximum utilization.

GPU workloads have become the backbone of modern AI infrastructure, but managing and upgrading GPU schedulers in Kubernetes remains risky and complex. This post explores how vCluster and NVIDIA’s KAI Scheduler together enable fractional GPU allocation, isolated scheduler testing, and multi-team autonomy, helping organizations innovate faster while keeping production safe.

Eliminating the “Cluster 1 problem” with vCluster Standalone v0.29 – the unified foundation for Kubernetes multi-tenancy on bare metal, VMs, and cloud.

vCluster Standalone changes the Kubernetes tenancy spectrum by removing the need for external host clusters. With direct bare metal and VM bootstrapping, teams gain full control, stronger isolation, and vendor-supported simplicity. Explore how vCluster Standalone (v0.29) solves the “Cluster 1 problem” while supporting Shared, Private, and Auto Nodes for any workload.

Auto Nodes extend Private Nodes with provider-agnostic, automated node provisioning and scaling across clouds, on-prem, and bare metal.

Kubernetes makes pods elastic, but node scaling often breaks outside managed clouds. With vCluster Platform 4.4 + v0.28, Auto Nodes fix that gap, combining isolation, elasticity, and portability. Learn how Auto Nodes extend Private Nodes with automated provisioning and dynamic scaling across any environment.

Dynamic, isolated, and cloud-agnostic autoscaling for every virtual cluster.

vCluster Auto Nodes brings dynamic, Karpenter-powered autoscaling to any environment, public cloud, private cloud, or bare metal. Combined with Private Nodes, it delivers true isolation and elasticity for Kubernetes, letting every virtual cluster scale independently without cloud-specific limits.

Kubernetes pods scale elastically, but node scaling often stops at the provider boundary. Auto Nodes extend Private Nodes to bring elasticity and portability to isolated clusters across clouds, private datacenters, and bare metal.

Pods autoscale in Kubernetes, but nodes don’t. Outside managed services, teams fall back on brittle scripts or costly overprovisioning. With vCluster Platform 4.4 + vCluster v0.28, Auto Nodes close the gap, bringing automated provisioning and elastic scaling to isolated clusters across clouds, private datacenters, and bare metal.

Kubernetes has pods and deployments covered, but when it comes to nodes, scaling breaks down across clouds, providers, and private infrastructure. Auto Nodes change that.

Kubernetes makes workloads elastic until you hit the node layer. Managed services offer partial fixes, but hybrid and isolated environments still face scaling gaps and wasted resources. vCluster Auto Nodes close this gap by combining isolation, just-in-time elasticity, and environment-agnostic portability.

A technical walkthrough of Private Nodes in vCluster v0.27 and how they enable true single-tenant Kubernetes clusters.

Private Nodes in vCluster v0.27 take Kubernetes multi-tenancy to the next level by enabling fully isolated, dedicated clusters. In this deep dive, we walk through setup, benefits, and gotchas, from creating a vCluster with Private Nodes to joining worker nodes and deploying workloads. If you need stronger isolation, simpler lifecycle management, or enterprise-grade security, this guide covers how Private Nodes transform vCluster into a powerful single-tenant option without losing the flexibility of virtual clusters.

A new name, the same mission, building the best Kubernetes tenancy tools for teams everywhere.

Loft Labs is now vCluster Labs, a name that reflects our focus on building the best Kubernetes multi-tenancy and infrastructure engineering tools. The same team, projects, and mission remain, but with a clearer brand aligned to our product, vCluster.

Dedicated, tenant‑owned nodes with a managed control plane, full isolation without running separate clusters.

Private Nodes complete vCluster’s tenancy spectrum: tenants connect their own nodes to a centrally managed control plane for full isolation, custom runtimes (CRI/CNI/CSI), and consistent performance, ideal for AI/ML, HPC, and regulated environments. Learn how it works and what’s next with Auto Nodes.

Rethinking Kubernetes scale: avoid the risks of etcd sharding with virtual clusters built for performance, stability, and multi-tenant environments.

Is your Kubernetes cluster slowing down under load? etcd doesn’t scale well with multi-tenancy or 30k+ objects. This blog shows how virtual clusters offer an easier, safer way to isolate tenants and scale your control plane, no sharding required.

Why covering the full Kubernetes tenancy spectrum is critical, and how Private Nodes bring stronger isolation to vCluster

In this blog, we explore why covering the full Kubernetes tenancy spectrum is essential, and how vCluster’s upcoming Private Nodes feature introduces stronger isolation for teams running production, regulated, or multi-tenant environments without giving up Kubernetes-native workflows.

Why sharding etcd doesn’t scale, and how virtual clusters eliminate control plane bottlenecks in large Kubernetes environments.

OpenAI’s outage revealed what happens when etcd breaks at scale. This post explains why sharding isn’t enough, and how vCluster offloads API load with virtual control planes. Benchmark included.

How vCluster Balances Kubernetes Cost Reduction With Real-World Performance

Can you really save millions on Kubernetes infrastructure without compromising performance? Yes, with vCluster. In this blog, we break down how virtual clusters reduce control plane overhead, unlock higher node utilization, and simplify multi-tenancy, all while maintaining lightning-fast performance.

A curated list of impactful, technical, and thought-provoking sessions to catch at KubeCon + CloudNativeCon India 2025 in Hyderabad.

KubeCon + CloudNativeCon India 2025 is back in Hyderabad on August 6–7! With so many exciting sessions, it can be hard to choose. Here are 5 standout talks you shouldn't miss, from real-world Kubernetes meltdowns to scaling GitOps at Expedia, and even why Kubernetes is moving to NFTables.

Unlocking Secure and Scalable Multi-Tenancy in Kubernetes with Virtual Clusters

Running multiple tenants on a single Kubernetes cluster can be complex and risky. In this post, Liquid Reply explores how vCluster offers a secure and cost-efficient solution by isolating workloads through lightweight virtual clusters.

Container breakouts on GPU nodes are real, and just three lines of code can be enough. Discover how vNode neutralizes vulnerabilities like NVIDIAScape without relying on VMs.

NVIDIAScape (CVE-2025-23266) is a critical GPU-related vulnerability that allows attackers to break out of containers and gain root access. While some respond by layering in virtual machines, this blog walks through a better approach, how vNode uses container-native sandboxing to neutralize such attacks at the kernel level without sacrificing performance. Includes a step-by-step replication of the exploit, and a demo of how vNode prevents it.

How shared clusters fall short, and why virtual clusters are the future of controller development.

Shared clusters are cost-effective, but when it comes to building and testing Kubernetes controllers, they create bottlenecks, from CRD conflicts to governance issues. This blog breaks down the trade-offs between shared, local, and dedicated clusters and introduces virtual clusters as the scalable solution for platform teams.

How Kubernetes can make GPU usage more efficient for AI/ML teams through MPS, MIG, and smart scheduling.

As AI and ML workloads scale rapidly, GPUs have become essential, and expensive resources. But most teams underutilize them. This blog dives into how GPU sharing in Kubernetes can help platform teams increase efficiency, cut costs, and better support AI infrastructure.

How virtual clusters and smarter tenancy models can reduce carbon impact while scaling AI workloads.

AI’s rapid growth is fueling a silent climate problem: idle infrastructure. This blog explores why multi-tenancy is key to scaling AI sustainably and how vCluster helps teams reduce waste while moving faster.

Keep your CI pipelines clean and efficient by automating Kubernetes resource cleanup with vCluster and Loft.

Leftover Kubernetes resources from CI jobs can drive up cloud costs and clutter your clusters. This guide shows how to automate cleanup tasks using vCluster, helping you maintain cleaner, faster CI/CD pipelines.

Keep your CI pipelines clean and efficient by automating Kubernetes resource cleanup with vCluster and Loft.

Leftover Kubernetes resources from CI jobs can drive up cloud costs and clutter your clusters. This guide shows how to automate cleanup tasks using vCluster, helping you maintain cleaner, faster CI/CD pipelines.

Why Namespace Isolation Falls Short for GPU Workloads, and How Multi-Tenancy with vCluster Solves It

Managing AI workloads on bare metal Kubernetes with GPUs presents unique challenges, from weak namespace isolation to underutilized resources and operational overhead. This blog explores the pitfalls of namespace-based multi-tenancy, why running a separate cluster per team is expensive, and how vCluster enables secure, efficient, and autonomous GPU sharing for AI teams.

Step-by-step guide to setting up a GPU-enabled Kubernetes cluster on GKE for scalable AI and ML workloads.

Running AI or ML workloads on Kubernetes? This tutorial walks you through setting up a GPU enabled GKE cluster, from configuring GPU quotas and node pools to testing workloads and optimizing for multi-team GPU usage with vCluster.

Cut Kubernetes costs by up to 91% using spot instances and vCluster, without compromising workload stability.

Spot instances offer massive savings but come with unpredictability. In this step-by-step guide, learn how to combine them with vCluster to build resilient, cost-effective Kubernetes environments for CI/CD, AI/ML, and more.

Compare vCluster and HyperShift to understand which solution delivers better multi-tenancy, efficiency, and compatibility in Kubernetes environments.

As platform engineering teams scale Kubernetes adoption, choosing the right multi-tenancy model becomes crucial. This article compares Red Hat's HyperShift and Loft Labs’ vCluster, highlighting the trade-offs in resource usage, ecosystem compatibility, portability, and vendor lock-in. Learn why vCluster’s lightweight, upstream-native approach is often better suited for modern internal platforms.

Enable multi-tenant Kubernetes service mesh with zero sidecars and seamless Istio integration using Ambient Mode and vCluster.

The v0.25 release of vCluster brings native support for Istio’s Ambient Mesh, enabling shared service mesh capabilities across multiple virtual clusters without sidecars. This update dramatically reduces resource overhead, simplifies operations, and boosts scalability in multi-tenant Kubernetes environments.

Octarine Unleashed: How Kubernetes 1.33 Changes Everything for Devs & Platform Teams

Kubernetes 1.33, codenamed "Octarine: The Color of Magic", lands with 64 feature updates. This blog dives into the most impactful ones and shows how to try them immediately with vCluster. From in-place pod resizing to ClusterTrustBundle, here’s what’s new.

From mainframes to Kubernetes: Why it might be time to let go of virtual machines.

Kubernetes changed how we build and deploy software, but are you still clinging to virtual machines out of habit? In this post, Scott McAllister walks through the evolution of enterprise computing, from mainframes to microservices, to help you rethink your current infrastructure choices. Is it time to go bare metal?

Virtual clusters with container native isolation on bare-metal

Traditional multi-tenancy in Kubernetes often leads to complex RBAC, noisy neighbors, and security headaches. This post breaks down why platform teams struggle with namespace-based isolation, and how virtual clusters offer a better path forward.

Why Moving from VMware to Kubernetes-native Infrastructure is Critical for Modern Enterprises

Discover why enterprises in 2025 are shifting from traditional VMware based virtual machines to modern, Kubernetes-native architectures. Learn how adopting Kubernetes closer to bare metal simplifies infrastructure, reduces costs, and enhances scalability and efficiency.

Well, well, well, KubeCon + CloudNativeCon EU 2025 has come to an end, and we made so many friends and memories! Let's review some of the exciting things our team did at KubeCon: Platinum Sponsors at Platform Engineering Day: The buzz around platform engineering and internal deve...

There's something new on the Ranch

What about Rancher? Does vCluster work on Rancher? How do we manage virtual clusters on Rancher? Hey, what about Rancher? These are just a few of the questions we have heard over the last couple of years at KubeCon, on the interwebs, and everywhere in between. The answer was alwa...

When we first launched vCluster in 2021, our mission was clear: make Kubernetes multi-tenancy easier, safer, and more cost-efficient. Since then, we've helped organizations around the globe manage Kubernetes with greater flexibility and security. But as Kubernetes usage expanded,...

KubeCon is around the corner, and we at LoftLabs are excited to be back for the 2025 Europe Edition! The projected number of attendees speaks to the scale of the cloud-native landscape, and we’re so excited to mark our presence with talks, activities, and being a proud sponsor. V...

Back it up and put it to sleep

I’m excited to present the updates coming with vCluster v0.24. In this post, we will cover Snapshots and Sleep Mode Improvements. Along with this post, we have a couple of videos that will demo both features. Snapshots Let’s start by talking about Snapshots. While you were always...

The ideal size of your Kubernetes clusters is a day 0 question and demands a definite answer. You find one giant cluster on one end of the spectrum and many small-sized ones on the other, with every combination in between. This decision will impact your organization for years to ...

Multi-tenancy in Kubernetes has been an ongoing challenge for organizations looking to optimize their cloud-native infrastructure. Over the years, the approach to multi-tenancy has evolved, from simple namespace isolation to virtual clusters and, more recently, full-fledged inter...

Kubernetes was designed for efficient resource sharing, but complexity escalates when multiple teams or users step into the same cluster. As organizations scale, the challenge isn't just about running workloads; it's about how those workloads coexist securely and efficiently.

Like a couple of innovative technologies, different people have different viewpoints on where WebAssembly fits the technology landscape. > WebAssembly (also called Wasm) is certainly the subject of much hype right now. But what is it? Is it the JavaScript Killer? Is it a new prog...

In a fast-paced development environment, having an isolated and ephemeral environment to test changes for every pull request (PR) is a game-changer. In this blog, I’ll walk you through setting up ephemeral PR environments using vCluster, enabling seamless testing of your applicat...

Multi-tenancy in Kubernetes is not a new cooncept; it simply refers to creating isolated spaces for different users, teams, or projects. Many organizations begin by using namespaces to isolate workloads, teams, or projects. However, they soon encounter limitations, such as challe...

vCluster for isolation and costs control

This post is the third and final in my series about running tests on Kubernetes for each pull request. In the first post, I described the app and how to test locally using Testcontainers and in a GitHub workflow. The second post focused on setting up the target environment and ru...