Kubernetes - A Unified Platform for VM and Container Workloads

The case for disruption: challenging the status quo

Most enterprise organizations are carrying the weight of a bloated virtual infrastructure estate. We observe virtualization platforms on one side and containers platforms on the other, with teams and governance models that rarely overlap between each other. This separation carries a real cost to organizations big and small. Companies operating siloed virtual machine and container environments may tend spend significantly more time and resources on operational overhead, face longer incident resolution times and struggle to enforce consistent security posture across their estate.

One platform to rule them all: Kubernetes as the universal workload layer

Kubernetes projects like KubeVirt, coupled with enterprise ready platforms such as Red Hat OpenShift Virtualization have made it technically viable to run virtual machines natively within Kubernetes, not as an afterthought, but as a first-class workload type. Organizations can now manage VMs and containers through a single application programming interface (API), a single scheduling layer and a single operational model.

WWT sees this pattern gaining rapid momentum among enterprises undertaking virtual infrastructure modernization, and it represents a clear path to a unified, future-ready architecture.

The journey: from fragmentation to a single operating model

Achieving this vision requires a deliberate, phased approach anchored in Infrastructure as Code. IaC not a best practice but as the foundational methodology that makes a unified Kubernetes platform operationally sustainable at scale. The journey typically unfolds in three stages:

First, assessing the existing workload estate and identifying candidates for VM-based Kubernetes orchestration. Second, establishing a hardened, GitOps driven Kubernetes foundation integrated with existing identity, security and monitoring systems. Third, migrating workloads incrementally while advancing containerization where appropriate.

WWT's Advanced Technology Center plays a critical role, giving clients a risk-free environment to validate this architecture before committing to production. The journey is not trivial, but it is now achievable, and the destination is well worth the effort.

Infrastructure as code as the operational backbone

Infrastructure as Code is not merely a tool in this journey — it is the philosophy that makes the unified platform model work at scale. When every component of the stack, from cluster configuration to VM definitions to network policy, is expressed as version-controlled code, teams gain a single source of truth. Drift is eliminated. Environments become reproducible. Onboarding a new workload becomes a pull request, not a ticket.

GitOps workflows bring software development discipline to infrastructure operations: every change is traceable, every rollback is clean and platform teams gain the velocity to move fast without breaking things. 

Efficiency gains: what consolidation delivers

The efficiencies unlocked by a unified Kubernetes platform are measurable. At the most immediate level, organizations eliminate the cost of parallel infrastructure stacks, reducing licensing overhead, consolidating tooling spend and shrinking the operational surface area. The deeper gains are organizational: when VMs and containers share a common platform, the artificial boundaries between infrastructure and application teams dissolve.

Solving the networking challenge: Kubernetes on Cilium and the future of virtual network infrastructure

One of the most persistent objections to consolidating VMs onto Kubernetes is advanced networking and security policies enforced throughout large VM estates. Traditional virtualization environments are built around sophisticated constructs in micro-segmentation, distributed firewalling, network isolation, stateful load balancing and deep traffic observability. These are elements that enterprises have spent years engineering and hardening. Replicating that capability in a Kubernetes-native model has historically been the gap that stopped migration efforts in their tracks.

Isovalent and Cilium can now close that gap. Built on Extended Berkley Packet Filter (eBPF), a Linux kernel technology that enables programmable, high-performance networking without the overhead of traditional agents or proxies. Cilium is trusted by the both the largest hyperscalers and enterprises alike. Isovalent's enterprise offering, Isovalent Networking for Virtualization, extends these capabilities directly to VM workloads running inside Kubernetes.

The practical implications are significant. Isovalent Networking for Virtualization enables micro-segmentation applying identity-aware network policies consistently across both VMs and containers. Rather than relying on IP addresses, Cilium uses identities for policy enforcement and a new construct type called Private Network for enhanced VM support, enabling a true Zero Trust networking model at the application instead of infrastructure layer. Think of it as a virtual private cloud (VPC) for Kubernetes.

On observability, Isovalent's Hubble Timescape component provides rich, real-time visibility into network flows across both workload types without requiring sidecar proxies or instrumentation changes. For enterprises leaving behind traditional enterprise virtualization, this is the architectural answer: How does one preserve enterprise-grade networking controls while moving to a unified Kubernetes platform.

The moment to act

Organizations that move now can build a durable multi-purpose infrastructure advantage. Those that wait risk accumulating technical, financial and organizational debt that compounds every dey. Kubernetes as a unified workload platform is not the future. It is now the present, and the window to lead rather than follow is now open.