Kubernetes won the container orchestration argument some time ago. That’s settled. What isn’t settled — and what comes up constantly on migration projects — is what you do when the Kubernetes platform you’re already running is no longer the one you want to be on. Nutanix ran a session on exactly this as part of its Migration in Motion webinar series, and it’s worth a look if container migration is anywhere on your roadmap.
The session
Migration in Motion: How to Move Containers is a 45-minute virtual session delivered by Nutanix professional services experts, subtitled “Practical insights on moving to Kubernetes and migrating Kubernetes across platforms.” The live run took place on Tuesday, 19 May 2026 (10:00 BST | 11:00 CEST | 13:00 GST), presented by David Rios Delgado, Senior Staff Enterprise Architect at Nutanix.
The registration page is still live, and this session is part of Nutanix’s on-demand Migration in Motion collection — so if the date has passed by the time you read this, the material is still reachable:
- Migration in Motion: How to Move Containers — event page
- The full Migration in Motion on-demand series
What it covers
Three areas, per Nutanix’s own framing:
- Migrating from other orchestration tools to Kubernetes. The legacy-platform problem — Docker Swarm, Mesos, OpenShift on an older footing, or a bespoke scheduler someone built in 2018 and then left the company.
- Moving Kubernetes platforms across providers, both cloud and on-premises. This is the one that has grown teeth: repatriation from public cloud, moving between managed Kubernetes services, or standing up an on-prem platform alongside an existing cloud one.
- Best practices, key considerations, and lessons learned. The part that’s hard to get from documentation.
Why container migration is harder than it looks
The pitch that containers make workloads portable is true in a narrow sense and misleading in a broad one. The container image really does move cleanly. Almost nothing else does.
What actually holds a Kubernetes platform in place is everything around the workload: persistent volumes and the CSI driver behind them, load balancer and ingress implementations that differ per provider, cluster-scoped custom resources, admission and policy controllers, service mesh configuration, secrets management, image registries and their credentials, RBAC bindings mapped to an identity provider, and the CI/CD pipelines that assume all of the above. A Deployment manifest is portable. A cluster is not.
Two things consistently cause the most pain in practice. The first is stateful workloads — anything with a PersistentVolumeClaim needs a real data migration story, and “we’ll just re-deploy it” stops being an answer the moment the data matters. The second is networking: ingress controllers, external DNS, and the assumption that a cloud provider will hand you a load balancer on request. On-premises, that becomes an explicit design decision rather than an API call.
The Nutanix angle
Nutanix’s position here is that the Nutanix Cloud Platform gives you one place to run these workloads across on-prem and cloud, so a migration becomes a move onto a consistent platform rather than a move from one provider’s assumptions to another’s. For anyone already running Nutanix for virtual machines, the relevant point is that container platforms and CSI-backed persistent storage sit on the same infrastructure the VMs do — which removes a fair amount of the storage and networking guesswork that makes these projects slip.
Whether that’s the right answer depends on where your workloads already live and how much of your platform tooling is tied to a specific provider. But the three questions the session organises itself around are the right ones to be asking before a container migration, regardless of which platform you land on.
Worth watching if…
You’re running Kubernetes on a platform you inherited rather than chose; you’re being asked what repatriating container workloads from public cloud would actually involve; or you’re running Nutanix for VMs and want to understand where containers fit alongside them. Forty-five minutes is a reasonable price for the lessons-learned section alone.
Session and series details above are from Nutanix’s own event pages, linked in full at the top of this post. Registration and availability are handled by Nutanix directly.