Running Kubernetes in production takes real work. Patching control planes, managing upgrades, and keeping clusters highly available pulls engineers away from shipping applications. Red Hat OpenShift on IBM Cloud removes that operational load. IBM runs the platform as a fully managed service, and your team keeps its attention on workloads. IBM’s May 2026 announcement added two managed services, Red Hat AI Inference and Red Hat OpenShift Virtualization Service, extending the platform into AI workloads and virtual machines.
This article explains how OpenShift on IBM Cloud works in practice, how responsibility splits between IBM and your team, and where the platform fits best. You’ll also see what the new services mean if you’re currently evaluating your options. Finally, we cover protecting the applications and data running on your clusters.
What Red Hat OpenShift on IBM Cloud Is
Before comparing platforms or pricing, it helps to be clear on what this service actually covers. Let’s define it plainly, then connect the underlying technologies for anyone new to the container space.
The Managed Service in One View
Red Hat OpenShift on IBM Cloud is a fully managed OpenShift Container Platform service. IBM operates the platform itself, including the infrastructure that keeps it running, while your team deploys and manages applications on top of it. Internally, IBM refers to the offering as Red Hat OpenShift Kubernetes Service (ROKS).
This is OpenShift on IBM Cloud delivered as a service. You skip the installers, control plane provisioning, and platform build-out entirely. IBM handles that layer and hands your team a working cluster from day one.
The same service also supports community Kubernetes clusters. If part of your organization wants vanilla Kubernetes and another part wants OpenShift, both can run under one managed offering. Keep in mind that even in a managed model, responsibility for application data typically stays with you, which is why questions about Kubernetes backup come up early in most platform evaluations.
How the Pieces Relate
Containers package an application with everything it needs to run, so it behaves the same on a laptop as it does in production. Kubernetes orchestrates those containers, deciding where they run, restarting them when they fail, and scaling them under load.
OpenShift sits on top of Kubernetes and adds enterprise tooling: developer workflows, built-in security controls, and an integrated console. Many production deployments also pair the platform with storage services such as OpenShift Data Foundation to handle persistent workloads. Red Hat OpenShift on IBM Cloud is that same OpenShift platform, operated and maintained for you on IBM Cloud. Think of it as the difference between owning a building and leasing a fully serviced office. The workspace is yours; the maintenance is not your problem.
How Red Hat OpenShift on IBM Cloud Works
The service works because responsibilities are divided cleanly: IBM runs the platform, you run the applications. Here is how that division plays out across day-to-day operations, cluster architecture, and the infrastructure choices available to you.
What IBM Manages vs. What the Client Controls
With a self-managed OpenShift Container Platform installation, your team owns everything, including control plane health, etcd backups, certificate rotation, upgrade sequencing, and node recovery. With Red Hat OpenShift on IBM Cloud, IBM takes full ownership of the master and control plane. That includes patching, version upgrades, automated recovery when components fail, and high availability across the stack.
Your side of the line is the workloads: You deploy applications, configure networking policies for your services, and decide when worker node updates get applied, so a patch never lands mid-release without your approval. IBM’s Site Reliability Engineering team manages the full stack around the clock, and SSH access to worker nodes is disabled by default. That last detail surprises some administrators, but it closes off a common attack path and reflects the managed model: If you never need to shell into a node to fix it, that door can stay shut.
The managed split is simple to remember: IBM keeps the platform healthy, you keep the applications running, and worker updates only happen on your schedule.
Clusters, Worker Pools, and Worker Nodes
Three terms are associated with the architecture:
- A cluster is the complete unit: a control plane plus the compute that runs your containers.
- Worker nodes are the individual machines, virtual or bare metal, where your pods actually execute.
- Worker pools group nodes of the same machine type, which means you can run one pool of general-purpose nodes for web services and another of GPU-backed nodes for inference jobs, all inside a single cluster.
The control plane itself runs three master replicas. If one fails, the other two keep the cluster operational while IBM’s automation replaces the lost replica. You never see the disruption, which is the entire point of paying for a managed service.
Infrastructure Options: VPC, Classic, and Satellite
OpenShift on IBM Cloud runs on three infrastructure types, and the right choice depends on where your workloads and data need to live. The table below summarizes each option, what it provides, and where it fits best.
| Option | What You Get | Best Fit |
|---|---|---|
| VPC | Network-isolated clusters by default, with the newest platform features arriving here first | Most new deployments; IBM's recommended starting point |
| Classic | Bare metal worker options and software-defined storage flavors | Teams with existing classic infrastructure or bare metal performance needs |
| Satellite | IBM-managed clusters running on your own data center hardware or in other clouds | Data residency requirements or hybrid setups spanning multiple environments |
All three deliver the same managed OpenShift experience, so operators and developers can move between environments without relearning tools or workflows.
Automated Red Hat OpenShift Data Protection & Intelligent Recovery
Perform secure application-centric backups of containers, VMs, helm & operators
Use pre-staged snapshots to instantly test, transform, and restore during recovery
Scale with fully automated policy-driven backup-and-restore workflows
Benefits and Common Use Cases
Managed operations are only part of the appeal. Teams choose Red Hat OpenShift on IBM Cloud for a combination of resilience, hardened security defaults, and tight integration with the rest of the IBM Cloud catalog. Here’s what stands out and where the platform tends to earn its place.
Why Teams Choose OpenShift on IBM Cloud
Availability comes first. Multizone clusters distribute worker nodes across zones within a region, so a zone outage costs you capacity rather than taking your application offline. IBM backs multizone clusters with a 99.99% availability SLA, and platform details (including compliance coverage such as SOC 2, ISO 27001, PCI DSS, and HIPAA readiness) are documented in the IBM Cloud documentation. Verify the specific certifications your auditors require before committing, since coverage can vary by region and configuration.
Security defaults matter just as much. Clusters ship with encryption for data at rest and in transit, image security scanning, and the locked-down node access covered earlier. You start from a hardened baseline instead of spending weeks building one. It still pays to layer your own data protection on top; a solid disaster recovery plan covers the application-level failures that platform hardening alone cannot.
Then there’s integration. IBM Cloud IAM handles access control across clusters and services, observability tooling connects without custom plumbing, and the wider catalog (e.g., databases, event streams, and watsonx for AI work) sits one service binding away. On cost, organizations holding existing OpenShift entitlements can often apply them here rather than paying twice for the same subscription.
If you’re weighing OpenShift on IBM Cloud against alternatives, a structured evaluation beats a feature checklist. The following sequence keeps the decision grounded in your actual workloads:
- Inventory your workloads: Flag anything with strict latency, data residency, or bare metal requirements, since that determines whether VPC, Classic, or Satellite fits.
- Map your compliance obligations: Check them against IBM’s published certifications for your target regions.
- Review your entitlements: Confirm whether existing Red Hat OpenShift subscriptions can transfer, which changes the cost math considerably.
- Run a pilot: Deploy a representative workload on a multizone cluster and simulate a zone failure before signing off.
Working through these steps surfaces dealbreakers early, while the pilot is still cheap to walk away from.
Where It Gets Used
Mission-critical applications are the most common fit: customer-facing systems where downtime translates directly into lost revenue and multizone resilience earns its keep. Application modernization follows closely, with teams decomposing legacy monoliths into containers and using OpenShift’s developer tooling to standardize how services get built and shipped. If your team is still weighing that shift, comparing containerization against virtualization is a useful starting point.
AI and machine learning workloads increasingly land on Red Hat OpenShift on IBM Cloud as well, drawn by GPU worker pools and proximity to watsonx. Regulated industries, finance and healthcare in particular, gravitate toward the platform because the compliance groundwork and encryption defaults reduce the audit burden they would otherwise carry themselves. A bank modernizing its payments stack and a hospital network containerizing patient-facing portals face very different constraints, yet both end up wanting the same thing: a hardened platform that someone else keeps healthy.
What's New: The May 2026 Managed Services on IBM Cloud
IBM keeps widening what “managed” means on this platform. The newest additions move past containers into AI model serving and virtual machines, two workload types that have historically demanded heavy operational effort from the teams running them.
The Announcement
On May 12, 2026, IBM introduced two new managed services on IBM Cloud: Red Hat AI Inference and Red Hat OpenShift Virtualization Service. IBM is extending the same managed model behind Red Hat OpenShift on IBM Cloud to AI inference and VM workloads, so teams running either get the platform’s capabilities without owning the plumbing underneath.
Red Hat AI Inference is a fully managed service for running production-grade AI models through OpenAI-compatible APIs, and it reached general availability on May 22, 2026. The design follows a models-as-a-service pattern: Models run as shared resources exposed through APIs, powered by vLLM and Red Hat AI’s inference engine. That directly targets the pilot-to-production gap, the point where enterprises build promising AI experiments and then stall because nobody wants to own GPU scheduling and runtime tuning at steady state.
Red Hat OpenShift Virtualization Service on IBM Cloud
The second service gives teams a managed path to migrate and run virtual machines at scale on OpenShift. It ships with the Migration Toolkit for Virtualization and automated lifecycle management, making it a practical bridge for organizations moving legacy VMs toward containerization at their own pace. The service runs on IBM Cloud VPC Bare Metal and is positioned for predictable performance and total cost of ownership. It has been generally available since June 26, 2026.
The table below summarizes what each new service handles and which operational burdens it takes off your team.
| Service | Workload | What IBM Removes from Your Plate |
|---|---|---|
| Red Hat AI Inference | Production AI model serving via APIs | GPU management and model runtime tuning |
| Red Hat OpenShift Virtualization Service | Virtual machines on OpenShift | Migration tooling setup and lifecycle operations |
Why This Matters for Teams Evaluating the Platform
If you’re comparing platforms right now, these launches signal direction. The managed footprint of Red Hat OpenShift on IBM Cloud now covers containers, VMs, and AI inference, which means fewer separate stacks to operate and fewer specialist skills your team has to maintain in-house. For organizations consolidating infrastructure, that consolidation happens under one operational model instead of three.
Protecting What You Run: Backup and Recovery Considerations
The managed model does not cover your applications and their data. IBM’s shared responsibility documentation assigns disaster recovery for both applications and data entirely to the customer, and notes that persistent storage for things like application logs is not configured by default. IBM does provide the hooks: storage classes, CSI snapshot support, object storage targets, and integrations with partner technologies such as Portworx. What it does not provide is a managed service that recovers a deleted namespace, a corrupted database, or a ransomware-hit workload for you. Assembling and operating that capability is your team’s job.
A managed platform protects the platform. Protecting the applications, data, and Kubernetes resources running on it is still your job.
This is where a purpose-built tool earns its place. Trilio’s OpenShift Backup and Recovery captures full snapshots of application data alongside Kubernetes objects, metadata, and configurations, so an entire environment can be restored accurately rather than piece by piece. Incremental backups save only what changed since the last run, which cuts storage costs and shortens backup windows. Backups can be scheduled across on-premises, hybrid, or cloud environments, and restores can target any x86_64 OpenShift cluster, keeping workloads portable across the VPC, Classic, and Satellite options covered earlier. Role-based access control, retention policy management, and backup status reporting round out the compliance side. To see how Trilio fits your OpenShift on IBM Cloud environment, schedule a demo for a walkthrough.
Learn why DETASAD chose Trilio’s self-service backup over others
Conclusion
Red Hat OpenShift on IBM Cloud is a reasonable choice when you want what OpenShift offers but would rather not carry the day-to-day operational burden yourself. IBM takes responsibility for the control plane, upgrades, and recovery, while you retain control over your workloads and decide when updates happen. Now that AI inference and VM management sit inside the same managed model, the platform can handle a wider range of workload types under a single operational approach than it could a year ago.
The one thing this managed service does not do for you is recover lost application data. IBM’s own responsibility documentation is explicit about it: disaster recovery for applications and data sits on your side of the line. Platform health and data recoverability are two separate concerns, and teams that treat them separately from day one avoid some hard lessons down the road. If you are considering OpenShift on IBM Cloud, run a pilot first, simulate a zone failure, and put your recovery process under real pressure before anything critical goes into production. A cluster you can rebuild is useful, but applications and data you can restore anywhere, on a timeline you control, are what actually keep the business running.
FAQs
Is Red Hat OpenShift on IBM Cloud fully managed?
Yes, IBM operates the control plane, applies platform patches, and handles automated recovery around the clock. Your team stays responsible for deploying applications, approving worker node updates, and protecting workload data.
What is the difference between ROKS and self-managed OpenShift?
ROKS is IBM’s internal name for its managed OpenShift service, where clusters arrive ready to use and IBM maintains the platform layer. With self-managed OpenShift, your engineers handle everything from installation to certificate rotation and etcd recovery.
Which infrastructure option should I choose for Red Hat OpenShift on IBM Cloud?
VPC suits most new deployments and receives new features first, while Classic fits teams needing bare metal performance. Satellite is the right pick when clusters must run in your own data center or another cloud for data residency reasons.
Does Red Hat OpenShift on IBM Cloud include backup for my applications?
Not as part of the managed service. IBM operates the platform and backs up cluster configuration in etcd, but application data recovery is documented as the customer’s responsibility. You can run the OpenShift API for Data Protection operator yourself against IBM Cloud Object Storage, though you own its installation, configuration, and testing, and it does not cover operators or etcd. Either way you need a dedicated tool that captures application data together with Kubernetes objects and configurations.
Can I run AI models and virtual machines on the same managed platform?
IBM’s May 2026 additions brought managed AI inference and a virtualization service into the same operational model as containers. This lets teams consolidate model serving, VMs, and containerized apps without maintaining three separate stacks.
