Red Hat OpenShift

What Red Hat OpenShift actually is

OpenShift is Red Hat's enterprise Kubernetes platform: a curated, supported distribution of Kubernetes with the operational and developer layers already assembled — an installer and lifecycle management, a web console and CLI, integrated CI/CD (OpenShift Pipelines on Tekton and GitOps on Argo CD), an internal container registry, build tooling, a service mesh, security context constraints and policy controls, and a certified Operator catalog. Red Hat introduced OpenShift in May 2011 as a Platform-as-a-Service for developers; it pivoted to Kubernetes with OpenShift 3 and is now sold as the application platform underneath Red Hat's hybrid-cloud story, run in production by a large share of the Fortune 500 and most regulated industries.

The honest framing is this: you are not buying infrastructure here, you are buying a supported platform layer. OpenShift runs on your own hardware (including bare metal), on virtualisation, on AWS, Azure, Google Cloud or IBM Cloud, or as a managed service operated by Red Hat's SRE team — OpenShift Dedicated, Red Hat OpenShift Service on AWS (ROSA), Azure Red Hat OpenShift (ARO), and OpenShift on IBM Cloud. The self-managed editions stack up in tiers: an entry-level enterprise Kubernetes runtime, the standard Container Platform (the full developer and operations toolchain), Platform Plus (multicluster security and management — Advanced Cluster Management, Advanced Cluster Security, Quay, Data Foundation), and a virtualization-only edition for migrating VMs onto Kubernetes (KubeVirt-based OpenShift Virtualization).

What separates it from rolling your own cluster is opinionation and accountability. Routes, BuildConfigs, ImageStreams, Security Context Constraints and the Operator Lifecycle Manager are OpenShift's own abstractions on top of Kubernetes, and one throat to choke when the platform breaks — Red Hat's support, with 24x7 severity-1 coverage on Premium. That is the product. Whether it is worth the subscription is a question of scale, compliance, and how much platform engineering you want to own.

Pricing: what you really pay

OpenShift pricing is mostly sales-mediated, and we will not pretend otherwise. Self-managed editions (Kubernetes Engine, Container Platform, Platform Plus, Virtualization Engine) are licensed by 2-core packs or bare-metal node entitlements with Standard or Premium support; complete list prices are quote-driven and Red Hat does not publish a public price table. Third-party analyses put 2026 list pricing in the region of $1,000–$2,500 per 2-core pack per year depending on edition and support tier — figures we treat as estimates from published analyses, not vendor numbers. The consequence is predictable: at fleet scale, licensing is the dominant cost. The same analyses model a 100-node cluster with 16-core workers at seven figures per year in subscriptions alone, on top of the compute you buy anyway.

The managed services do have published rates. When Red Hat cut OpenShift Dedicated pricing in October 2020, the published numbers were a control-plane fee of $0.03/hour (about $263/year per cluster) and worker nodes from $0.076/hour — roughly $667/year for a 4 vCPU worker on a three-year commitment, billed annually for OpenShift Dedicated, with on-demand hourly rates higher (current marketplace rates cited in 2026 reviews run around $0.33/hour for an m5.large-class worker and $0.65/hour for an m6a.xlarge-class worker). Cloud infrastructure costs are billed separately via AWS, Azure, Google Cloud or IBM Cloud.

There are cheaper doors in. A no-cost Developer Sandbox exists for learning, OKD is the free community distribution for non-production use, a 30-day trial path is offered, and the "Level Up" programme gives eligible RHEL customers a free one-year Platform Plus subscription (2-core/4 vCPU, Standard support, non-renewable). None of these are a free tier of the production product. Our practical advice: budget from a real quote, not from list-price rumours, and push for marketplace committed-spend deals — Red Hat discounts are negotiated, and nobody pays list at volume. We track the euro-denominated Kubernetes platforms on /category/kubernetes-managed for easier cross-comparison.

Performance & infrastructure

OpenShift's performance profile is the hardware you attach it to. The platform itself is a scheduler, an overlay network, an ingress layer and a build system; it neither sells compute nor publishes benchmark numbers for it. What is verifiable is the breadth of deployment targets: installer-provisioned and user-provisioned infrastructure on all major clouds, VMware, OpenStack, bare metal (including a Red Hat Bare Metal as a Service option and IBM Cloud Bare Metal), IBM Z/LinuxONE, and single-node and two-node edge topologies. NVIDIA GPU workloads are supported through the NVIDIA GPU Operator (drivers, device plugin, container toolkit, GPU feature discovery, DCGM monitoring) with GPU passthrough, MIG, time-slicing and vGPU options, and NVIDIA-certified bare-metal GPU servers are a documented deployment class — note that Red Hat sells no GPU capacity itself, and NVIDIA supports the GPU Operator, not Red Hat.

Networking is OVN-Kubernetes by default, with full IPv6 single-stack and IPv4/IPv6 dual-stack support, IPsec for intra-cluster traffic, egress IPs and firewalling, hardware offloading, and multicast. In our view the dual-stack story is genuinely complete on supported platforms — and Red Hat explicitly does not support disabling IPv6, which tells you where their networking priorities sit. Storage runs through the Container Storage Interface with OpenShift Data Foundation as the Red Hat-blessed option (including an S3-compatible object gateway as an add-on).

For managed services, the reliability commitment is written down: 99.95% uptime SLA on OpenShift Dedicated, ROSA and ARO, rising to 99.99% on OpenShift on IBM Cloud, with Red Hat SREs operating the control plane (they have run Kubernetes at scale since 2015) and 24x7 Premium support included. That is a real, credit-backed SLA — a different category from the self-managed subscription, where availability is your own operational record. We have not benchmarked OpenShift clusters; our /benchmarks numbers cover the underlying hosts from other providers, and platform overhead is a per-workload question we would not generalise.

Where it falls short

Cost, unambiguously. Per-core subscriptions that scale with worker count make OpenShift the most expensive way to run Kubernetes per unit of compute, and the pricing opacity compounds it: enterprise list prices are not public, discounts are negotiated, and total cost depends on edition, support tier, add-ons (ACM, ACS, Quay, Data Foundation), professional services and training. If your workload does not need the compliance and support apparatus, upstream Kubernetes plus a managed control plane is a fraction of the price.

Second, weight and lock-in. The opinionated layer is thick: Routes, BuildConfigs, ImageStreams and SCCs are OpenShift constructs that require rework if you ever leave the platform. Upgrades are managed and safe but move at Red Hat's cadence, the platform demands real cluster-administration skill, and the abstractions that protect enterprises frustrate engineers who know vanilla Kubernetes. Restricted-v2 security constraints reject manifests that run fine elsewhere, which is either the point or the annoyance depending on your day.

Third, buying friction. Self-managed pricing requires a sales process; the free paths (Developer Sandbox, Level Up) are deliberately time-boxed or tier-limited; and the portfolio is genuinely hard to navigate — four self-managed editions, four managed services, and overlapping add-on bundles. We regularly see teams buy Platform Plus when Container Platform would do.

Who should pick this

We would pick OpenShift when the organisation needs a supported, certified, auditable Kubernetes platform rather than a cluster: regulated industries (finance, government, healthcare, telecom) where vendor support and certification evidence are functionally mandatory, VMware refugees consolidating VMs and containers onto one platform via OpenShift Virtualization, hybrid estates that must run the same platform on-premises and across several clouds, and large platform teams standardising hundreds of clusters with ACM and ACS. The ROSA/ARO route is our pick when you want AWS or Azure integration and a single bill without operating the control plane yourself.

We would not pick it for a startup, a single team, a cost-sensitive workload, or any organisation with strong Kubernetes skills and no compliance driver. There, managed Kubernetes from a hyperscaler or a European provider is cheaper, lighter, and free of the abstraction tax.

Key takeaways

• Enterprise Kubernetes platform (launched May 2011 as a PaaS, now Kubernetes-based) with integrated CI/CD, registry, service mesh, policy controls and a certified Operator catalog.
• Runs anywhere: on-prem, bare metal, VMware, all major clouds, plus managed services (Dedicated, ROSA, ARO, IBM Cloud) operated by Red Hat SREs.
• Managed SLA is 99.95% uptime (99.99% on IBM Cloud) with 24x7 Premium support — a written, credit-backed commitment.
• Self-managed pricing is quote-driven; third parties estimate roughly $1,000–$2,500 per 2-core pack per year; managed rates published at $0.03/hour control plane plus workers from $0.076/hour.
• Full IPv6 single-stack and dual-stack support via OVN-Kubernetes; NVIDIA GPU workloads via the GPU Operator on certified hardware.
• Strong platform with real lock-in (Routes, BuildConfigs, SCCs) and the highest per-core cost in our /category/kubernetes-managed comparison — worth it for compliance, rarely for cost.

Sources

• https://www.redhat.com/en/technologies/cloud-computing/openshift
• https://www.redhat.com/en/blog/red-hat-openshift-dedicated-price-reduced-by-75-on-average-and-sla-improved-to-99.95
• https://www.redhat.com/en/resources/application-services-subscription-guide-detail
• https://access.redhat.com/level-up/red-hat-openshift-container-platform
• https://docs.redhat.com/en/documentation/openshift_container_platform/4.15/html/networking/ovn-kubernetes-network-plugin
• https://docs.redhat.com/en/documentation/openshift_container_platform/4.22/html/hardware_accelerators/nvidia-gpu-architecture
• https://www.redhat.com/en/about/press-releases/red-hat-delivers-platform-service-cloud-open-source-developers
• https://www.redhat.com/en/blog/the-openshift-journey-a-tale-of-cloud-devops-microservices-and-containers
• https://tech-insider.org/openshift-vs-kubernetes-2026/
• https://www.rfp.wiki/cloud-computing/cloud-native-application-platforms/red-hat/cloudflare