Overview: Implement a multi-architecture OpenShift cluster with s390x LPAR
You probably didn’t wake up thinking, “I need to add some IBM Z mainframes to my Kubernetes cluster.” But maybe you should’ve. Integrating s390x (a.k.a. IBM Z architecture) workers into your Red Hat OpenShift cluster isn’t just possible. It’s actually useful.
Running containers on mainframes lets you take what some might consider legacy estate – your COBOL apps, your DB2 monoliths – and layer in shiny new microservices without forklifting the whole thing into an overpriced hyperscaler sandbox. You get cloud-native workflows, but on iron that was built for uptime and isolation. Bonus points: s390x punches way above its weight in I/O-heavy workloads, and in some cases, can outcompete cloud infra on cost per transaction.
In this learning path, we’ll go over the steps necessary to add an s390x worker LPAR to an x86 OpenShift cluster. There are different ways to achieve the same outcome, but the reference architecture for this learning path post is based on Figure 1:

We will install the operating system onto a Fiber Channel storage system that is connected to the mainframe.
Prerequisites:
- An existing OpenShift cluster patched to support multi-architecture.
- A logical partition (LPAR) that can communicate with the OpenShift cluster over a network.
- A
jumphost
to serve as an HTTP and SFTP server.
In this learning path, you will:
- Install Red Hat Enterprise Linux CoreOS (RHCOS) on an LPAR.
- Join the node to your existing cluster.
- Walk through deploying a s390x-architecture workload.