Why and how to build Kubernetes clusters with mixed CPU architectures
summary
The talk explains why and how to build Kubernetes clusters that mix CPU architectures (ARM64, x86, ppc64le, and s390x). It covers using Kubernetes and OpenShift to add heterogeneous worker nodes so applications run on the optimal architecture, and discusses designing container systems for demanding environments, with reference to IBM hardware platforms like LinuxOne and POWER (ppc64le).
AI-written summary · about
chapters (8)
- 0:00Introduction
- 2:15Need for mixed architectures
- 9:09Mixed CPU in OpenShift
- 15:35Adding foreign nodes
- 20:46Scheduler and container images
- 25:44Image tag pitfalls
- 29:02Heterogeneous cluster benefits
- 32:37Future and summary
related talks
36mBare metal Kubernetes on mixed x64 and ARM64
24mPropagating Programming Paradigms: Lifting High Performance Compute Into K8s
38mRunning Multi-Arch Builds & Tests in OpenShift CI/CD
05mLightning Talk: Not One Size Fits All, How to Size Kubernetes Clusters - Jeff Sloyer, IBM
37mAutoscaling a Multi-Platform Kubernetes Cluster Built with kubeadm [I]
08mLightning Talk: Orchestrating Kubernetes Clusters on HPC Infrastructure
20mMulti-Arch VM deployment options with KubeVirt + CDI
36mMulti-Cloud Machine Learning Data and Workflow with Kubernetes