Kubernetes Scalability: Federation & Cluster API
summary
Conde Nast runs a global platform across seven Kubernetes clusters (aiming for nine) serving ~250 million pages per month and ~37,000 article views per minute, and prototypes two solutions: Federated Deployment with per-cluster overrides (demo shows a federated namespace and nginx deployment with cluster placement and overrides — replicas 3 vs 5, image nginx:1.17-alpine, annotations) and Cluster API machine/machineSet/MachineDeployment CRDs using kubeadm bootstrap + AWS infra provider to create masters (t3.large) and workers (t2/t2.micro) from a central management cluster.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 0:15Overview
- 3:51What is Federation
- 10:14Demo
- 19:38China Russia
- 20:45Bootstrap Tools
- 21:23Cluster API
- 21:41Cluster Lifecycle
- 22:40How Cluster API Works
- 22:59Controllers Required
- 24:00Provisioning Clusters
- 24:45Cluster API CRDS
- 26:16Demo Setup
- 26:28Cluster API Overview
- 36:09Summary
- 37:23References
- 37:37Contact Information
related talks
25mKubernetes Scalability: Federation & Cluster API - Katie Gamanji, Condé Nast International
13mKeynote: A Journey to a Centralized, Globally Distributed Platform – Katie Gamanji
19mKeynote: Kubernetes Federation
34mKubernetes Multi-Cluster Operations without Federation
30mScaling Kubernetes to Thousands of Nodes Across Multiple Clusters, Calmly
42mQuestions answered: Kubernetes Fundamentals and its Advanced Platform Capabilities
33mMcK8s: Container Orchestration in Kubernetes Multi-Clusters
37mOne Cluster to Rule Them All - Multi-Tenancy for 1k Users