Scaling up to your first 10 million users - The Kubernetes Edition
summary
Kubernetes practical checklist for your first 10M users with concrete limits and fixes: follow documented scale caps like 110 pods per node (soft limit), 5,000 nodes, 150k pods, design for pod density ~50–70 pods/node, enable node DNS caching and custom networking or IPv6 dual‑stack for >100 nodes, push images through ECR pull‑through cache, use tooling like Carpenter autoscaler, kube-node-usage and kube-capacity, separate instance types/spot/GPU, favor GitOps, infra-as-code, and optimize daemonsets and dependency lifecycle.
AI-written summary · about
chapters (21)
- 0:00Introduction and agenda
- 1:45Cluster sizing considerations
- 3:32Pod density and resource limits
- 6:42DNS caching at scale
- 9:55Node DNS caching for performance
- 10:37Networking for 10+ nodes
- 11:19Custom networking for 50+ nodes
- 12:21IPv6 for 100+ nodes
- 13:03Image registry optimizations
- 15:11Image duplication and caching
- 16:12Autoscaling strategies
- 17:16Instance differentiation at scale
- 18:00Container vs serverless workloads
- 19:27Operations and cluster reliability
- 20:50Access control and IAM roles
- 21:34GitOps and deployment reliability
- 22:16Demon set optimization
- 22:56Cluster access management
- 23:38Change management and dependencies
- 25:46Summary and scaling best practices
- 28:17Q&A session
related talks
32mScaling Kubernetes Application to 1M Users
37mOne Cluster to Rule Them All - Multi-Tenancy for 1k Users
34mPanel: Improving and Managing Kubernetes at Scale
42m2000 Nodes and Beyond: How We Scaled Kubernetes to 60,000-Container Clusters
50mScale Kubernetes to Manage 60K Nodes Through Architectural Extension
15mDemo: Resource(Cost) optimization on planet scale Kubernetes clusters.
30mScaling Kubernetes to Thousands of Nodes Across Multiple Clusters, Calmly
35mTo Infinite Scale and Beyond: Operating Kubernetes Past the Steady State