Kubernetes with a single master
summary
Claims that a production workload can keep serving on a single-master kubeadm with embedded etcd cluster, demonstrated by a destroyed master node live demo where a guestbook app continues to respond. Reviews practical tradeoffs: kube-apiserver loss freezes control-plane actions, CoreDNS placement matters, backups and recovery testing are mandatory. Presents an alternative using kine backed by RDS/Cloud SQL to avoid etcd clustering and outsource DB ops, plus cost, availability and operational staffing considerations.
AI-written summary · about
chapters (12)
- 0:00Start
- 0:39PLATINUM SPONSORS
- 1:34Kubernetes with a (single) master
- 2:09Kubernetes Experience - Angel Barrera
- 4:43Agenda
- 10:20Single master with
- 22:03Considerations
- 24:06Where will be located?
- 26:20What will it host?
- 28:09Who is going to manage it?
- 29:07How much budget you've got?
- 29:56Conclusions
related talks
42mMastering Kubernetes: Challenges, Strategies, and Best Practices
06mKeynote: Distributed Systems Simplified Using Kubernetes
36mKubernetes Day 2: Cluster Operations [I]
37mManaging and Running Multiple Kubernetes Clusters in Hybrid Setups
44mKubeCeption! A Story of Self-Hosted Kubernetes
35mKubernetes, Lessons Learnt
36mUnderstanding Scalability and Performance in the Kubernetes Master
01h 31mInside the Kube: A Guided Tour of Kubernetes Cluster Setup