Benefits of Taking the Less Traveled Road with Containers Infrastructure
summary
After almost a year on OpenShift Origin the team migrates to vanilla Kubernetes, citing a week-long outage during an upgrade and slow two‑week role provisioning as drivers. Key changes: control‑plane processes run under systemd instead of containerized controllers, config moved to Puppet instead of Ansible for reproducible rollouts, a small leader‑election actor handles single‑instance control services across 3–5 masters, and networking drops Open vSwitch so there are addressable pods and no kube-proxy on nodes. Observability adds SLOs (pod start‑up time) and synthetic end‑to‑end probes.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 0:55How we started
- 2:42Replicability
- 5:27Feature Set
- 8:34Observability
- 12:06Master all components
- 12:42Learning
- 13:54Recap
related talks
35mNavigating the Kubernetes Odyssey: Lessons from Early Adoption and Sustained...
36mHow to run like a startup with enterprise Kubernetes
46mAround & After Kubernetes: The Principles and Ideas that Guide Us
41mOpenShift: a Kubernetes Story in an Enterprise Environment
28mOpenShift for the Enterprise or What Could Possibly Go Wrong
31mMigrating Legacy Infrastructure to Kubernetes: Olark’s Lessons From the Trenches
05mIGNITE: Kubernetes/OpenShift (in way too short of time)
49mThe state of on-prem Kubernetes