18 Months Before the Mast
summary
Production runs on Kubernetes 0.17 on CoreOS + Fleet with images built via SBT into Jenkins and stored in Artifactory, going live June 2015 and growing from 35 to ~80 nodes across nine sites. Major wins: devs push immutable containers and iterate minutes instead of hours. Major failures: kube-proxy user-space causes high latency until iptables is enabled, container logging throughput tops under ~1,000 events/s, and CloudFormation ALB changes can delete an Auto Scaling Group. Migration to AWS, CFSSL for TLS, and ASG autoscaling largely work.
AI-written summary · about
chapters (4)
- 0:00Start
- 2:45Late autumn 2014: "The Deployment Problem"
- 12:00Organizational decision: divide and conquer
- 39:17Questions
related talks
17mDeployments We Can Believe In
35mKubernetes journey
36mFrom Screen to Pods: Bootstrapping a Cloud Agnostic System using Kubernetes [I]
39mStrangling the Legacy with Kubernetes
36mPractical Cloud Native: What Works, What Doesn't
32mAutomating and Testing Production Ready Kubernetes Clusters in the Public Cloud
38mBuilding the Perfect Infrastructure with Kubernetes
33mNavigating the Seas of Transformation: Migrating from EC2 to Kubernetes