Gokart to Grand Prix: How It Feels To Actually Go Fast
summary
Deus replaces a brittle fleet+etcd+cloudconfig platform with Kubernetes and many smaller git repos plus Jenkins CI and parallel test workers, cutting a 90 minute cycle to 15 minutes by getting 4 minute builds, 1 minute deploys and 10 minute tests. A cache of prewarmed clusters called the Cates claimer speeds cluster provisioning. The change is incremental over six months and it measurably works — six times faster lap times and substantially fewer test failures.
AI-written summary · about
chapters (23)
- 0:00Intro
- 1:17Who are we
- 2:21Why are we here
- 3:37Deus
- 5:10Summary
- 6:34Breaking Bad
- 7:20How did we get here
- 7:59When engineers start to engineer
- 9:12Managing containers
- 10:18What value do you want
- 11:02Formula One Pit Crew
- 11:53Build Time
- 13:02Deployment
- 13:50Tire Warmer
- 14:57Warm Kubernetes Clusters
- 15:41Deployment Time
- 16:37Software Testing
- 17:57Parallel Testing
- 19:35Development Cycle
- 19:48Going Fast
- 21:03Growth is never comfortable
- 22:31Questions
- 26:22Speed isnt all about speed
related talks
25mFrom Duct Tape To Declarative: Playtika's Platform Overhaul at Unicorn Scale
33mFrom Slow To Pro: Accelerating Kubernetes Native Devel...
35mKubernetes journey
39mStrangling the Legacy with Kubernetes
34mKubernetes from Dev to Prod at GoEuro
28mPushing Git repos over the speed limit.
35mYocto Continuous Integration in a Kube
48mProgressive Delivery in the Kubernetes Era