Humpty Dumpty A story of API DevOps Gone Wrong
summary
Compares a failed production deploy to Humpty Dumpty and claims cultural fixes beat finger‑pointing: enforce environment parity (dev/staging/prod), give developers read‑only access, store configs in source control, add everyone to an on‑call rotation, and simplify releases to one‑click deploys. Cites reliability math — five nines means ~5 minutes downtime per year — recommends two‑hour pairing, small incremental features, blameless postmortems within 48 hours, and shared ownership; these changes reduce incidents and change how engineers write and deploy code.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:20You see, Humpty is a deploy.
- 3:40Change is hard.
- 6:32Bridge the Skills Gap
- 7:10Read-Only Logs
- 7:44Consistency
- 8:03Source Control
- 8:20On-Call Rotation
- 15:27Questions
related talks
57mGo Away Or I Will Replace You With A Very Little Shell Script
47mOvercoming Obstacles to DevOps
46mContinuous delivery antipatterns
41mContinuous delivery in 10 easy mistakes
43mTo make mistakes is human. To auto-deploy it is DevOps.
28mDevOps Stories; Think outside of your Pipelines
49mDevOps Ten Years After: Review of a Failure with John Allspaw and Paul Hammond
26mHuman Trust in Continuos Deployment