10 Learnings From Running Production Infrastructure at Google
summary
Ten concrete lessons from running Google production: reliability requires early design choices and a constant voice at the table; change is the top cause of outages so minimize per-change risk and deploy when people can fix problems; use SLOs tied to user experience to decide priorities; prefer cattle not pets and heavy automation to eliminate toil; require blameless postmortems, fast rollbacks, on-call escalation, and refuse no haunted graveyards by deleting fragile, neglected code.
AI-written summary · about
chapters (16)
- 0:00Intro
- 5:26Culture
- 7:001. Reliability can't be taken for granted
- 10:422. Cattle vs. Pets
- 14:113. Blamelessness
- 16:154. Measure what matters
- 19:22A word on Ops
- 20:165. Failure modes
- 21:596. No heroes
- 25:587. Automation
- 27:55Change is constant
- 28:058. Change is No. 1 reason for outages
- 30:429. Outages are inevitable
- 34:3010. No haunted graveyards
- 36:53What did we learn?
- 38:10Outro
related talks
43mGetting Your System into Production and keeping It There
44mPlanning for & Handling Failures from Open Hardware, Aviation, to Production at Google
39mOops. I broke the Google, now what?
58mCultivating Production Excellence
42mHow DevOps Accelerates "Idea to Prod" at Google and What It Means to You
31mRadical Ideas Enterprises Can Learn From The Cloud
40mBuilding Secure and Reliable Systems at Scale
42mYou put out the fire on production, what's next?