How we Won the Super Bowl (Commercials)
summary
A small engineering team prepares for a surprise Super Bowl ad by prioritizing exact fixes: raising the AWS SDK socket limit, deploying PG bouncer to reduce DB CPU from connection churn, and migrating to a new VPC and k8s cluster to avoid IP exhaustion. Non‑essential services get turned off and waiting rooms plus fallbacks remain for signup. Practice ads reveal real traffic, rapid CI deploys (spiking to 75 deploys in a day) let iterative changes land, and the system survives with capacity to spare.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:14The Farmer’s Dog background
- 3:02Project Daisy secrets
- 4:04Resilience engineering surprises
- 5:09Robustness vs resilience
- 6:35Rasmussen model
- 8:21Workload and capacity
- 10:31Playoff testing
- 12:17Team structure
- 13:44Continuous deployment safety
- 15:31Deployment pipelines
- 16:35Adaptive capacity
- 17:18Dynamic reconfiguration
- 18:46Shared goals under pressure
- 20:10Signup flow bottlenecks
- 21:54Post-crisis optimizations
- 22:56Turning off services
- 23:38Failure preparation
related talks
26mHow We Prepped Our Infrastructure for a Super-Bowl Sized Traffic
28mReliability Lessons from Running High Traffic Services for More Than 10 Years
54mAdrian’s Greatest Hits, B-Sides & Re-issues
32mHow We Survived (and Thrived) During The Pandemic and Helped Millions...
26mRisky business: taking risks in production
01h 04mOpening Keynote [Full]
24mSleeping soundly even when scaling your app to 200k request/seconds
31mDevOps in the Cloud: Case Studies of Amazon.com teams and their resilient architectures