Architecting for Reliability
summary
Panelists describe concrete steps: they replace brittle DNS blue‑green flips with ALB canaries, move complex logic from Lambda back into the cluster, and use Terraform plus sidecars for consistent rollout. One team records full request/response traces to replay production bugs in isolated environments; another enforces APM, profiling and custom metrics via Datadog. They run automated rebuilds every two weeks for 18 months, run game days and chaos testing, and face integration with roughly 90 backend systems while preparing a fleet of 250 new aircraft.
AI-written summary · about
chapters (9)
- 0:00Introductions
- 2:34Architecting for simplicity
- 6:15Consistency and automation
- 11:56Testing and observability
- 16:00Scale and observability
- 21:04Predictability and debugging
- 25:21Handling ephemeral issues
- 32:23Measuring reliability
- 38:57Closing thoughts
related talks
46mDeploying and maintaining software for today's infrastructure
45mDeveloping a Chaos Architecture Mindset
31mDevOps in the Cloud: Case Studies of Amazon.com teams and their resilient architectures
31mHow We Went From Two Major Outages to 99.98% Reliability in 6 Months
25mLive Q&A Session #2
42mPlanning To Fail
24mLive Q&A Session #3
48mChaos Engineering Panel