Static stability using availability zones
summary
Defines static stability as over‑provisioning so services keep running without launching new instances when an Availability Zone fails, showing EC2’s split control plane / data plane design, two architectures—active‑active (stateless, load balanced) and active‑standby (stateful, warm standby)—and the concrete rule to over‑provision by 50% for three AZs. Describes zonal NAT Gateway and zonal deployment calendar practices and claims these choices let infrastructure survive an AZ impairment with no recovery actions.
AI-written summary · about
chapters (10)
- 0:00Static stability intro
- 2:06AWS infra and AZs
- 5:40Static stability concept
- 8:53Control vs data plane in EC2
- 11:27Over-provisioning for AZ resilience
- 13:54Active-standby for stateful services
- 16:20AZ independence in EC2
- 18:49Regional vs zonal dependencies
- 23:23Lessons learned and patterns
- 25:09Closing and resources
related talks
12mMulti-Zone Service Architecture for 99.99% Uptime
49mPatterns for resilient, highly available cloud architectures
32mCreate a cloud-ready environment
33mMany Layers of Availability
54mResilient by Design
28mHighly Available Cloud Foundry Deployment for Allstate
22mFailing Over without Falling Over
07mHow SingleStore saves 56% on Amazon EC2 with no DevOps hours invested