Certainty Among the Chaos
summary
Amazon’s finding that 100 ms ≈ 1% revenue becomes an experimentable hypothesis: use gremlin dial to change CPU, memory, pods, or quiesce 20% of containers and use app dynamics meter to measure checkout latency, user abandons and ticket volume. Start with the highest-traffic transactions, write outcome statements (e.g. reduce pod memory by Y% → checkout +X ms; 100 ms slow → 20 abandons or +15% tickets), then iterate to find where capacity changes buy revenue or save cost.
AI-written summary · about
chapters (9)
- 0:00Introduction and speaker background
- 1:52Finding certainty in chaos
- 3:38Chaos computing explained
- 5:20Starting with load and volume tests
- 7:05Correlating capacity to revenue
- 9:10Testing beyond resiliency
- 10:14Tools for capacity management
- 11:36Next steps for implementation
- 13:40Q&A session begins
related talks
23mHow to tame chaos effectively?
36mCloud Assumes You Know What a Request Will Cost
37m73,000 Pods a Day, Lessons From Misadventures In Multi-Tenant
39mResilience & Chaos Engineering in the Cloud
56mGetting all the 99.99(9) you always wanted
36mMetrics - What's your code actually doing?
26mImprove application resilience with chaos engineering
37mChaos Engineering: The Fine Art of Breaking Stuff in Production on Purpose