Benefiting from chaos engineering at Capital One (PRT026)
summary
Capital One runs four event types — regional testing events (planned, teams must participate in at least two per year), monthly game day sessions, self‑service experiments, and no‑notice events where PRE injects failures without team prep — all driven by an experiment designer, a catalog of predefined experiments and a fault controller (AWS FIS, Gremlin, Steadybit or SSM scripts) that can stop a run. Experiments are objective‑first (e.g., time to promote an RDS read‑replica, survive 120% peak after an AZ loss, or tolerate intermittent network degradation), produce automated pass/fail metrics and actionable remediation.
AI-written summary · about
chapters (7)
- 0:00Intro
- 0:32Failures happen - Don't panic - Be prepared!
- 2:01Integration of chaos fault injection into resiliency testing events
- 4:48Chaos engineering components critical to our long-term resiliency testing strategy
- 7:33Our objective-driven chaos experiment framework
- 9:52Simulate real-world incidents to validate resiliency readiness objectives
- 11:23Key session takeaways
related talks
08mChaos Engineering at Citi Bank
26mContinuous Chaos in DevOps
57mChaos Engineering Primer
35mImprove resilience with automated chaos engineering
46mChaos Experiments - Datacenter Stress Testing
47mImproving Business Resiliency with Chaos Engineering
18mChaos Engineering: Break It On Purpose
27mChaos Engineering: The Path to Reliability