Microservice Meltdown
Keep checkout alive when dependencies fail. Diagnose the bottleneck, spend a limited intervention budget, then replay the evidence.
Challenge a friend: SRE vs Chaos →On-call briefing
Auth is the universal choke point. 60 req/s steady traffic. A single +50 requests/tick spike fires at tick 60. SRE: scale Auth or shed load before its queue fills.
You have 4 scaling moves, 6 breaker changes and 6 load-shed adjustments. Pause to investigate. The incident runs for two simulated minutes, even through outages, so recovery matters.