Use this when your service is growing but you can't answer 'what happens if this fails?' for any component. Describe your system architecture to get chaos test designs for network failures, DB connection loss, disk exhaustion, and more — with expected behaviors and auto-recovery procedures.