Graceful degradation — CCA-F Exam Prep

PencilPrepPencilPrep
L2.30|Graceful degradation
1/12
Countdown
A production monitoring dashboard at 2:47am. Three panels cascading from yellow to red. Panel 1: 'Claude API latency: 12s (normal: 2s).' Panel 2: 'Error rate: 34% (normal: 0.1%).' Panel 3: 'Customer complaints: 847 in last hour.' A countdown timer in the corner: 'SLA breach in: 00:13:00.' Dark control room.

2:47am. The Claude API starts returning errors on 34% of requests. 847 customers are stuck.

The customer service agent has no fallback. When the API errors, customers see 'Something went wrong. Please try again.' They try again. Same error. They try again. Same error. They leave.

Meanwhile, the database tool is timing out on 15% of lookups. Conversations that worked fine at 2pm are failing at 2am because the database is under maintenance. No fallback. No cached responses. No human handoff. Just errors.

Every component failed individually. The system failed completely. Because nobody planned for failure.