Episode 10. Full scenarios. Day 10. Four timed mocks. No new facts. Only reps.

Rules. Fifteen minutes each. Two minutes to Understand out loud. Three minutes to Prepare. Five minutes to Identify with a committed recommendation. Two minutes to Support. Three minutes spare. Say: my recommendation is, in every single mock. No hedging.

Mock 1. Aurora PostgreSQL primary failed over at 2am and the app never recovered. Connections hang. Diagnose it.

Understand. Ask: writer endpoint or custom hardcoded host? Pool behavior on DNS change? Failover event time? Reader versus writer split correct?

Prepare. Say the data: failover events, endpoint DNS TTL, pool idle lifetime versus DNS cache, connection error mix.

Say your translation line: Same instinct as SCAN listener failover: the endpoint moved and the pool did not follow.

Identify. Rank: stale DNS in pool, idle connections to the old writer, missing retry with backoff. Fix: use the cluster endpoint, shorten DNS and idle lifetimes, add failover aware retry. Support: chaos failover drill quarterly, alarm on failover events.

Mock 2. DynamoDB costs tripled after a launch. Same traffic shape. Explain it.

Understand. Ask: on demand or provisioned? Scans introduced? GSI backfill running? Item sizes grown?

Prepare. Say the data: consumed versus provisioned units, top operations by cost, scan volume, item size histogram.

Identify. Rank: new scans, oversized items, GSI write amplification. Fix: replace scans with queries, compress or trim items, right size GSIs. Support: cost anomaly alarm, weekly top operation review.

Mock 3. Cross Region Aurora Global Database lag is growing. Reads are stale.

Understand. Ask: write volume change? Large transactions? Network path change? One secondary or all?

Prepare. Say the data: replication lag metric, redo volume, long transactions, network throughput.

Identify. Rank: write flood, giant transaction blocking replay, binlog or storage throughput cap. Fix: split the giant writes, throttle the flood, verify instance and network headroom. Support: lag alarm with paging threshold.

Mock 4. RDS storage filled overnight and the instance stopped. Recover it.

Understand. Ask: storage autoscaling on? What grew: logs, temp, audit? Can it start?

Prepare. Say the data: free storage metric, log sizes, temp usage, snapshot recency.

Identify. Rank: WAL or binlog retention too high, runaway temp from a bad query, audit log flood. Fix: extend storage to restart, purge or rotate logs, fix the query, enable autoscaling. Support: free storage alarm at twenty five percent, log rotation review.

Your Day 10 lab: run all four mocks with a timer. Record yourself, find one missing recommendation sentence, and redo that mock.
Tomorrow, Episode 11: behavioral and Leadership Principles. Stories that score. Understand. Prepare. Identify. Support.