Reliability & Operations

N+1 and N+2 redundancy, active-passive vs active-active, health checks, failover, and cell-based isolation.

Timeouts from percentiles, retry budgets, backoff with jitter, circuit breakers, and hedged requests.

Token and leaky buckets, fixed and sliding windows, distributed limits, per-tenant quotas, and load shedding.

Rolling, blue-green, and canary deploys, feature flags, backward-compatible changes, and rollback.

Metrics, logs, and traces, RED and USE, cardinality, distributed tracing, percentile traps, and symptom-based alerts.

Backups vs replicas, point-in-time recovery, RTO and RPO, restore drills, and region failures.

Load vs stress testing, headroom, peak provisioning vs autoscaling, game days, and fault injection.