Cloud

Configure ECS deployment circuit breakers and rollback while understanding health checks, last-known-good requirements, events, and database limits.

Choose Amazon EventBridge or SQS by separating event routing from durable work ownership, then design retries, ordering, backpressure, and dead-letter recovery.

Design CloudWatch alarms that treat missing data correctly, tolerate isolated samples, reduce notification noise, and remain actionable during real incidents.

Design Amazon S3 lifecycle policies that control storage cost without mistaking delete markers, transitions, or asynchronous expiration for immediate deletion.

Use AWS Secrets Manager rotation with bounded client caches, version-stage semantics, least-privilege access, and deliberate connection reload behavior.

Operate Amazon SQS dead-letter queues as a recovery workflow with deliberate retry thresholds, alarms, diagnosis, controlled redrive, and idempotency.

Separate ECS task and execution roles, scope application permissions to real operations, and diagnose credential and authorization failures without embedding AWS keys.

Design system-level resilience for .NET workloads on AWS with deadlines, bounded retries, jitter, circuit breakers, idempotency, and useful observability.

Design direct-to-S3 uploads with presigned URLs while constraining object keys, expiration, content, authorization, and post-upload processing.

Design reliable SQS consumers in .NET with idempotent processing, deliberate visibility timeouts, safe acknowledgement, and recoverable failures.