Why file-transfer status should be decoupled from the transfer itself

A team asked whether to replace a Service Bus trigger with an HTTP trigger so a single request could report a file's location across Azure, an on-premises hop, and customer delivery. The author argues that operational excellence is typically the first Well-Architected pillar sacrificed under deadline pressure, and that a synchronous design would be fragile given uncontrolled networks and customer downtime. The recommended fix is a correlation ID per file, with each stage writing a small timestamped record to cheap queryable storage, so status reads return quickly without waiting on the transfer. Long-running callers can receive a 202 Accepted with a status URL to poll. The on-premises agent should push its own status events outbound-only, since it is the authority on the final stages and the cloud must not infer delivery from a mere handoff.