How a 3-box architecture diagram hid a 7-hop pipeline — and why it mattered

A software team at ALSC discovered their C4 container diagram showed only three components for a batch upload flow, while the actual runtime process involved seven distinct execution steps across Node.js functions, blob storage, Service Bus, and Hangfire background jobs. The oversimplification went unnoticed until a production incident forced engineers to manually correlate logs across four separate teams, consuming significant time with no clear trace of where the failure occurred. The team identified that C4's container-level diagrams are designed to show deployable units and structural relationships, not runtime execution sequences, and dynamic diagrams that would have captured the full flow were never prioritized. To resolve the gap, the team implemented distributed tracing using OpenTelemetry, though a key challenge was that Hangfire's Redis-backed queue does not propagate request context by default, requiring explicit code-level instrumentation. Once full trace propagation was established across all seven hops, incident resolution shifted from a half-day manual log-grepping exercise to a quick review of a single trace in Application Map.