Plain logs from five services interleave into noise — impossible to follow one request. Switch on structured logs with a shared trace id, filter to one id, and the request's whole journey across services reconstructs itself.
Two requests are running at the same time — a booking (A) and a report download (B). Toggle structured logging and try to follow just one of them.
Log stream
00:00:00 POST /book received
00:00:00 GET /report received
00:00:00 token validated (alice)
00:00:00 token validated (bob)
00:00:00 creating booking #88213
00:00:00 fetching report #5521
00:00:00 charge timed out after 8s
00:00:00 report ready
00:00:00 fallback: pending payment
00:00:00 200 OK
🌀 No ids, two requests mixed together — which 'timed out' line belongs to the booking? You can't tell. This is most logging.
What just happened
▹Plain text logs from many services interleave into noise. When two requests run at once, their lines mix together and you can't tell which line belongs to which request.
▹Structured logs attach fields — service, level, and crucially a trace_id (correlation id) created at the edge and passed to every service. Filter by one trace_id and the entire journey of that single request reconstructs itself across services, in order.
▹Also: never log secrets, tokens, or full patient records. Log enough to debug (ids, actions, status, latency) and nothing that would itself be a breach if the logs leaked.