Observability · Flashcard
A 300 ms trace holds a 250 ms database span. What does that show?
- AWhere the time went — that call dominates the request
- BThat the database is saturated — it needs more capacity
- CThat the other spans ran concurrently — alongside the query
- DNothing yet — a span's duration excludes time spent waiting
Why this is the answer
Each span carries its own start and end timestamp, so comparing a child's duration with the trace's shows which step the latency sits in — here the database call accounts for most of it. Why it is slow is a further question a single trace cannot answer; parallelism has to be read off the timeline rather than assumed; and a span's duration is wall-clock, waiting included.
Official docs