DOP-C02 Practice Question 36: X-Ray Trace Map Latency

X-Ray correlates the segments of one request across every service it touches, which is what "which hop was slow on *this* request" requires. The documentation describes the map's purpose as "identify services where errors are occurring, connections with high latency, or traces for requests that were unsuccessful", showing "service nodes that serve requests, upstream client nodes... and downstream service nodes." You can "use the response distribution histogram to filter traces by duration" and then open the matching traces, so the slow tail can be isolated instead of being averaged into the rest, and the timeline shows which segment consumed the time.

Option A looks at resource utilization, not request attribution. Container metrics can hint that a service is under pressure, but they cannot tell you that a particular checkout spent four seconds waiting on one downstream call.

Option B makes the problem worse. Summing per-component durations aggregates across all requests, so the slow tail stays buried exactly as it is today, and metric math has no notion of a single request's path through the system.

Option D averages again, and averages per log group. Even a per-request duration in each log group cannot be joined across services without a shared trace identifier, which is the thing X-Ray provides.

Reference: Using the X-Ray trace map