What the Service Map Shows
The Service Map is X-Ray's most powerful visualization. It is not a diagram you draw — it is auto-generated from the traces X-Ray has collected. Every node is a service. Every edge is a connection between services that X-Ray has observed in real traffic.
Each node displays:
- Average latency (in milliseconds)
- Request rate (requests per minute)
- Error rate (4xx client errors, shown in yellow)
- Fault rate (5xx server errors, shown in red)
- Throttle rate (429 throttling errors, shown in purple)
Each edge (connection between nodes) shows the same metrics for that specific call path.
How to Read a Service Map During an Incident
- Find the red/purple nodes first. Red means faults (your code or a downstream service is returning 5xx). Purple means throttling (a downstream service is rate-limiting you).
- Follow the edges upstream. If DynamoDB is purple (throttled), the Lambda calling it will show elevated latency. The API Gateway calling that Lambda will show elevated latency. The root cause is DynamoDB, but the symptom appears everywhere upstream.
- Click a node to filter traces. Clicking DynamoDB on the Service Map filters the trace list to only traces that touched DynamoDB — immediately narrowing your search space.
- Use the time range selector. Correlate the Service Map with your deployment timeline. If the map turned red at 14:32 and you deployed at 14:30, you have your culprit.
Inferred vs. Instrumented Nodes:
Instrumented nodes (your Lambda, your ECS service) have solid borders. Inferred nodes (DynamoDB, S3, external HTTP endpoints) have dashed borders. Both appear on the map, but only instrumented nodes can provide segment-level detail from their own perspective.