Cross-service propagation
Follow one action from the browser through the API gateway, backend services, and database calls in a single distributed trace.
Platform · Traces
See the user behind every backend error.
When a backend service throws, Traceway shows the user's session replay, the cross-service trace, and the exact span that failed. No log-digging. No guessing what happened.
Service facts
The route, by the numbers
What a single user action carries across the whole system.
- 1trace ID, end to end
- Generated by the frontend SDK, forwarded by every service via the W3C traceparent header.
- 4signals linked
- Spans, logs, exceptions, and session replay tied together by that shared trace ID.
- 3async brokers covered
- Kafka, RabbitMQ, and SQS propagate context through message headers, so workers join the original trace.
Through-routed
One trace ID rides the whole way.
From the browser to the last database call, every service forwards the same context. And because Traceway is MIT licensed with full Cloud parity, the whole route works identically on your own metal.
- Propagation
- W3C Trace Contexttraceparent/tracestate headers; any OpenTelemetry-instrumented service is picked up automatically.
- Waterfall
- Every span, every hopDuration and status per hop; click any span for its attributes, logs, and events.
- Replay link
- AutomaticA backend 500 is paired with the frontend replay of the user hitting submit — no manual reproduction.
- Async
- Kafka · RabbitMQ · SQSA consumer's spans land in the producer's trace, so the whole workflow reads as one story.
On this line
Every span. Every hop.
A request becomes a waterfall: duration, status, and the exact failing span, with logs and attributes one click down.
Span waterfall drill-down
Each hop shows duration and status; open any span for its attributes, logs, and events.
Replay on the same ticket
When your payment service returns a 500, you see the user clicking Checkout and hitting submit — linked by trace ID, across browser and server.
Impact propagation
The Impact Score extends across service boundaries, so a degrading upstream service surfaces in every downstream consumer.
Honest fine print
Both ends earn the full picture.
Backend-only distributed tracing works on its own. Linking the user's session replay to a backend error takes the frontend SDK and the backend middleware sharing one trace ID.
Next stop
Trace every hop, not just the first.
Connect an SDK, propagate one trace ID, and see the whole story.