Understand exactly where time is spent.
P50/P95/P99 percentiles for every endpoint, span waterfall traces, and per-endpoint drill-downs. Know exactly what's slow and why.
One score. Five signals. Zero guesswork.
The Impact Score combines five service-level indicators into one automatic priority for every endpoint. It takes the max across all five — if any single signal is bad, the endpoint surfaces immediately.
Drill into any request
Request/response details, waterfall traces, and custom context tags. Understand the exact state of your application for every single trace — not just aggregates.
- Detailed request/response payloads
- Waterfall trace view with span attributes
- Custom context and tag facets
- Filter by status, latency, or tag
Endpoint analytics at a glance
See P50, P95, and P99 percentiles for every endpoint in your application. Quickly identify which routes are fast and which need attention, with throughput and error-rate breakdowns.
- P50 / P95 / P99 latency percentiles
- Throughput + error rate per route
- Historical trend comparison
- Per-endpoint slow threshold override
Span waterfall view
Break down every request into its component operations. See exactly which database query, external API call, or middleware is adding latency, with precise timing for each span.
- Operation-level timing breakdown
- Visual waterfall timeline
- Pinpoint bottlenecks instantly
- Expand spans to see logs + attributes
Know what counts as slow
Set a slow-endpoint threshold globally or per route. Traceway uses it as the apdex anchor — a deliberate knob so /api/health and /api/checkout can be judged on their own terms.
Questions about performance
What is the Impact Score?
The Impact Score is Traceway's automatic prioritization system. It combines five service-level indicators — an inverted apdex variant, error rate floor, P99 latency floor, client error floor, and volume error floor — into a single 0-100 score for every endpoint.
It takes the max across all five, so if any single signal is degraded that endpoint surfaces immediately. You can adjust the slow endpoint threshold per endpoint to tune the apdex calculation for routes with different latency profiles.
For incident management, notification rules fire when the Impact Score or error rate crosses a threshold, and route alerts to Slack, GitHub Issues, email, Pushover, or custom webhooks.
What percentiles does Traceway track?
Traceway calculates P50 (median), P95, and P99 latency percentiles for every endpoint. This gives you a clear picture of both typical and worst-case response times for your application.
How does the waterfall view work?
The waterfall view shows every span (database query, external API call, middleware, etc.) within a single request as a timeline. You can see how long each operation took and where they overlap, making it easy to identify the slowest part of any request.
How are the apdex thresholds computed?
Traceway uses a slow-endpoint threshold as the apdex anchor. Requests under 750ms are Good, under 1500ms are Tolerable, and over 1500ms (or returning 5xx) are Bad. You can override the threshold globally or per-endpoint to account for routes that have different latency profiles — /api/health and /api/checkout rightly judged on their own terms.
Where do server metrics live now?
Server metrics (CPU, memory, goroutines, GC) moved to the dedicated Metrics page. Performance stays focused on latency, percentiles, and waterfall span analysis.