Follow every request across your services. Find bottlenecks in seconds with waterfall visualizations, service maps, and latency percentiles.
# Trace waterfall — one request, many services GET /api/orders/42 247ms ├─ api-gateway 3ms ├─ auth-service 12ms │ └─ redis.get 1ms ├─ order-service 180ms │ ├─ postgres.query 45ms │ └─ inventory-service 120ms ⚠ slow │ └─ postgres.query 115ms ⚠ N+1 └─ payment-service 48ms └─ stripe.charge 42ms
See the full picture of how requests flow through your system.
See every span in a request's journey. Drill down to database queries, HTTP calls, and queue processing times.
Auto-generated dependency graph of your services. See request rates, error rates, and latency between every pair.
P50, P90, P95, P99 — track latency distribution per endpoint. Spot regressions before they become incidents.
Accept traces via OTLP (OpenTelemetry), Zipkin, and Jaeger protocols. No vendor lock-in.
Capture exceptions with full stack traces linked to the request trace. Group errors by root cause.
Search traces by service, operation, duration, tags, or error status. Find the needle in millions of spans.
Instrument once, understand everything.
Add OpenTelemetry SDKs to your services (auto-instrumentation available for most frameworks).
Spans stream to Koder Trace via OTLP. We index, correlate, and build the service graph automatically.
Explore traces in the waterfall view. Identify slow services, N+1 queries, and error hotspots.
Set latency and error rate thresholds per service or endpoint. Get notified via Koder Notify.
Send traces using the OpenTelemetry protocol. Works with every OTEL SDK and auto-instrumentation library.
# Configure OTEL SDK to send to Koder Trace export OTEL_EXPORTER_OTLP_ENDPOINT=\ "https://trace.koder.dev" export OTEL_EXPORTER_OTLP_HEADERS=\ "Authorization=Bearer $TOKEN" export OTEL_SERVICE_NAME="my-service" # Auto-instrument a Node.js app npx @opentelemetry/auto-instrumentations-node \ node app.js # Or use the Python SDK opentelemetry-instrument \ --traces_exporter otlp \ python app.py
Pay for what you trace.
OTLP (OpenTelemetry) over gRPC and HTTP, Zipkin v2, and Jaeger Thrift. We recommend OTLP for new deployments.
OpenTelemetry provides auto-instrumentation for Java, Python, Node.js, .NET, Go, Ruby, PHP, and more. Check the OTEL registry.
Head-based sampling is done in your SDK. Our Pro plan adds tail-based sampling: we keep 100% of errors and slow traces, sample the rest.
Yes. Inject trace IDs into your logs and Koder Trace links them automatically. Works with Koder KDB log model.
See every request. Find every bottleneck.