See every API call before your users do
Traceloop traces every request end-to-end — latency, errors, payloads, and dependencies — so you catch the slow endpoint and the 500 spike in seconds, not after the support ticket. Instrument in one line, ship with confidence.
Free tier forever · No credit card · OpenTelemetry-native · Self-host option
Trusted by backend teams shipping at scale
Observability that speaks HTTP
Metrics tell you something broke. Traceloop tells you which endpoint, which dependency, which payload, and which deploy — in one trace view built for the people who own the API.
One-line instrumentation
Drop in our SDK for Node, Python, Go, Ruby, or Java — or point your existing OpenTelemetry exporter at us. Full request tracing in under five minutes, no code rewrite.
End-to-end distributed traces
Follow a single request across every service, queue, and database hop. See exactly where the 800ms went and which downstream call timed out — waterfall view included.
Endpoint-level latency & errors
P50/P95/P99 and error rates broken out per route and method, automatically. Sort by slowest, spot regressions against last deploy, and drill straight into the offending traces.
Alerts that page on symptoms
Set SLO-based alerts on latency, error rate, or throughput per endpoint. Route to Slack, PagerDuty, or webhook — and land in the exact trace that tripped it.
Payload & header capture
Inspect request and response bodies, headers, and status codes on any trace — with automatic PII redaction — so reproducing a bug doesn't mean guessing at inputs.
Your data, your rules
SOC 2 Type II, per-field redaction, configurable retention, and a self-hosted option for teams that can't send traces off-prem. Sampling controls keep costs predictable.
From black box to full visibility in three steps
Instrument in one line
Install the Traceloop SDK or redirect your OpenTelemetry pipeline. Traces start flowing immediately — no sidecar to babysit, no agent to tune.
Watch traces stream in
Every request appears in a live waterfall: services, DB queries, external calls, timings, and status. Filter by route, status, or duration to find the needle fast.
Alert, debug, and ship
Set SLO alerts, get paged on real symptoms, and jump from alert to root-cause trace in one click. Fix the slow query before it becomes an incident.
“We had metrics and logs but still spent hours reconstructing what a request actually did. With Traceloop I open one trace and the slow downstream call is right there. Our mean-time-to-resolve dropped from ~40 minutes to under 5.”
Pricing that scales with your traffic
Priced per span ingested. Every plan includes tracing, alerting, and the full SDK library. No seat limits, ever.
Free
For side projects and small services.
- 5M spans / month
- 3-day retention
- All SDKs & OTel ingest
- Endpoint latency & error views
- Community support
Pro
For teams running production APIs.
- 100M spans / month included
- 30-day retention
- SLO alerts (Slack, PagerDuty, webhook)
- Payload capture & PII redaction
- Deploy markers & regression tracking
- Priority support
Enterprise
For high-scale & regulated teams.
- Everything in Pro
- Volume span pricing
- Self-hosted / VPC deployment
- SSO, RBAC & audit logs
- Custom retention & sampling
- SLA & dedicated support
Questions engineers ask us
How is this different from my metrics dashboard?+
Metrics tell you a graph moved. Traceloop tells you why: the exact request, the service hop that stalled, the query that ran slow, and the payload that triggered it. It's built for debugging individual requests, not just watching aggregates.
Do I have to rip out my current setup?+
No. Traceloop is OpenTelemetry-native, so if you already emit OTel spans you just point the exporter at us. No SDK lock-in, and you can run it alongside your existing metrics and logging tools.
What languages and frameworks are supported?+
First-class SDKs for Node.js, Python, Go, Ruby, Java, and .NET, with auto-instrumentation for the common web frameworks, HTTP clients, and databases. Anything that speaks OpenTelemetry works out of the box.
How do you keep sensitive data out of traces?+
Payload and header capture runs through configurable, per-field redaction before anything is stored, and you control sampling and retention. Regulated teams can self-host in their own VPC so traces never leave their network. We're SOC 2 Type II certified.
Will this blow up my bill at scale?+
Pricing is per span ingested with generous included volumes, and head/tail sampling controls let you keep the interesting traces while dropping the noise. You set a monthly cap, so costs stay predictable even during a traffic spike.
Stop debugging blind
Start free with 5M spans a month — no card required. Instrument your first service and watch the traces roll in within minutes.
Start free