Beacon Trace Start free trial

API observability · Seattle, Washington

Never miss a signal.

Beacon Trace measures API latency hop by hop, groups every error into one issue, and routes the alert to the engineer on call — so an incident arrives as a page, not a surprise.

Start free trial Talk to an engineer

SDKs for Node · Python · Go

beacon-trace trace stream
routePOST /v1/orders
slowest spanpayments-service
routed toplatform on-call
trace stream · p99 per route spike detected → alert routed

01 — Instrumentation

One import per service.

Install the SDK for your runtime, name the service, and traces start arriving with your traffic. Nothing else to deploy and no dashboard to build before you can read a request.

  1. 01

    Install the SDK

    One package per runtime — Node, Python or Go — imported at your service entry point.

  2. 02

    Name the service

    Set the service name once. Every request through it starts producing spans.

  3. 03

    Read the first trace

    Latency, errors and throughput arrive with your traffic, attributed to route and span.

A developer's hands on a keyboard in a dark room, green monitor light across the desk
first service instrumented Node · Python · Go
Nodenpm package
Pythonpip package
Gogo module

02 — Latency & distributed tracing

Find the hop that burns.

Every request becomes a trace you can read span by span. The slow hop is not buried at the bottom of a log file — it is the highlighted bar on the screen, with the route and the service that owns it.

  • Latency tracingEvery route and every percentile, in one view per service.
  • Distributed tracingFollow one request across services and queues to the span that slowedit down.
  • Live, not averagedSee which service is slow right now, not what last week's mean looked like.
Rows of server racks in a dark data centre, small green status lights glowing
every instrumented service, one trace view span-level latency

03 — Errors & on-call

Every error grouped. Every page routed.

Error tracking and alerting sit on the same trace data, so an issue and the alert it raised point at the same span — and the engineer who gets paged can see it before opening a terminal.

  • Grouped by fingerprint

    Occurrences that share a fingerprint collapse into one issue, so a thousand sparks read as one flame instead of a thousand notifications.

  • Linked to the failing span

    Every issue keeps the trace and the exact span that threw, along with the request that carried it.

  • Triaged where it happened

    Resolve, mute or assign an issue to a teammate without leaving the trace view.

An engineer at night seen from behind, silhouetted against monitors showing green line graphs
incident triage

Alerting and on-call routing

  1. 01

    Detect

    Latency or error rate crosses the threshold set for that service.

  2. 02

    Match

    Rules match on service, route or severity — and silence what should not wake anyone.

  3. 03

    Route

    The alert goes to the rotation that owns the service, with the trace attached.

  4. 04

    Escalate

    No acknowledgement? It moves to the next responder until someone takes it.

04 — Service dashboards

The whole system on one screen.

Latency, errors and throughput for every service side by side, so the first question of an incident — what changed? — is already answered on the wall.

See it on your stack
An operations engineer watching a wall of monitors in a dark operations room
operations view service dashboards

latency

Percentiles per route, refreshed as the traffic moves.

errors

New groups surface the moment they start sparking.

throughput

Request volume across every instrumented service.

Get started

Light your beacon.

Start a free trial and point your first service at Beacon Trace, or talk to an engineer about how your stack is wired.

Start free trial Talk to an engineer