Self-hosted observability

Your logs, metrics, and traces — in one honest pane.

backwork unifies logs, metrics, distributed traces, requests, incidents and alerts into a single self-hosted dashboard. Multi-tenant, OpenTelemetry-native, and engineered to never show you green when your pipeline is down.

Open source One Docker command No credit card

backwork.dev/overviewLive
REQ/S
1.2k
ERRORS
0.3%
P95
142ms
12:04:21infoGET /checkout 200 · 142ms
12:04:21warnretry upstream payments (1/3)
12:04:20errPOST /charge 500 · timeout

Built on the open standards you already trust

OpenTelemetryLokiVictoriaMetricsJaegerPrometheusVectorGrafana-compatibleDocker
The problem

Observability shouldn't cost more than the thing you're observing.

Per-GB pricing punishes you for having traffic. Five separate tools never agree. And too many dashboards quietly show green when the data behind them stopped flowing. backwork is the opposite: one stack you run yourself, one pane that tells the truth.

One tool, the whole picture

Everything you need to watch a system.

Logs, metrics and traces share one timeline, one search and one set of alerts — no tab-juggling.

Logs

Every line, searchable.

Full-text and LogQL search across every container, with level, service and time filters and a live volume chart.

12:04:21INFOGET /checkout 200 · 142ms
12:04:21WARNretry upstream payments (1/3)
12:04:20ERRPOST /charge 500 · timeout
12:04:19INFOcache hit user:8842
12:04:18INFOGET /api/orders 200

Metrics

Host to container to app.

node-exporter and cAdvisor metrics plus RED rates from real traces — PromQL underneath.

checkout-api
auth
payments
postgres

Traces

Follow every request.

OpenTelemetry-native spans with a waterfall and a dependency map.

GET/checkout200POST/charge500GET/orders200GET/cart304

Requests

RED, at a glance.

A flat HTTP feed that jumps straight to the trace.

Error spike critical
checkout-api · 5xx rate ↑ · detected 2m ago

Incidents

It tells you first.

Crash, OOM, restart-loop and error-spike detection from real signals.

Alerts

Routed where you'll see them.

Rules to Slack, Discord, webhook, email or SMS — with a no-data state so an outage never fakes “all clear.”

SlackDiscordWebhookEmailSMSno data → suppressed
Why backwork

Built different, on purpose.

Four decisions that set it apart from the per-gigabyte status quo.

01

Own your data

Self-hosted on your own box with one Docker command. Your telemetry never leaves your infrastructure — no per-GB bill, no third-party retention.

02

Multi-tenant, SaaS-ready

Organizations, teams, roles and per-project ingest tokens. Every tenant's logs, metrics and traces are isolated to the byte.

03

Honest by design

No demo data, no fake green. A data-source outage reads as “no data,” never “healthy” — so you trust the dashboard when it matters.

04

One command to ship

curl | sh drops an agent on any host for logs and metrics. Point any OpenTelemetry SDK at one endpoint for traces. That's it.

How it works

Live in three steps.

1

Create a project

Spin up an organization and get a per-project ingest token, scoped to your tenant.

# in the dashboard
Projects → New project
2

Ship your telemetry

One command for logs & metrics on any host. One env var for OpenTelemetry traces.

curl -fsSL backwork.dev/install.sh | sh
3

Watch it live

A unified dashboard that auto-refreshes, detects incidents and routes alerts where you'll see them.

OTEL_EXPORTER_OTLP_ENDPOINT=/otlp
Honest by design

It never shows you green when your pipeline is down.

Most tools coerce a missing metric to zero and paint it healthy. backwork distinguishes no data from all clear — across overview, incidents and alerts — so an outage in the telemetry path is visible, not hidden.

Typical tool — collector offline 6m
checkout-apiOperational
metric coerced to 0 → looks healthy
backwork — same outage
checkout-apiNo data
source unreachable → alert suppressed, not “OK”
Pricing

Free to self-host. Pay only for convenience.

Run the whole thing yourself at no cost, or let us host and scale it for you.

Self-host
Free
your hardware, open source
Cloud
Free
managed, free in beta
Enterprise
Let's talk
SSO, SLA & data residency
FAQ

Questions, answered.

Everything worth knowing before you deploy.

Is backwork open source?+

Yes. The full stack lives in a public repo — clone it and run it yourself for free, forever.

Is it really self-hosted?+

Yes. backwork runs on your own host via Docker Compose, and your telemetry stays on your infrastructure.

Does it work with OpenTelemetry?+

Natively. Point any OTLP exporter at a single endpoint and your traces, requests and APM metrics appear automatically.

What's it built on?+

Battle-tested open standards — Loki for logs, VictoriaMetrics for metrics, Jaeger for traces and Vector for collection — unified behind one dashboard.

Can my whole team use it?+

Yes. Create organizations and teams, assign roles, and issue per-project ingest tokens that keep each tenant's data isolated.

How is it different from Grafana or Datadog?+

One integrated pane instead of a wall of dashboards, self-hosted economics instead of per-GB pricing, and states you can trust — it never shows green when your pipeline is down.

Stop renting your observability.

Deploy backwork on your own host in minutes and own your telemetry end to end.