Oriel documentation

Oriel is a self-hosted, OpenTelemetry-native observability platform. It accepts OTLP traces, metrics, logs, and profiles; stores telemetry in ClickHouse; stores metadata in PostgreSQL; and uses Valkey for runtime cache, rate limits, series limits, ruleset fanout, and pub/sub.

Oriel overview showing request throughput, latency, active alerts, service health, and slow operations.

Start a local stack

Run Oriel with PostgreSQL, ClickHouse, and Valkey, then create the first workspace and send OTLP telemetry.

Start with the Compose quickstart.

Use the product

Work inside the Oriel scope model, query telemetry with OQL, inspect traces, logs, metrics, and profiles, and build alerts and dashboards.

Start with the scope model and Explore and OQL.

Operate a deployment

Choose a deployment path, configure storage, TLS, auth, migrations, readiness, backups, retention, and security controls.

Start with Choose an install path and Architecture.

First successful path

Run the local stack:

cd deploy/compose
docker compose -f local.yml up --build

When the stack is healthy, bootstrap the first user and workspace:

docker compose -f local.yml exec -e ORIEL_BOOTSTRAP_PASSWORD='change-me' oriel \
  oriel admin bootstrap --email you@example.com
docker compose -f local.yml exec oriel oriel admin create-workspace --slug demo

The local endpoints are:

Endpoint	URL
UI and API	`http://localhost:6743`
OTLP gRPC	`localhost:4317`
OTLP HTTP	`http://localhost:4318`

Next, create a project and environment in Settings, mint an ingest token, and send telemetry with an OpenTelemetry SDK or Collector. See Send telemetry.

Runtime shape

Oriel runs as one binary with four runtime roles:

Role	Purpose
`all`	Runs ingest, query/API/UI, and worker in one process. This is the default documented path.
`ingest`	Runs OTLP gRPC and HTTP receivers.
`query`	Runs the UI, management API, query API, health checks, and metrics endpoint.
`worker`	Evaluates alert rules and rolls up service-map edges.

The binary requires PostgreSQL metadata storage, ClickHouse telemetry storage, and Valkey runtime cache/pub-sub. Postgres is the durable metadata source of truth; Valkey holds rebuildable runtime state.