Run it locally
Bring up ClickHouse, the schema, PulsusDB, and an OpenTelemetry Collector with Docker Compose. Open the quickstart →
The observability data platform
One open source platform for logs, metrics, traces and profiles. Ingest with OpenTelemetry, keep PromQL, LogQL and TraceQL, and let PulsusDB turn them into SQL for ClickHouse.
PulsusDB is in active development. Interfaces can change and production readiness is still being validated.
Start with the task in front of you
Section titled “Start with the task in front of you”Deploy to Kubernetes
Install the Helm chart in single-process or split writer/reader mode. Read the Helm guide →
Send telemetry
Configure an OpenTelemetry Collector to send logs, metrics, traces, and profiles. Configure the collector →
Connect Grafana
Use the Prometheus API and optional compatibility endpoints with existing dashboards. Configure datasources →
Built from the query backwards
Section titled “Built from the query backwards”PulsusDB gives each signal its own ClickHouse schema, ordering strategy, indexes, and read path. Its stateless Rust processes can run together as one binary or split into independently scalable writer and reader tiers.
OpenTelemetry-native
OTLP is the primary ingest path. The Collector keeps instrumentation concerns outside the database.
Familiar query languages
Keep PromQL, LogQL, and TraceQL. PulsusDB parses and plans them against ClickHouse.
Read-path first
Purpose-built tables, skip indexes, shard-aware lookups, and exact rollup selection reduce unnecessary reads.
One binary, three modes
Start with all, then scale ingestion and queries separately with writer and reader modes.