// the find
parseablehq/parseable
Parseable is an open source, unified infrastructure observability platform built in Rust on a data lake architecture. It tracks logs, metrics, traces, and events across apps, agents, and systems, reducing storage costs by up to 90% through columnar telemetry compression.
Parseable is a self-hosted log/metrics/traces store written in Rust that uses Parquet on object storage (S3, GCS, Azure Blob) as its persistence layer instead of a proprietary index format. It targets teams who want Grafana/Loki-style observability without the storage costs or the operational complexity of running Elasticsearch or ClickHouse. A single binary handles ingest, query, alerting, and dashboards.
The data lake architecture is the real differentiator: because data lands as Parquet on S3, you own your data outright and can query it with any Arrow-compatible tool outside Parseable itself. Native OTel ingest support means zero-config drops from any OTel Collector without custom exporters. The separation of ingestor and querier into distinct stateful sets in the Helm chart is the right call for production — you can scale write throughput without touching query capacity. Build provenance attestation via GitHub Actions shows the team takes supply-chain security seriously, which matters for infrastructure software.
PromQL support listed as a feature alongside SQL is a bit of a yellow flag — layering PromQL semantics onto a batch-oriented Parquet store means metric queries won't behave like Prometheus at high cardinality or fine resolution. The RBAC model appears to be homegrown rather than integrating with an existing policy engine like OPA, which tends to become a maintenance burden as permission requirements grow. There is no apparent WAL or durable staging guarantee before object storage flush, so an ingestor crash mid-batch likely drops recent events. The enterprise split (separate Helm charts with higher version numbers) suggests some features are being held back commercially, which means the open-source version may hit capability walls in production without it being obvious upfront.