Overview
OpenObserve (often called O2) is an open-source observability platform written in Rust. Instead of Elasticsearch-style indexes, it writes logs, traces and metrics as compressed Parquet files to S3-compatible object storage, which it says cuts storage cost sharply for large log volumes. It runs as a single binary for small setups or as a cluster for scale.
It covers logs with full-text and SQL search, metrics with SQL or PromQL, OpenTelemetry traces, real user monitoring with session replay, error tracking, dashboards, alerts, pipelines and LLM observability, so it can replace an ELK or Loki stack plus a tracing backend.
The open-source edition is free, and even the Self-Hosted Enterprise edition with SSO and RBAC is free below 50 GB a day. OpenObserve Cloud charges $0.50 per GB ingested on an annual commitment plus $0.01 per GB queried, with no per-seat fees.
Pricing and plans
checked 28 Sept 2026Free tier · no card
Open Source edition is free forever to self-host; the Self-Hosted Enterprise edition is free up to 50 GB ingested per day. The cloud has a 14-day trial only.
| Plan | Price | What you get |
|---|---|---|
| Open Source | $0 | AGPL-3.0, self-hosted, single binary or HA cluster |
| Self-Hosted Enterprise | $0 up to 50 GB/day | SSO, RBAC, federated search, audit trail, redaction; above 50 GB/day contact sales |
| Cloud Professional | $0.50/GB ingested + $0.01/GB queried | Logs, metrics, traces, RUM, replay, errors; metrics kept 15 months, other data 30 days |
| Enterprise | Custom | Bring your own bucket, volume discounts, SLAs, BYOC |
What it does
- Parquet on object storage
- Columnar storage on S3, GCS or Azure Blob keeps long retention cheap.
- SQL and PromQL
- Query logs and traces with SQL and metrics with SQL or PromQL.
- OpenTelemetry native
- Accepts OTLP logs, metrics and traces, plus Fluent Bit, Vector and syslog.
- RUM and session replay
- Core Web Vitals, frontend errors and session replay.
- Alerts and incidents
- Scheduled, real-time and anomaly alerts grouped into incidents.
- Single binary
- Starts in minutes as one binary; scales to an HA cluster.
Strengths and limitations
Strengths
- Free self-hosted editions, including enterprise features below 50 GB a day.
- Low storage cost for large log volumes using object storage.
- No per-seat or per-host fees on the cloud.
- Covers logs, metrics, traces and frontend monitoring in one tool.
Limitations
- Cloud has only a 14-day trial, no permanent free tier.
- The advertised cloud ingest price assumes an annual commitment.
- AGPL licence may need review for companies that modify and offer it as a service.
- No uptime checks or status pages.
Who it suits
Good for
- Teams with large log volumes replacing Elasticsearch or Splunk
- Self-hosters wanting one backend for logs, metrics and traces
- Security and platform teams keeping long retention
Look elsewhere if
- Developers wanting simple hosted error tracking with a free tier
- Uptime monitoring
Alternatives to OpenObserve
| Tool | From | Free option | Stages |
|---|---|---|---|
| Amazon CloudWatch AWS-native metrics, logs, alarms and tracing | Usage-based | Free tier | |
| Checkly Synthetic and uptime monitoring as code | $24/mo | Free tier | |
| Grafana Cloud Hosted metrics, logs, traces and dashboards | $19/mo | Free tier | |
| New Relic Full-stack observability platform | $10/mo | Free tier | |
| Rollbar Error tracking and session replay | $15.83/mo | Free tier | |
| Sentry Error tracking and application monitoring | $26/mo | Free tier |
Questions
Is OpenObserve free?
The open-source edition is free to self-host, and the Self-Hosted Enterprise edition is free up to 50 GB of ingestion a day.
How much does OpenObserve Cloud cost?
$0.50 per GB ingested with an annual commitment plus $0.01 per GB queried, with 30 days of log retention and 15 months of metrics included.
Can OpenObserve replace Elasticsearch for logs?
It is designed to, storing logs as Parquet on object storage and querying them with SQL.
Does OpenObserve support OpenTelemetry?
Yes, it ingests OTLP logs, metrics and traces natively.
Sources read for this page