# Performance Benchmarks

> Head-to-head benchmarks of OpenObserve against Elasticsearch, ClickHouse, Prometheus, and Grafana Mimir on identical hardware: ingestion throughput, query latency, storage footprint, and cost.

Source: https://openobserve.ai/benchmarks/

---

Real-world performance metrics from head-to-head deployments on identical hardware.

## Benchmark reports

### OpenObserve vs ClickHouse: One Billion Log Records

One billion log records, 2 TiB of NDJSON, identical bytes written to ClickHouse and two OpenObserve instances at once. 19 queries, full-text indexes on both sides, 3.4x to 4.6x faster on a third less disk.

[Read the benchmark](/blog/openobserve-vs-clickhouse-one-billion-logs-benchmark/)

### OpenObserve vs Prometheus and Mimir

About 2.2 billion samples pushed through one OpenTelemetry Collector into Prometheus, Grafana Mimir, and OpenObserve at the same time, then queried with identical PromQL.

[Read the benchmark](/blog/openobserve-vs-prometheus-mimir-metrics-benchmark/)

### OpenObserve vs Elasticsearch

1.1 TB of Kubernetes-format log data streamed to both Elasticsearch and OpenObserve on identical AWS hardware - storage, CPU, RAM, and query performance compared.

[Read the benchmark](/blog/elasticsearch-openobserve-benchmarking/)

## Benchmark FAQ

How the tests are run and what the numbers mean for your stack.

### How are OpenObserve benchmarks run?

Each benchmark streams the same dataset to both systems on identical cloud hardware, then measures ingestion throughput, query latency across representative queries, CPU and memory during ingest and query, and storage on disk or object storage. The full methodology, dataset, and configuration are published with every report so the results can be reproduced.

### Why is OpenObserve's storage footprint so much smaller than Elasticsearch?

OpenObserve stores data as Apache Parquet on object storage instead of building inverted indexes. Columnar compression on log data yields up to 140x lower storage cost for the same volume, which is the single largest cost lever for logs-heavy workloads.

### Do you have benchmarks against Datadog, Grafana, or Dash0?

Not yet as published head-to-head performance runs; SaaS platforms do not expose comparable infrastructure to measure. For those vendors we publish cost and feature comparisons instead. See OpenObserve vs Datadog, OpenObserve vs Grafana, and OpenObserve vs Dash0, or use the Datadog cost calculator with your own volumes.

### Can I reproduce these results on my own data?

Yes. OpenObserve is open source and runs as a single binary, so you can replay your own logs, metrics, or traces into it in minutes. The demo team can also run a proof of concept against your real ingest volume.

## See the numbers on your own data

OpenObserve unifies logs, metrics, traces, and frontend monitoring in one open-source platform - at a fraction of the cost of legacy tools.

- Get Started for Free
- [Book a Demo](/demo/)