OpenTelemetry data stored in ClickHouse
SigNoz is an observability platform that takes in logs, metrics and traces over OpenTelemetry and stores all three in ClickHouse, a columnar database. From a service's latency chart, the UI leads to the traces behind it, then to the logs attached to those traces. Dashboards are built with a visual query builder, in PromQL or in plain ClickHouse SQL. The backend is written in Go and the frontend in TypeScript. The company behind it, SigNoz Inc., also sells a hosted version.
The comparison of open source observability platforms sets out what each one stores, on which database, and at what volume it starts to make sense.
Built-in APM and traces past 10,000 spans
- Standard ingestion: applications are instrumented with OpenTelemetry and send data over OTLP. Prometheus and a list of integrations also feed data in.
- APM out of the box: per-service latency, error rate, throughput, Apdex score, top endpoints, database calls and external API calls.
- Very large traces: a dedicated view handles traces above 10,000 spans, and the pricing page lists it in the free edition.
- MCP server: coding agents can query production telemetry, self-hosted instances included.
Five services to run, SAML only on paid plans
Five services to run, 4 GB of RAM to try it. The Docker and Linux installs start ClickHouse, ClickHouse Keeper, PostgreSQL for metadata, SigNoz's OpenTelemetry Collector and the backend. The docs ask for at least 4 GB of memory for "a small test installation" and point production sizing to a separate resource planning page.
Production sizing adds up to 56 cores and 152 GiB. The resource planning page lists three collectors at 4 cores and 16 GiB each, two ClickHouse shards at 16 cores and 32 GiB, three ZooKeeper nodes, PostgreSQL and the backend. It does not say what ingestion volume those numbers are meant for. OpenObserve ships as a single Rust binary that writes Parquet to S3 object storage. Traceway runs in one container on SQLite or DuckDB, and only needs ClickHouse and PostgreSQL in server mode.
MIT, except the enterprise code. The LICENSE file puts the whole repository under MIT apart from the ee/ and cmd/enterprise/ directories. Those fall under the SigNoz Enterprise license: running them in production requires a subscription for the right number of user seats.
Several features are paid only. The pricing page keeps anomaly detection, SAML, multiple ingestion keys and one-click AWS integrations out of the Community edition. Fine-grained RBAC is in beta on Enterprise only, and Noz, the built-in AI assistant, exists only on SigNoz Cloud. The README marks that cloud as the recommended option: the Teams plan starts at $49 a month, with logs and traces billed at $0.30 per GB and kept for 15 days. Enterprise starts at $4,000 a month, as a dedicated cloud, inside the customer's own cloud account (BYOC) or self-hosted with support.
Who the README says it's for: teams leaving a patchwork of tools. According to its README, SigNoz is often adopted by teams moving off a stack of single-purpose tools or a commercial platform with unpredictable pricing. Netdata comes at monitoring from the other end: an agent on every machine, collecting per-second metrics.
A minor release every week, still 0.x. Thirteen minor versions shipped between 0.133.0 on July 15 and 0.145.0 on October 5, 2026. The repository dates back to January 2021 and has never tagged a 1.0. The install.sh script and the repo's Docker Compose files are deprecated in favor of Foundry, a separate installer licensed under AGPL-3.0.