SigNoz Review: A Practical Open-Source Alternative to Datadog for APM and Observability

If your Datadog invoice has ever landed on a Friday afternoon and ruined your weekend, you have probably already Googled “open-source Datadog alternative.” SigNoz is almost certainly what the search engine handed you, and for good reason — it is one of the few tools in that category that actually gives you APM, distributed tracing, log management, and metrics in a single UI rather than asking you to stitch together Prometheus, Grafana, Loki, and Jaeger yourself. Whether it is the right call for your team depends on how much operational overhead you are willing to absorb, and this review will be specific about that trade-off.

What SigNoz actually is (and what it is built on)

SigNoz is an open-source observability platform built on two technology choices that define nearly everything about its performance characteristics and its maintenance burden: OpenTelemetry for instrumentation and ClickHouse for storage.

OpenTelemetry-native means SigNoz takes OTel’s semantic conventions seriously rather than treating them as an afterthought. If you instrument your services with the OTel SDK — which you should be doing regardless of where you ship telemetry — you get full fidelity in SigNoz without vendor-specific agents or proprietary SDKs. You also avoid the custom-metrics surcharges Datadog levies when you step outside its ecosystem. Datadog does support OTel, but you have to dig for it; SigNoz is built around it.

ClickHouse is a columnar database originally developed at Yandex and used in production by Uber and Cloudflare. It is very fast for the kinds of queries observability generates — high-cardinality filtering across billions of trace spans, full-text log search — but it is not a managed service you can forget about. It needs memory, storage planning, and occasional tuning as your data volume grows.

The GitHub repository (v0.125.1 as of May 2026) sits at around 27,000 stars with active weekly releases. The project is written primarily in TypeScript and Go, with the frontend and the backend query layer each being substantial codebases. This is not a side project — the release cadence is real, and the recent v0.121.1 release shipped an MCP server that lets AI assistants like Claude and Cursor query observability data through natural language, which is a sign the team is tracking where developer tooling is heading.

The three signals in one place

SigNoz’s core pitch is correlating logs, metrics, and traces without leaving the UI. In practice this means:

• Traces: Flamegraph and Gantt chart views for distributed traces. You can filter by service, span attributes, or duration, and jump from a slow span directly into the relevant logs.
• Logs: ClickHouse-backed log storage with a query builder, PromQL support, and raw ClickHouse SQL for power users. Full-text search is fast at scale because of the columnar storage.
• Metrics: Prometheus-compatible metric ingestion. Dashboards are customizable, and you can write PromQL alongside a point-and-click query builder.
• Exceptions: Language-agnostic exception tracking that shows stack traces grouped by type, with a count timeline.
• LLM observability: A newer addition — tracking token usage, latency, and prompt/response pairs for AI applications. This is genuinely useful if you are running LLM pipelines in production and want cost visibility alongside latency data.

Pricing: self-hosted vs. cloud vs. the hidden cost of ops

SigNoz has three meaningful tiers.

Community Edition is self-hosted, MIT-licensed, no data caps, no license fee. You pay for your own infrastructure — compute, storage, network egress — and you operate ClickHouse yourself.

Teams (Cloud) starts at $49/month, which buys you a credit toward usage. After that credit, you pay $0.30/GB for logs and traces ingested, and $0.10 per million metric samples. There is no per-host or per-container pricing, which immediately makes the bill easier to reason about than Datadog’s layered per-host model. Regional options include US, EU, and India. A 30-day free trial covers all features.

Enterprise starts around $4,000/month for cloud, with custom pricing for managed self-hosted. If you are comparing to a large Datadog contract, this tier is worth benchmarking directly.

The comparison that SigNoz’s own site makes — and that third-party cost analysis sites like CubeAPM corroborate — is that for a mid-size team with several dozen APM hosts and a few terabytes of logs per month, SigNoz Cloud can be significantly cheaper than Datadog. The exact multiplier depends on your specific usage, so treat any “X times cheaper” claim as a prompt to run your own numbers, not as a guarantee.

Where the self-hosted pricing calculation gets complicated is engineering time. Running ClickHouse at production scale is non-trivial. You need to size nodes correctly for your ingestion rate, manage storage growth, handle upgrades, and be on call when the monitoring stack itself has a problem. Cost estimates from independent sources suggest this can add up to tens of thousands of dollars per year in engineering labor for a mid-size organization before you count the cloud compute. If your team has dedicated DevOps or SRE capacity, this may be fine. If you are a small team where everyone is also shipping product features, operating ClickHouse is a real second job.

Where SigNoz works well, and where it does not

SigNoz is a strong choice when:

• Your team is already instrumenting with OpenTelemetry, or willing to commit to it. The platform’s value compounds if you are using OTel semantic conventions consistently.
• You have the engineering bandwidth to operate ClickHouse, or you are willing to pay for the Cloud tier.
• You want to avoid Datadog’s per-host and custom-metrics billing surprises.
• You are building AI-enabled applications and want LLM observability in the same tool as your APM data.
• Data residency matters — the EU and India storage regions for Cloud are a genuine differentiator over some competitors.

SigNoz is harder to recommend when:

• Your team is small and already stretched. The self-hosted path requires real commitment; the Cloud path is easier, but you are back to paying a vendor.
• You primarily need infrastructure monitoring (host metrics, container orchestration dashboards) rather than application-level APM. Datadog’s infra monitoring has more depth in this specific area.
• You need turnkey integrations with many non-standard data sources. Datadog’s integrations catalog is enormous; SigNoz is more focused and relies on OTel for coverage.
• Your organization is used to Datadog’s UI for log exploration and dashboarding. The SigNoz interface is functional and has improved substantially, but some users find the daily log search workflow requires more steps than they are used to.

The UI reality

The SigNoz frontend has improved meaningfully over the past year, but it is fair to say it does not yet match the polish of Datadog or Grafana Cloud for common daily-driver workflows. The query builder is capable, and PromQL plus raw ClickHouse SQL give power users a genuine escape hatch. But navigating between services, drilling into traces, and correlating across signals sometimes requires more clicks than you expect. This is not a dealbreaker — it is the kind of thing you adapt to — but set expectations accordingly if you are evaluating it against a tool your team already knows well.

Is it worth switching?

The answer depends almost entirely on your bill and your ops capacity. If you are paying a Datadog bill that has become a recurring complaint in engineering leadership meetings, SigNoz Cloud is worth a serious trial — the 30-day window with full feature access is enough to instrument a real service and stress-test the query experience with actual production data. If you are self-hosting and have a capable DevOps team, the Community Edition is genuinely production-ready and gives you full control.

What SigNoz is not is a magic cost reduction with no trade-offs. The trade-off is real: you get transparent, predictable pricing and open standards in exchange for either managing ClickHouse yourself or accepting the Cloud tier’s pricing model. That is a reasonable deal for many teams. It is worth being clear-eyed about it.