Datadog alternatives: 6 open source observability platforms you can self-host

Most teams start looking for a Datadog alternative when the bill arrives. Datadog prices by host, by indexed log volume, by custom metric and by product, so costs grow in several directions at once and are hard to predict. The open source alternatives below let you keep metrics, logs and traces on infrastructure you control, and most of them speak OpenTelemetry, so you can instrument once and change the backend later. We read each project’s license and latest release on November 5, 2026.

Quick comparison

Platform	License	Covers	Storage	Best for
SigNoz	MIT (core), enterprise code in ee/	Metrics, logs, traces, alerts, dashboards	ClickHouse	The closest all-in-one Datadog replacement
Grafana (with Prometheus, Loki and Tempo)	AGPL-3.0	Dashboards over any data source	Whatever backends you choose	Teams that want to pick each piece
OpenObserve	AGPL-3.0, separate enterprise edition	Logs, metrics, traces, real user monitoring	Parquet files on disk or S3	Cheap long-term log storage
HyperDX (ClickStack)	MIT	Logs, traces, metrics, session replay	ClickHouse	Teams who already run ClickHouse
Coroot	Apache-2.0	Metrics, logs, traces, profiles via eBPF	ClickHouse and Prometheus	Kubernetes clusters with no instrumentation
Prometheus	Apache-2.0	Metrics and alerting	Its own time-series database	A metrics foundation for any of the above

SigNoz: the all-in-one replacement

SigNoz is the project that most deliberately copies the Datadog experience: one interface for application performance monitoring, logs, metrics, distributed traces, dashboards, alerts and exception tracking, with service pages that show latency, error rate and throughput per endpoint. It is built on OpenTelemetry from the start, and stores everything in ClickHouse, a columnar database that handles high-cardinality data well. You can query with a visual builder, PromQL or ClickHouse SQL.

The core is MIT licensed; code under the ee/ directory has a separate license, and SigNoz sells a hosted cloud and an enterprise self-hosted tier with extras such as fine-grained access control. Version 0.147.0 shipped on October 22, 2026, so the project moves fast. Self-hosting runs on Docker, Kubernetes or Linux. Choose SigNoz when you want to move off Datadog without assembling a stack yourself.

Grafana with Prometheus, Loki and Tempo: build your own stack

Grafana is the dashboard most engineers already know. On its own it stores nothing: it visualizes data from Prometheus for metrics, Loki for logs, Tempo for traces and dozens of other sources, including commercial databases. That flexibility is the reason to choose it. You can keep an existing Prometheus setup, add log search later and build dashboards that mix business data with system metrics.

Grafana is AGPL-3.0 (with some Apache-2.0 exceptions listed in its licensing file), and version 13.2.4 shipped in October 2026. The cost is operational: you run and upgrade several systems, each with its own storage and scaling rules. Choose this route when you have people who enjoy owning infrastructure, or when you already run part of it.

OpenObserve: low-cost storage for lots of logs

OpenObserve is written in Rust and stores data as Parquet files on local disk or in S3-compatible object storage, which is where its cost advantage comes from: object storage is far cheaper than the fast disks a search cluster needs. It handles logs, metrics, traces and real user monitoring with session replay, and you query logs with SQL and metrics with SQL or PromQL. It ships as a single binary, so a first installation takes minutes.

The open source edition is AGPL-3.0; a separately licensed enterprise edition adds features such as federated search across clusters. Version 1.1.2 shipped on October 20, 2026. Choose OpenObserve when log volume, not features, is what makes your current bill painful.

HyperDX: search and dashboards on ClickHouse

HyperDX is now part of ClickHouse’s ClickStack observability bundle. It gives you fast full-text and property search across logs and traces, dashboards, alerts, live tail and browser session replay, all on top of a ClickHouse database. It works with your existing ClickHouse schema, which matters if you already store events there. The all-in-one Docker image includes ClickHouse, the OpenTelemetry collector and MongoDB for metadata, and the README suggests at least 4 GB of RAM to test it.

The repository is MIT licensed. Choose HyperDX when your team already runs ClickHouse, or when debugging by searching logs and traces is more important to you than prebuilt APM pages.

Coroot: observability without instrumenting your code

Coroot takes a different approach. Its agent uses eBPF to gather metrics, logs, traces and continuous profiles from the Linux kernel, so you get a service map of your whole system without adding SDKs to every service. On top of that data it runs built-in inspections that explain common problems, tracks service level objectives, compares each Kubernetes deployment with the previous one, and estimates cloud cost per application.

Coroot is Apache-2.0, with a commercial enterprise edition sold separately, and version 1.28.1 shipped on October 21, 2026. eBPF needs a reasonably recent Linux kernel, so it fits Kubernetes and Linux servers, not serverless functions. Choose Coroot when you have many services and little time to instrument them.

Prometheus: the metrics foundation

Prometheus is not a complete Datadog replacement, but almost every option above can read or write its data. It scrapes metrics from your services and exporters, stores them in its own time-series database, evaluates alerting rules and hands alerts to Alertmanager. It is a graduated Cloud Native Computing Foundation project, Apache-2.0 licensed, and version 3.15.1 shipped on October 21, 2026.

Start with Prometheus if you mostly need infrastructure metrics and alerts, then add Grafana for dashboards. For long-term storage across many clusters, you will eventually add a backend such as Mimir or Thanos.

Worth knowing about

• Netdata has a GPL-3.0 monitoring agent that collects thousands of metrics per second with almost no setup. Its dashboard, however, is distributed under the Netdata Cloud UI License, which is not an open source license, so we do not list it in the directory.
• Uptrace is an AGPL-3.0 APM built on OpenTelemetry and ClickHouse. Its latest release in October 2026 was a release candidate for version 2.1, so check the stable version before you deploy.

How to choose

• You want one product that feels like Datadog: SigNoz.
• You already use Prometheus and Grafana dashboards: extend that stack with Loki and Tempo.
• Your problem is log volume and storage cost: OpenObserve.
• You already run ClickHouse: HyperDX.
• You run Kubernetes and cannot instrument every service: Coroot.

Whatever you choose, instrument with OpenTelemetry rather than a vendor SDK. Your telemetry pipeline then stays the same if you switch backends, and you can send data to two systems at once while you compare them. Need uptime checks as well? Uptime Kuma is in the directory. You can browse every project in the directory.