Best open-source alternatives to Datadog

Langfuse is an open source platform for LLM application engineering aimed at teams building AI products. It provides a centralized place to instrument applications, inspect traces, monitor LLM calls, and debug complex user sessions, while also supporting prompt management, evaluation workflows, dataset-based benchmarking, and playground-driven iteration. The project is designed for both managed cloud use and self-hosted deployments. The README highlights quick local setup with Docker Compose, production deployment on Kubernetes via Helm, and integration with popular tooling such as OpenAI, LangChain, LlamaIndex, and custom SDK-based workflows. It also emphasizes an API-first approach with OpenAPI specs, Postman collections, and typed SDKs for Python and JS/TS.

SigNoz is a self-hostable observability platform aimed at teams that need a single place to inspect application logs, metrics, and traces. It positions itself as an open-source alternative to commercial monitoring products such as Datadog and New Relic, and it emphasizes OpenTelemetry-based instrumentation so users can collect telemetry with a vendor-neutral standard. The project provides several production-focused capabilities, including APM charts, centralized log search, distributed tracing, dashboards, alerting, exception monitoring, and LLM observability. The README indicates that users can deploy it with Docker or on Kubernetes via Helm, and that they can choose between self-hosted usage and SigNoz Cloud, depending on how much maintenance they want to manage.

Dozzle is a web-based log viewer for Docker environments. It is intended for people who want to inspect container logs live without storing them, and it emphasizes a small footprint and quick access to running services. The project supports a range of deployment styles centered on Docker, including standalone containers, Docker Compose, Swarm mode, and an agent mode for monitoring multiple Docker hosts. The README also describes authentication support, configuration through environment variables and CLI flags, and development setup for contributors working on the Go backend and Vue frontend.

Opik is an open-source platform from Comet for teams building LLM-powered products. It is aimed at developers who need to trace model calls, evaluate outputs, monitor production behavior, and improve prompts or agent workflows across the full lifecycle of an application. The project combines observability, evaluation, and optimization features in a single system. It supports trace logging, feedback annotation, datasets, experiments, LLM-as-a-judge metrics, dashboards, and guardrails, and it can be run either through Comet’s cloud offering or self-hosted with Docker Compose or Kubernetes/Helm for local and scalable deployments.

Beszel is a lightweight server monitoring platform aimed at users who want a simple way to view host and container metrics through a web interface. It presents a dashboard for connected systems, tracks historical performance data, and includes configurable alerts and notifications. The project is organized around a central hub and per-host agents. The hub is a web application built on PocketBase, while agents run on monitored systems and send metrics back to the hub. The README indicates support for multi-user access, OAuth/OIDC authentication, automatic backups, and API access, making it suitable for shared or self-hosted monitoring setups.

Manifest is an OpenClaw plugin for routing LLM requests to the most cost-effective model. It is aimed at users who want to reduce model usage costs while keeping request handling and observability integrated into their existing OpenClaw setup. The project operates in both cloud and local modes. In cloud mode, users sign up for a hosted service and provide an API key; in local mode, the dashboard runs on the user’s machine and telemetry remains local. Manifest emphasizes privacy by keeping prompt content out of telemetry, and it supports a large set of model providers including OpenAI, Anthropic, Google Gemini, Ollama, and OpenRouter.

HyperDX is an observability application that helps engineers investigate production issues by searching and visualizing telemetry data stored in ClickHouse. It is presented as a core component of ClickStack and is positioned as a fast, intuitive alternative for working with logs and traces on top of an existing ClickHouse cluster. The project is aimed at teams that need a unified way to work with logs, metrics, session replays, traces, dashboards, alerts, and APM-style request and database monitoring. It can be deployed as an all-in-one container or as part of a broader ClickStack setup, and it supports OpenTelemetry-based instrumentation through its collector endpoint and SDK integrations for common languages and platforms.

OpenSearch is an open-source search and observability suite designed to help users organize and analyze unstructured data at scale. It is positioned as an enterprise-grade platform, suggesting it is intended for production environments and larger deployments rather than lightweight local use. The README primarily serves as a project entry point, linking to downloads, documentation, community support, and contributor resources. It does not describe detailed installation steps or runtime architecture, but it does identify the project as Apache 2.0 licensed software maintained by the OpenSearch community.

Agenta is an open-source platform aimed at engineering and product teams that need to develop, test, and operate production LLM applications. It combines prompt management, evaluation workflows, and observability tools in one self-hostable product, with a hosted cloud option also available. The project is designed to help teams collaborate on prompts and configurations, run structured evaluations with human or automated feedback, and monitor runtime behavior such as cost, latency, and traces. It also supports a wide range of models and integrations, and exposes both UI-based and programmatic evaluation workflows for different kinds of users.

HOMER 10 is a carrier-grade observability platform for VoIP and RTC monitoring. It is designed to ingest signaling, real-time communication events, logs, metrics, and traces, then provide instant search, correlation, and drill-down analysis for troubleshooting and monitoring deployments at scale. The project targets enterprises, voice network operators, VoIP service providers, and traffic carriers that need to inspect SIP and RTP/RTCP traffic across heterogeneous systems. HOMER 10 is positioned as a headless observability solution built around standard protocols and visualization tools such as Grafana, and it emphasizes compatibility with HEP/EEP-based capture agents and external observability backends.

Gatus is a developer-oriented health dashboard designed to monitor services through active health checks rather than passive traffic-based metrics. It is aimed at operators who want to detect problems before users are affected, especially in environments where critical endpoints need to be probed directly. The project supports checks over HTTP, ICMP, TCP, and DNS, and can evaluate responses using conditions such as status code, response time, certificate expiration, body content, and IP address. It also integrates alerting through providers like Slack, Teams, PagerDuty, Discord, Twilio, and others, making it suitable for production monitoring workflows. Gatus appears built for self-hosted use in infrastructure such as Kubernetes clusters, with deployment options including Docker and Helm. The README also documents an API section, metrics, badges, TLS, authentication, and remote instances, indicating that it can be configured as a flexible monitoring backend for teams and services.

Kuvasz is an open-source, self-hosted monitoring platform for tracking website and service uptime as well as SSL certificate validity. It is aimed at users who want a modern interface for managing checks, viewing status information, and presenting service health through public or private status pages. The project exposes a REST API for programmatic management and integrates with common notification channels such as email, Discord, Slack, Telegram, and PagerDuty. It also supports metrics export to OpenTelemetry and Prometheus, making it suitable for teams that want to fold uptime data into an existing observability stack. The README emphasizes that it is free and open source, and points users to a deployment guide for setup.

Bencher is a continuous benchmarking platform for teams that want to detect performance regressions before they reach production. It combines a CLI, API server, and web console so benchmark results can be collected from local machines or CI jobs, then tracked and analyzed over time. The project is aimed at developers and engineering teams using benchmark suites in languages such as Rust, Python, JavaScript, Go, C++, Java, and others. Its workflow centers on running existing benchmark harnesses, storing measurements, and using graphs, thresholds, and analytics to spot changes across branches, testbeds, and measures. Bencher can be self-hosted or used through the hosted Bencher Cloud service.

Rootprint is a self-hosted log management platform aimed at engineering teams that need a focused interface for searching and exploring logs and traces. It combines object-storage-backed index search with OpenTelemetry ingestion, so users can send logs and traces from common observability pipelines and query them from a single web UI. The project includes team-oriented access controls, admin tools, saved views, field filters, share links, and export capabilities. It also exposes an API and ships with a Quickwit-powered backend, making it suitable for organizations that want to operate their own log search stack without sending telemetry to a hosted SaaS.

Nimbus is a self-hosted homelab dashboard aimed at users who want a centralized way to monitor and organize their services. It combines service health monitoring with a configurable, visually polished interface, offering features such as real-time status checks, uptime graphs, theming, and drag-and-drop service tiles. The project is designed for multi-user homelabs and includes local accounts, OAuth2 sign-in, RBAC, and an admin panel for managing users. It can be deployed quickly with Docker Compose and relies on PostgreSQL for persistence, while also exposing Prometheus metrics for integration with monitoring stacks. A managed cloud hosting option is mentioned, but the core project is intended for self-hosted use.

Elasticsearch is a distributed search and analytics engine designed for large-scale production workloads. It is positioned as the foundation of Elastic’s Open Stack and is aimed at users who need fast, relevant search across massive datasets, along with vector search and machine-learning-oriented use cases such as retrieval augmented generation. The README describes several ways to run it, including a managed Elastic Cloud deployment, local Docker-based setup for development, and building from source with Gradle. It also explains how to interact with the system through REST APIs, language clients, curl, and Kibana’s Dev Tools console. Common uses highlighted in the README include logs, metrics, application performance monitoring, security logs, and general full-text search.