Telegraf is a Go-based, MIT-licensed open-source agent that collects, processes, aggregates, and writes metrics, logs, and other arbitrary data, and it is aimed at engineers who run their own monitoring and observability pipelines.

What it is

Telegraf is a plugin-driven data collection agent maintained under the InfluxData organisation. It runs as a single process that reads data from configured inputs on an interval, passes that data through optional processors and aggregators, and writes the result to one or more configured outputs on a flush interval. More than 300 plugins ship with the project, spanning system monitoring, cloud services, message passing, industrial device protocols, and generic transports. Configuration lives in TOML, which keeps the setup explicit and readable rather than requiring code to wire a pipeline together.

The concrete problem it solves is the collection tier of a metrics pipeline. Instead of writing and maintaining a bespoke scraper or shipper for every source, a user declares which inputs and outputs they want in TOML and lets the agent handle scheduling, buffering, and delivery. It replaces hand-rolled collection scripts and per-source exporters in the monitoring ecosystem, and it also covers the log side through inputs such as File, Tail, and Directory Monitor. Where a user needs behaviour no plugin provides, the project supports user-defined code to collect, transform, and transmit data.

Key capabilities

• Ships over 300 plugins covering inputs, processors, aggregators, and outputs, including system monitoring for CPU, Memory, Disk, Network, SMART, Docker, and Nvidia SMI.
• Reads industrial and device data through OPC UA and Modbus plugins.
• Consumes messaging systems with AMQP, Kafka, and MQTT plugins, and exposes Prometheus and OpenTelemetry monitoring integrations.
• Collects networking telemetry via Cisco Telemetry MDT and gNMI.
• Covers universal transports and stores with Exec, HTTP, HTTP Listener, SNMP, and SQL plugins.
• Handles logs through File, Tail, and Directory Monitor inputs.
• Compiles into a standalone static binary with no external dependencies, and is configured entirely through TOML.

Who uses it and how

• Infrastructure teams that need one agent on every host to gather CPU, memory, disk, network, Docker, and SMART metrics without installing a separate collector per source.
• Platform and SRE groups running Kafka, MQTT, or AMQP pipelines that must bridge broker data into a time-series store or another sink.
• Industrial and operational technology teams pulling device data over OPC UA or Modbus into the same monitoring stack as their server metrics.
• Networking teams ingesting streaming telemetry from Cisco devices via Telemetry MDT or gNMI.
• Windows estates, using the Event Log, Management Instrumentation, and Performance Counters inputs alongside the same TOML configuration model used elsewhere.

Getting started

Install from the install guide, which covers binary builds, Docker images, RPM and DEB packages, and other builds, then follow the quick start to define a TOML configuration and pass it to the agent. A Docker image is published as library/telegraf on Docker Hub.

How it compares

The facts provided name no comparable tools and no list of paid products this project replaces, so Telegraf stands alone in this registry.

When to use it — and when not to

A self-hoster operates the agent itself, its TOML configuration, and the delivery path to whichever outputs are configured, so the operational burden is the pipeline rather than a database. Teams that want a hosted, configuration-free collection service, or that only need a narrow single-purpose exporter, should look elsewhere instead of adopting a general agent. The project is actively maintained with a recent push, but the documentation is spread across per-plugin READMEs under the plugins directory, so users should expect to read individual plugin documentation rather than one consolidated reference.