influxdata/telegraf

Telegraf is an agent, a program that runs continuously in the background, that collects data from various sources, processes it, and sends it somewhere else for storage or analysis. It is commonly used for system and infrastructure monitoring: for example, measuring CPU and memory usage on a server and sending those numbers to a database where you can graph them over time. What makes Telegraf flexible is its plugin system. It has over 300 plugins organized into input plugins (where data comes from, things like system CPU/memory/disk stats, Docker containers, databases, network devices, IoT sensors via Modbus or OPC UA, messaging systems like Kafka and MQTT, and services like Prometheus and OpenTelemetry), processor plugins (for transforming or filtering data in transit), and output plugins (where data goes, such as InfluxDB, a time series database, or other destinations). You pick the plugins you need by listing them in a configuration file written in TOML (a simple, readable settings format), and Telegraf reads from all configured inputs on a schedule and flushes the collected data to outputs. Telegraf compiles to a single standalone binary with no external dependencies, making it straightforward to deploy on a server or as part of a containerized setup. It is written in Go. You would use Telegraf when you want a single, consistent tool to pull metrics and logs from many different systems, servers, databases, cloud services, industrial equipment, and funnel them into a monitoring or alerting platform without writing custom collection scripts.