Grafana Alloy

Alloy is the agent you install on servers or Kubernetes to collect metrics, logs, traces, and profiles and ship them to whatever backend you use. It is Grafana Labs' distribution of the OpenTelemetry Collector with Prometheus pipelines built in. Apache 2.0 and free, with no paid edition. Anyone still running Grafana Agent or Promtail is headed here: both hit end of life, and Grafana points everyone to Alloy.

It runs as a systemd service, a Windows service, or a Kubernetes DaemonSet, on Linux, Windows, macOS, and FreeBSD. Config is Alloy's own syntax, not YAML, which is the real learning cost. Grafana's sizing rule is roughly 0.4 CPU cores and 11 GiB of memory per million active series. Set a storage path or buffered data disappears on restart. Minor releases land every three weeks, and they do ship breaking changes.

Free at every team size. The cost is the backend: self-hosted Prometheus and Loki, or Grafana Cloud, which has a free tier and then Pro from $19/mo plus usage. Fleet Management, the remote config for many collectors, is a Grafana Cloud feature. For something vendor-neutral, run the upstream open-telemetry/opentelemetry-collector or vectordotdev/vector.

The catch: it reports anonymous usage stats by default until you pass --disable-reporting. And the engine that accepts stock Collector YAML is marked experimental, so a straight OpenTelemetry config is not a safe path yet.