// the find
grafana/alloy
OpenTelemetry Collector distribution with programmable pipelines
Grafana Alloy is an OpenTelemetry Collector distribution that replaces the old Grafana Agent, adding a custom HCL-like configuration language (Alloy syntax) on top of the standard OTel collector component model. It targets teams already in the Grafana stack who want more expressive pipeline logic than vanilla OTel YAML allows.
The expression-based config language is genuinely better than OTel's YAML for complex pipelines — you can reference component outputs directly and build conditional logic without pile-ups of processors. The built-in UI for visualizing and debugging live pipelines is a real time-saver when chasing why metrics are dropping. Clustering with automatic workload distribution is first-class, not bolted on — useful for large scrape fleets where you'd otherwise hand-partition targets. The release cadence (minor every three weeks) is fast enough that upstream OTel component support stays reasonably current.
The Alloy config syntax is a proprietary DSL you have to learn and maintain tooling for — if you later move off Grafana, your configs don't transfer. The `remotecfg` centralized configuration feature pulls config from Grafana Cloud; self-hosted central config is not well-supported, so it's a soft lock-in. The component library is heavy with Grafana-ecosystem integrations (Loki, Pyroscope, Mimir) while more neutral destinations get less polish. Migrating from Grafana Agent Static requires running the `convert` CLI and then manually auditing the output — the conversion is not always clean for complex configs.