How Alloy works on Grafana Labs

Alloy in an observability setup

Fri, 14 Aug 2026 09:40:11 -0500

Alloy in an observability setup

A typical observability setup has three layers:

Data sources: the applications and infrastructure that generate telemetry
Collection: the tools that gather and process that telemetry
Storage and visualization: the backends that store the data, and the frontends you use to query and explore it

Alloy runs in the collection layer

Alloy sits in the collection layer, between your data sources and your backends. Data flows from your sources, through Alloy, and on to storage.

This collection layer is where you decide what telemetry to gather, how to clean it up, and where to send it.

How telemetry enters Alloy

Fri, 14 Aug 2026 09:40:11 -0500

How telemetry enters Alloy

Alloy gathers telemetry data by scraping, receiving pushed data, or tailing a source.

Scrape

Alloy opens a scheduled connection to a target and requests the current data. A common example is scraping Prometheus endpoints to pull metrics.

Push

Alloy opens a listener and passively waits for the source to send data. A common example is receiving OpenTelemetry Protocol, or OTLP, data that an application pushes to Alloy.

Tail

Alloy watches a file on disk for new appended content and reads it incrementally. A common example is reading new lines in a log file as they are written.

Which of these you use depends on the signal and the source. Alloy can also discover endpoints automatically and integrate with specialized systems.

What Alloy does

Fri, 14 Aug 2026 09:40:11 -0500

What Alloy does

In the collection layer, Alloy performs three main jobs: it collects telemetry, transforms it, and sends it to your backends.

Collect telemetry data

Alloy gathers telemetry from sources in your infrastructure. In addition to scraping, receiving pushed data, or tailing a source, it can also:

Automatically find the endpoints to collect from in Kubernetes or Docker (service discovery)
Integrate with specialized systems such as databases and message queues

Transform and process data

Before data reaches storage, Alloy can reshape it. It can:

Filter out unwanted data or redact secrets
Add identifying fields, often called labels, and other context
Standardize attribute names across services
Keep only a subset of high-volume data (sampling)
Convert between formats
Route different data types to different destinations

Send to backends

Alloy delivers the processed telemetry to the systems you choose:

Grafana Cloud
Self-managed Grafana stack components such as Mimir, Loki, Tempo, and Pyroscope
Prometheus-compatible databases
OpenTelemetry-compatible backends

You can send to more than one destination from a single configuration.

Resources

Fri, 14 Aug 2026 09:40:11 -0500

Resources

Use these resources to learn more about what Alloy is and how it works.

Documentation

Video

Grafana Alloy for Beginners series