Slide 4 of 11

Loki

Loki: Cost-effective log aggregation

What it’s for: Logs including application logs, system logs, and audit trails.

Good to know

  • Query language: LogQL, which is similar to PromQL
  • Loki indexes labels rather than full log content, which lowers storage cost. You filter by labels first, then search within that subset, instead of running arbitrary full-text search.
  • It shares Prometheus’s label model, so your Prometheus labels carry over.

Best for

  • Teams already using Prometheus
  • Cost-conscious log aggregation
  • Correlating logs with metrics

Documentation

View the full documentation.

Loki data source

Script

Loki is Grafana’s log aggregation system, and it takes a fundamentally different approach than traditional log systems. Here’s the key insight: Loki only indexes labels (metadata about your logs), not the log content itself. That means much lower storage costs.

The limitation? You can’t do arbitrary full-text search across all your logs. Instead, you filter by labels first, then search within that subset. For most log use cases, this works well.

If you’re already using Prometheus for metrics, Loki is the natural companion. They share the same label model, so your Prometheus labels carry over to Loki. The query language is LogQL, which is similar to PromQL.

And because Loki is built by Grafana Labs, you get native integration. It scales horizontally and works in both Grafana Cloud and self-managed environments.

For cost-conscious log aggregation, especially alongside Prometheus, Loki is a strong fit.

Previous Next