Grafana Loki

Grafana Loki is a log aggregation system built for teams that need to collect, store, and query logs at scale without paying the storage costs that come with full-text indexing. Instead of indexing every word in every log line, Loki indexes only the metadata labels attached to each log stream. That single architectural choice makes it significantly cheaper to run than alternatives, while still delivering fast queries through its own query language, LogQL.

Loki pairs naturally with Grafana for visualization and Prometheus for metrics, which means you can move between logs, metrics, and traces in a single UI without context-switching between tools. If you're already running a Prometheus-based observability stack, Loki slots in with minimal friction.

Key capabilities:

• Label-based indexing keeps storage requirements far lower than full-text index solutions, especially at petabyte scale
• Object storage backend (S3, GCS, Azure Blob, and others) for durable, cost-effective log retention
• LogQL query language lets you filter, aggregate, and build metrics directly from log lines
• Alerting integration evaluates incoming log data and routes alerts through Prometheus Alertmanager
• Promtail agent handles log collection with the same service discovery mechanism Prometheus uses, plus labeling and filtering before ingestion
• Real-time log tailing so you can watch logs as they arrive
• No ingestion format requirements means you can format and parse at query time rather than at write time
• Multi-tenancy built in, suitable for teams running shared infrastructure

Loki is horizontally scalable and designed to stay operationally simple. It's released under AGPLv3, has over 12,000 GitHub stars, and is actively maintained by Grafana Labs alongside a community of 400+ contributors. Self-hosted deployment is the open source path; Grafana Labs also offers a managed cloud version and an enterprise tier for organizations with stricter data localization requirements.