GreptimeDB Documentation

GreptimeDB is an open-source observability database that stores metrics, logs, and traces in one engine. Run it as the single OpenTelemetry backend in place of Prometheus, Loki, and Elasticsearch, keep the data on object storage, and query all of it with SQL and PromQL.

• Getting Started

  Install GreptimeDB, write your first rows, and run your first query.

• User Guide

  Ingestion protocols, queries, pipelines, flows, and running GreptimeDB in production.

• For AI Agents

  Run GreptimeDB from a coding agent, with no custom integration.

• Reference

  SQL syntax, functions, and data types.

Ask about GreptimeDB

Installation, ingestion, queries, configuration — answered from this documentation.

Why GreptimeDB

Replace three systems with one. Metrics, logs, and traces go into one columnar engine with native OpenTelemetry support, instead of a separate store, query language, and operational overhead for each signal.

Pay object-storage prices. S3, Azure Blob, and GCS are the primary store, and compute scales independently of them. Columnar storage and observability-tuned compression cut cost by up to 50x — OceanBase Cloud keeps 300 TB of logs and audit data on GreptimeDB and reports 60%+ lower storage cost after moving from Loki.

Keep your collectors and dashboards. Prometheus remote write and OTLP on the write path, PromQL and the Jaeger API on the read path, each on its own HTTP endpoint. SQL arrives over the MySQL and PostgreSQL wire protocols. Grafana dashboards keep working. Each protocol page states what it covers and what it does not.

Give agents one query interface. In Agent RCA Bench, six models investigated the same 14 incidents on GreptimeDB and on Prometheus, Loki, and Tempo: 40% fewer wrong diagnoses, 48% fewer input tokens read, about 45% lower cost to run.

More in Why GreptimeDB and Observability 2.0 and wide events.

Build with AI agents

The quickstart guide is hosted as a fetchable Markdown file. A coding agent reads it at runtime, with no installation required.

AGENT QUICKSTART

Read https://docs.greptime.com/SKILL.md and follow the instructions to use GreptimeDB with your AI agent — deploy, configure, ingest, and query.

• MCP Server — query tools an agent can call: SQL, TQL, and range queries, read-only by default. It also manages pipelines and dashboards, which write.
• Skills — pipelines, flows, triggers, table design, and performance diagnosis, in the Agent Skills format.
• Semantic layer — greptime.semantic.* table metadata recording the signal, source, and unit behind each table.
• llms.txt — a structured index of the whole site; append .md to any page URL for its raw Markdown.

Keep exploring

• Tutorials: end-to-end walkthroughs, such as monitoring a Kubernetes cluster.
• GreptimeDB Enterprise: read replicas, workload isolation, RBAC, audit logging, and disaster recovery.
• Contributor Guide: internals, for anyone working on GreptimeDB itself.
• Roadmap: the project's direction and planned work.
• Release Notes: every released version.
• FAQ: common questions about deployment, ingestion, and queries.