Welcome to the MariaDB Community Docs
A modern MariaDB knowledge base for beginners, AI developers and cloud users
What is this?
A community-driven, multilingual MariaDB documentation site. We're not trying to replace the official mariadb.com/kb reference — that has its place. Instead this site is task-driven and scenario-driven, aimed at:
- Beginners picking up a relational database for the first time
- AI developers / vibe coders using Cursor, Claude Code or Windsurf and needing LLM-friendly schemas, MCP wiring, vector search
- Cloud users trying to pick between AWS RDS, Aliyun, Tencent, Azure, SkySQL
Why not just read the official docs?
The official site is comprehensive but stuck in the early-2010s information architecture:
- Many entry points, high barrier, no "what should I do right now?" guidance
- Limited Chinese / Korean / Spanish / Portuguese coverage
- Almost nothing on agent-era workflows (MCP, vectors, NL → SQL)
- Cloud guidance is scattered across each vendor's own docs
We try to fill those gaps.
Where to start?
5-minute Quickstart
Docker / Homebrew / one-click cloud.
For beginners
Database, table, index, transaction — explained plainly.
For AI developers
LLM-friendly schemas, MCP, safe agent write patterns.
MariaDB on the cloud
Pick the right managed flavor.
Versions & forks
Mainline, LTS, MySQL diffs, Percona landscape.
Operations
Backups, monitoring, migration from MySQL/PG.
Current version snapshot
- Latest LTS: MariaDB 12.3 LTS (released 2026-05) — production pick
- Previous LTS still maintained: 11.8 LTS (VECTOR GA) and 11.4 LTS (supported until 2029)
- Rolling mainline: 12.0 / 12.1 / 12.2 are all EOL now, superseded by 12.3 LTS; the next rolling series hasn't started — just use 12.3 for the newest features
- MySQL upstream: fully diverged — MariaDB no longer merges from MySQL
See Versions & forks.
Contributing
This site is open source. PRs welcome on GitHub for translations (ko / es / pt), corrections, and new topics.
💡 AI-friendly tip: every page on this site will soon be available as plain markdown by appending
.mdto its URL, so your agent can ingest it directly.