Get Started

MariaDB gives your applications and agents one durable place for their data: a single source of truth that scales with your workload, runs anywhere, and is built on curated open source. These docs cover MariaDB Server, both Community and Enterprise, plus the MariaDB Platform components that extend it.

Run MariaDB Server anywhere. Your code stays the same. Your application connects through a MariaDB connector, and your coding agent connects through the MariaDB MCP server. Both reach MariaDB Server, Community or Enterprise. MariaDB Server runs on any of four environments. You run it with Docker on your laptop, with Linux on your servers and VMs, or with the operator on your Kubernetes cluster, and MariaDB Enterprise Platform adds MaxScale, Galera Cluster, ColumnStore, Exa, and GridGain to any of these. Or MariaDB runs it for you in MariaDB Cloud. Switching environments changes only your connection settings: host, user, password, and TLS.

The Get Started pages follow one small online shop. It starts as a single products table that an application queries, and it grows into an application account, backups, a replica, and a weekly sales report. The pages are written for the developer who builds that shop, so each step is SQL or code that you run yourself. The people who keep the shop online and the people who report on its sales use the same server, and their part of the story has pages of its own.

Run databases for a team? As a database administrator, start with Deployment to choose where the server runs, then go to Operations to back it up, restore it, and add a replica. Report on the data? As an analyst, go to Analytics, which moves the shop's sales history into a columnar table and joins it to the live tables in one query.

Pick a tab to get a server running, either with your coding agent or by hand. The Get Started pages follow the Docker tab, and each other tab says what changes for you. On Windows or macOS without Docker, install MariaDB Server with the Windows MSI package or with Homebrew on macOS, then continue at step 2.

Add the MariaDB plugin to Claude Code:

claude plugin marketplace add mariadb/ai-plugins
claude plugin install dev@mariadb

Then start Claude Code in your project and paste this:

Set up MariaDB for this project.

1. Use the MariaDB sandbox tools to deploy the latest
   MariaDB version they offer, on port 3310, with root
   password "sandbox". If the port is busy, pick the next
   free port and tell me.
2. Create a database called shop with a products table
   (id, name, price) and add ten rows.
3. In a Python virtual environment, install MariaDB
   Connector/Python with `pip install --pre mariadb`, then
   write a script that prints every product under 50.
   Run it.
4. Show me how to connect to the sandbox with the mariadb
   client. The sandbox has no TLS, so include --skip-ssl.

The first run downloads the mariadb-shell client and a MariaDB server binary, about 75 to 105 MB in total depending on your platform, and the sandbox runs natively, so you don't need Docker. The --pre flag installs the MariaDB Connector/Python 2.0 release candidate, which needs no C library, so it suits a sandbox. Step 3 below uses the GA release for application code. The sandbox doesn't include the mariadb command-line client, so install it from the Linux tab or your package manager. Using Codex, Pi, Hermes Agent, or OpenCode? The MariaDB AI plugins README has the install steps for each.

The prompt creates the shop database with ten products of its own, so skip step 2 below, and use port 3310 and the password sandbox in step 3.

docker run -d --name mariadb -p 3306:3306 \
  -e MARIADB_ROOT_PASSWORD=changeme \
  mariadb:lts

until docker exec mariadb healthcheck.sh --connect >/dev/null 2>&1; do
  sleep 2
done

docker exec -it mariadb mariadb -uroot -pchangeme

This runs the official MariaDB image from Docker Hub. On first start, the container spends a few seconds creating its data directory, so the until loop waits for the image's healthcheck.sh script to report that the server accepts connections. For volumes, configuration, and the other environment variables, see Installing and Using MariaDB via Docker.

# Debian and Ubuntu
sudo apt install mariadb-server

# RHEL, Rocky Linux, AlmaLinux, Fedora
sudo dnf install mariadb-server
sudo systemctl enable --now mariadb

sudo mariadb

Your distribution's package might be a few releases behind. To get the current release, add the MariaDB package repository first.

A package install gives root no password. The root account signs in through Unix socket authentication, so open the client with sudo mariadb, and run the step 3 script after the Server page creates the shop_app account.

kubectl create secret docker-registry mariadb-enterprise \
  --docker-server=docker.mariadb.com \
  --docker-username=<email> \
  --docker-password=<customer-download-token>

helm repo add mariadb-enterprise-operator https://operator.mariadb.com
helm install mariadb-enterprise-operator-crds \
  mariadb-enterprise-operator/mariadb-enterprise-operator-crds
helm install mariadb-enterprise-operator \
  mariadb-enterprise-operator/mariadb-enterprise-operator \
  -f values.yaml

The MariaDB Enterprise Kubernetes Operator is part of MariaDB Enterprise Platform. It pulls images from docker.mariadb.com with your Customer Download Token, which the mariadb-enterprise secret holds. The values.yaml file names that secret under imagePullSecrets. See Install the operator with Helm.

The operator alone runs no database yet. Before step 2, create a MariaDB resource as the Kubernetes tab on the Deployment page shows.

Sign in at cloud.mariadb.com and launch a Serverless service. It's ready in seconds.
Click Manage on the service, choose Manage allowlist, and add your IP address. A new service blocks all external traffic until you do.
Click Connect on the service and copy the host, port, user, and password.
Connect:

mariadb --host <host> --port <port> \
  --user <user> -p --ssl-verify-server-cert

The MariaDB Cloud quickstart walks through each screen, and Configuring Firewall covers the allowlist.

In steps 2 and 3, use the host, port, user, and password from the Connect dialog, and turn on TLS in the script with ssl_verify_cert=True. The Cloud page shows the full script.

From Install to Your First Query

Three steps take you from a running server to rows coming back in Python.

Run MariaDB Server

Use the Docker tab above to follow along page by page, or another tab with the change it notes. Deciding where to run it in production? Deployment walks through the options.

Create Some Data

Open the mariadb client and run:

CREATE DATABASE IF NOT EXISTS shop;

CREATE TABLE shop.products (
  id    INT PRIMARY KEY AUTO_INCREMENT,
  name  VARCHAR(100) NOT NULL,
  price DECIMAL(8,2) NOT NULL
);

INSERT INTO shop.products (name, price) VALUES
  ('Desk lamp', 39.00),
  ('Standing desk', 249.00),
  ('Notebook', 4.50);

The MariaDB basics guide covers databases, tables, and your first statements in more depth.

Query It From Python

This example uses Python. MariaDB has connectors for Node.js, Java, C#, C++, and more, listed under Connect From Your Language.

Install MariaDB Connector/Python:

pip install mariadb

The current GA release, MariaDB Connector/Python 1.1, builds against MariaDB Connector/C 3.3.1 or later, so install Connector/C first if your system doesn't have it. Then run:

import mariadb

conn = mariadb.connect(
    host="127.0.0.1",
    port=3306,
    user="root",
    password="changeme",
    database="shop",
)
cur = conn.cursor()
cur.execute("SELECT name, price FROM products WHERE price < ?", (50,))

for name, price in cur:
    print(f"{name}: {price}")

conn.close()

On the sandbox, use port 3310 and the password sandbox. The Python connector quickstart picks up from here with transactions, connection pools, and error handling.

Keep Your Agent in the Docs

This documentation is also a Model Context Protocol (MCP) server, which is how coding agents reach tools and data outside the editor. Add it once, and your agent can search and read current MariaDB docs while it works with you, so it stops guessing at syntax.

claude mcp add --transport http mariadb-docs \
  https://mariadb.com/docs/~gitbook/mcp

Already running MariaDB? The Server docs cover administration and the full SQL reference. For MaxScale, clustering, and analytics, head to MariaDB Platform.

Build With MariaDB

Jump to a feature, find the connector for your language, or clone a sample app and run it.

AI and Vector Search

Store embeddings in a native VECTOR column and run similarity search in SQL, next to the rest of your data.

JSON

Keep flexible documents in the same tables as relational data, and query both with one statement.

Temporal Tables

Let the database keep every version of a row, then query your data as it was at any point in time.

Modern SQL

Recursive and non-recursive common table expressions, window functions, and sequences, for queries that go past a basic SELECT.

Columnar Analytics

Load operational data into ColumnStore tables in the same database, and run analytical queries there instead of in a separate warehouse.

Sample Apps

Open source apps on GitHub, from to-do lists to RAG, in Python, Node.js, Java, .NET, and C++.

Connect From Your Language

Python

MariaDB Connector/Python

Node.js

MariaDB Connector/Node.js

Java

MariaDB Connector/J

PHP

PHP's MySQL extensions

C# and .NET

MariaDB Connector/NET

C++

MariaDB Connector/C++

Pick a Path

Each path is a page that walks through one job in order and links to the docs you need along the way.

Build an AI Application

Store and search vectors, connect your AI framework, and give an agent access to the database.

Develop an Application

Connect from your language, write the SQL, and build the schema out from there.

Analyze Your Data

Run analytics at scale with ColumnStore and Exa, and pick the right storage engine for each reporting table.

Move to MariaDB

See what's different, then bring your schema and data over from MySQL, PostgreSQL, or Amazon RDS.

Deploy MariaDB

Install from packages, run on Kubernetes with the operator, or launch a managed service on MariaDB Cloud.

Run It in Production

Back up and restore, run clustered and routed topologies, and tune performance. For courses and certification, see MariaDB University.

Get Help

Community

Ask questions and meet the people who build MariaDB on mariadb.org and the MariaDB Zulip chat.

Report a Bug

File bugs and feature requests for MariaDB Server, the connectors, and MaxScale in MariaDB Jira.

Get Support

Support for MariaDB Cloud and MariaDB Enterprise Platform customers.

Read Offline

Download the documentation as PDFs to read offline, one per documentation set, with working cross-references.