Database monitoring fundamentals on Grafana Labs

Start monitoring databases

Wed, 07 Oct 2026 10:00:46 -0500

Welcome

Grafana Cloud gives you two ways to monitor your databases: Database Observability and database integrations.

If you run MySQL or PostgreSQL and want to understand what’s happening inside a query, Database Observability gives you that visibility. If you want metrics, dashboards, and alerts across a broader set of databases, database integrations have you covered.

Two monitoring approaches

Database Observability: query-level visibility for MySQL and PostgreSQL, so you can see how tracked queries perform, not just aggregate database metrics
Database integrations: database-level metrics, prebuilt dashboards, and alerting rules across MongoDB, Redis, Memcached, Cassandra, Oracle, and the SQL databases

What this journey covers

Database Observability: what it shows, how it works, and how to set it up for MySQL or PostgreSQL
Database integrations: what they provide and how to set them up for MongoDB, Redis, Memcached, Cassandra, MySQL, PostgreSQL, and Oracle

Supported databases

Database Observability: MySQL and PostgreSQL, both self-managed and on managed services (Amazon RDS, Aurora, Google Cloud SQL, and Azure Database)
Database integrations: MongoDB, Redis, Memcached, Cassandra, MySQL, PostgreSQL, and Oracle

How Database Observability and integrations differ

Wed, 07 Oct 2026 10:00:46 -0500

Two different layers

A database integration and Database Observability aren’t two versions of the same product, and you don’t choose one over the other. They monitor different layers of the database and answer different questions.

	Database Observability	Database integration
What it monitors	Tracked queries	The database instance as a whole
Question it answers	How do tracked queries perform, and why?	Is the database healthy?
What you see	Rate, errors, and duration for tracked queries, explain plans for SELECT queries, wait events, and query samples	Aggregate, instance-level metrics such as connections, throughput, and replication, with prebuilt dashboards and alerts
Databases	MySQL and PostgreSQL	Many, including MongoDB, Redis, Memcached, Cassandra, Oracle, MySQL, and PostgreSQL

Prerequisites

Wed, 07 Oct 2026 10:00:46 -0500

For hands-on exercises

Both approaches need a free Grafana Cloud account (sign up free) and Grafana Alloy, which collects data and sends it to Grafana Cloud. What you need beyond that depends on which approach you’re setting up.

Database integration requirements

For a database integration you need a running instance of a supported database: MongoDB, Redis, Memcached, Cassandra, Oracle, MySQL, or PostgreSQL. You configure an exporter for that database as part of setup.

Database Observability version requirements

For Database Observability, you need:

A running MySQL 8.0 or later or PostgreSQL 14 or later database
A database user with monitoring privileges
Versions shown in the following table

Requirement	Details
Grafana Alloy	Version 1.17.0 or later
MySQL	Version 8.0 or later (MySQL 5.7 and MariaDB aren’t supported)
PostgreSQL	Version 14 or later, with the `pg_stat_statements` extension enabled in each database you monitor

Database Observability is available on the Grafana Cloud free tier and needs the Editor or Admin role. It is billed as a per-host hourly fee for each connected database, plus metrics and logs usage. On the free tier, metrics and logs, including query samples, are retained for 14 days. See Database Observability pricing.

The monitoring user needs two kinds of access. Query statistics need the base privileges: for MySQL, PROCESS, REPLICATION CLIENT, and SELECT on performance_schema; for PostgreSQL, the pg_monitor and pg_read_all_stats roles. Schema details and explain plans also need access to the tables you want to see. Grant SELECT and SHOW VIEW on each MySQL schema you monitor, or on all schemas if you want every schema. For PostgreSQL, grant USAGE on each schema and SELECT on all tables in that schema, or grant pg_read_all_data if you want every object.

If you’re running MySQL 5.7, MariaDB, or an older PostgreSQL version, Database Observability isn’t available for those.

A database integration can still monitor the database instance’s overall health, such as connections, throughput, and replication status. However, that’s a different capability, not a substitute. An integration reports aggregate, instance-level metrics, not the per-query explain plans, wait events, and query samples that Database Observability provides.

Recommended prior knowledge

Topic	Why it helps
Basic Grafana navigation	You’ll use dashboards and the Configuration page.
Terminal or command line basics	Alloy installation uses the CLI.
Basic database administration	You’ll create a monitoring user and configure permissions.

Just exploring? You can complete the journey without hands-on practice. The concepts are valuable on their own.