Quickstart
1. Start PMM Server (needs Docker; the script installs it if it's missing)
curl -fsSL https://www.percona.com/get/pmm | /bin/bash
Open https://<server-ip> and log in as admin / admin, then set a new password.
2. Connect a database (on the database host; Debian/Ubuntu shown)
wget https://repo.percona.com/apt/percona-release_latest.generic_all.deb
sudo dpkg -i percona-release_latest.generic_all.deb
sudo percona-release enable pmm3-client release
sudo apt update && sudo apt install -y pmm-client
sudo pmm-admin config --server-insecure-tls --server-url=https://admin:<password>@<server-ip>:443
sudo pmm-admin add mysql --username=pmm --password=<pass> --query-source=perfschema
This assumes a fresh server with its self-signed certificate. For production, use a trusted certificate (drop --server-insecure-tls) and register with a service account token instead of the admin password.
Open Query Analytics and your queries appear within a minute. The pmm database user needs these privileges. Guides for RHEL, Docker, PostgreSQL, MongoDB, Valkey/Redis, Amazon RDS/Aurora, Azure and Google Cloud are in the full quickstart.
Other ways to run PMM Server: Docker · Podman · Kubernetes (Helm) · AWS Marketplace
Product tour
Every query ranked by load, with examples, EXPLAIN plans and table stats. See the Query Analytics docs.
For MySQL, PostgreSQL, MongoDB, Valkey/Redis, ProxySQL and HAProxy. Built on Grafana.
Runs where your databases run: bare metal, VMs, Kubernetes, Amazon RDS/Aurora, Azure and Google Cloud.
Why PMM?
| PMM | Prometheus + Grafana, self-built | Hosted DB monitoring |
|---|
| Per-query analytics | Built in | Build it yourself | Yes |
| DB dashboards, advisors, alert templates | Built in | Assemble by hand | Yes |
| Where your data lives | Your servers | Your servers | Vendor cloud |
| Cost | Free, open source | Free + your time | Per host, per month |
Compared with tools you know
Prometheus + exporters
Good if you want to assemble it yourself. PMM ships the exporters, curated dashboards, query analytics, advisors and alert templates as one install with one upgrade path.
Datadog Database Monitoring
SaaS, priced per database host. PMM is self-hosted, so query text and samples stay on your servers.
pganalyze
Commercial and PostgreSQL only. PMM covers PostgreSQL, MySQL, MongoDB and Valkey in one tool.
How it works
PMM Server (Grafana, VictoriaMetrics, ClickHouse, PostgreSQL) stores metrics and query data. PMM Client (pmm-agent plus exporters) runs next to each database and sends them. Details are in the architecture reference.
FAQ
PMM is built on Grafana and VictoriaMetrics (PromQL-compatible) and adds the database parts: exporters configured for you, query analytics stored in ClickHouse, curated dashboards, advisors, alert templates and backups. It installs and upgrades as one product.
The client reads statistics the database already collects (Performance Schema or the slow log, pg_stat_statements or pg_stat_monitor, the MongoDB profiler or log), and you choose the query source for each service. See choosing a query source.
Yes. Add the instance from the PMM web interface; nothing is installed on the database host. For RDS, PMM also reads Enhanced Monitoring metrics through the AWS API. See Connect Amazon RDS.
No. Every feature is in the open source release. Percona sells support, not features.
PMM Server: pull the new image for Docker or Podman, or run helm upgrade for Helm. PMM Client: your package manager. See the upgrade guide.
Community and support
If PMM saves you time, give the repo a ⭐. It helps other DBAs find it.
License
