Hands-On PMM 3: Building a Professional Database Observability Stack

Most monitoring systems tell you if a host is up, but they can’t tell you why your database is underperforming. Percona Monitoring and Management (PMM) is an open-source observability platform built to troubleshoot MySQL, PostgreSQL, MongoDB, Valkey, and Redis at the query level. “Why is the database slow?” is the most common question a DBA will face, and this 3-hour workshop provides the technical framework to answer it using PMM 3.

In this interactive lab, you will fork a template repo and:

• Hunt the Bottleneck: Use Query Analytics (QAN) to identify slow or resource-intensive queries by measuring execution time and frequency.

• Signal vs. Noise: Perform a deep-dive review of the most critical PMM dashboards. We will identify which specific graphs indicate immediate health threats and which ones are just background noise.

• Audit Health: Run Percona Advisors locally for automated configuration, security, and performance recommendations.

• Secure by Design: Deploy PMM 3 using its new rootless container architecture on Kubernetes.

• Correlate Load: Connect database metrics with system-level CPU and I/O to find the source of lag during simulated incidents.

This session requires a laptop and a GitHub account. The goal is for you to leave with more than just a functional PMM 3 deployment. You’ll gain a hands-on framework for correlating performance signals across your stack, allowing you to stop guessing and start fixing when a database starts slowing down.