Percona adds high-availability PMM to cut monitoring outages

Percona has launched a high-availability deployment option for Percona Monitoring and Management in version 3.10.0, aiming to remove a single point of failure in database monitoring.

Called PMM High Availability, the new option is designed for Kubernetes environments and runs three PMM servers behind a single address. If the active server, pod or node fails, another instance takes over, allowing monitoring and alerting to continue without requiring users or connected services to change endpoints.

The move addresses a gap that can emerge in production database environments where core systems are built with replication and failover, but the monitoring layer still relies on a single server. In those cases, teams can lose alerting when the monitoring server fails, leaving them unaware of problems until customers report them.

The software is free and open source, with no additional licence fee beyond PMM itself. It is installed through two Helm charts, and upgrades can be carried out with rolling updates so maintenance does not take the monitoring system offline.

The release also adds backup and restore orchestration for the full monitoring stack. Backups can be sent to S3-compatible storage or a shared volume, with restore options either in place or into another Kubernetes namespace.

The architecture has been validated on Amazon Elastic Kubernetes Service and Red Hat OpenShift, including Red Hat OpenShift Service on AWS. PMM 3.10.0 also adds ARM64 support for PMM Server, widening deployment options for organisations using ARM-based infrastructure.

The system is aimed at infrastructure teams, database administrators, site reliability engineers and platform engineers responsible for production database estates. It is positioned for organisations with strict uptime requirements and established on-call processes that want to remove single points of failure from operational tooling.

Reliability gap

Peter Farkas, Chief Executive Officer of Percona, said the issue often appears in environments where the databases themselves have already been hardened against outages.

“When a single-instance PMM server fails, alert rules stop running, meaning critical notifications may never reach on-call teams. In the worst cases, teams may only learn of an incident when a customer notifies them. This is why database observability should be held to the same standards of availability and reliability as the systems it supports. With PMM HA, we're helping organizations close that gap, making what were once silent outages visible, and delivering some much-needed resilience to these essential systems,” Farkas said.

Percona Monitoring and Management is the company's database observability platform. The new high-availability option extends the product rather than replacing its standard deployment model, giving users an alternative setup for environments where continuous visibility is essential during incidents or maintenance.

Under the new design, clients can queue metrics during failover and send them once connectivity resumes. This is intended to reduce breaks in monitoring data and preserve graph continuity when one PMM instance becomes unavailable.

Open source model

Percona is also using the launch to underline its open source approach in monitoring software as well as database tooling. Organisations can adopt the high-availability architecture without extra software licensing costs, a point that may appeal to teams already running open source databases and wanting operational tools built on a similar model.

Percona has built its business around open source database software, support and services. It works across technologies including MySQL, MariaDB, PostgreSQL, MongoDB, Valkey and Redis, and also offers Kubernetes operators for cloud-native database deployments.

The launch reflects a growing focus among infrastructure teams on the resilience of support systems around core applications, not just the applications and databases themselves. As operational estates become more automated, the reliability of alerting and visibility tools has become more closely tied to incident response and customer impact.

PMM High Availability is now generally available as part of Percona Monitoring and Management 3.10.0.