Research Notes

Percona Strengthens Database Operations Without the Giant Platform

Research Finder

Find by Keyword

Percona Strengthens Database Operations Without the Giant Platform

PMM High Availability supports organizations that want database flexibility and reliable operations without adopting a broader data platform.

11/18/2026

Key Highlights

  • Percona serves developers and organizations that want database choice without the scope and commitment of a large, integrated data platform.
  • PMM High Availability reduces dependence on a single monitoring server.
  • Recovering buffered metrics preserves diagnostic history, while timely incident response also requires current visibility and working alerts.
  • The Kubernetes architecture introduces operational requirements that customers must weigh against their availability needs.

The News

Percona announced general availability of Percona Monitoring and Management High Availability on October 27, available in PMM 3.10.0. The Kubernetes-based deployment uses three PMM servers behind a shared address, with automatic failover when the active server fails. Percona says clients buffer metrics during the transition and submit them when connectivity resumes. PMM HA is free and open source, with the architecture validated on Amazon EKS and Red Hat OpenShift, including ROSA.

Analyst Take

Not every organization needs a giant data platform. Some developers want to choose their databases, decide where to run them, and get help keeping them healthy. That is where Percona fits.

That choice still comes with production responsibilities. A database can remain available while the team responsible for it loses the ability to see what is happening. PMM High Availability addresses a practical weakness in that setup. Customers choosing flexibility still need dependable monitoring, backup, and recovery capabilities.

During an incident, database administrators need evidence to decide whether to intervene, wait, or reverse a recent change. If monitoring disappears at the same time, they lose diagnostic information and confidence in their next step. An application that still responds may already be slowing down or approaching a resource limit. Waiting for users to report the problem is an expensive substitute for a working alert.

Getting the metrics back afterward is useful. Getting the warning in time to act is more useful. Customers should evaluate how quickly current data and useful alerts become available after a failure, alongside whether the measurements eventually arrive.

The tradeoff is operational effort. A replicated monitoring environment requires resources, maintenance, and recovery procedures. For an organization already running Kubernetes and supporting demanding database availability commitments, that may fit existing practices. A smaller team should assess whether it can manage the additional infrastructure effectively.

Percona’s customers may want to avoid a large data platform, but they also want to avoid assembling every operational capability themselves. The question is whether PMM HA delivers that resilience with an amount of work those teams can reasonably manage.

What Was Announced

PMM HA provides redundancy across the monitoring service and its supporting data stores. Percona’s architecture documentation describes an active PMM server with standby replicas, traffic routing through HAProxy, and operator-managed storage for metrics, query analytics, and configuration data. Another server can take over without requiring clients to discover a new endpoint.

The implementation matters because replicating the dashboard alone would leave other failure points untouched. Monitoring depends on stored measurements and configuration as well as the service presenting them. Customers therefore need to assess the deployment as a complete system, including its storage and Kubernetes dependencies.

PMM 3.10.0 also adds backup and restore orchestration for the HA installation, using S3-compatible storage or a shared volume. Built-in alerts cover conditions including loss of a leader, unstable leader elections, and storage capacity pressure, helping teams identify when the monitoring environment itself is becoming unhealthy.

Looking Ahead

Percona does not need to become a mega data platform to have a meaningful role in enterprise database operations. Its opportunity is to serve developers and organizations that value database choice and want practical help running what they have selected. PMM HA supports that position, provided the deployment remains manageable for the teams it serves.

Free software removes a licensing consideration. Customers will still judge the offering on infrastructure cost, maintenance effort, and recovery behavior. Clear deployment guidance and well-documented failure boundaries will matter, particularly for organizations with limited platform engineering resources.

Percona has room to grow by serving customers who want more choice in their database architecture. The test is whether it can make that choice easier to live with. PMM HA should help teams spend less time building and maintaining their own monitoring recovery procedures. Customer experience will show whether the added Kubernetes complexity is worth it.

Author Information

Stephanie Walter | Practice Leader - AI Stack

Stephanie Walter is a results-driven technology executive and analyst in residence with over 20 years leading innovation in Cloud, SaaS, Middleware, Data, and AI. She has guided product life cycles from concept to go-to-market in both senior roles at IBM and fractional executive capacities, blending engineering expertise with business strategy and market insights. From software engineering and architecture to executive product management, Stephanie has driven large-scale transformations, developed technical talent, and solved complex challenges across startup, growth-stage, and enterprise environments.