Azure Monitor Health Model (Preview): What's New!

Azure Monitor Health Model is a modern solution that combines telemetry, architecture, and business context to offer valuable health insights for your workloads. It continually gathers signals across dependencies to deliver a single, actionable health status. This not only reduces unnecessary alerts but also helps teams become more proactive in their operations, providing a clearer view of their systems, enhanced insights, and quicker troubleshooting.

It aims to solve the pressing operational question: ‘Is my system/service/application in good shape?’ and ‘Which underlying component is affecting health?’

The latest update introduces a more adaptable, workload-focused discovery process (leveraging application insights topology and Azure Resource Graph queries, alongside user and system flow designs) and improves the creation of health signals (utilising recommended signals, importing existing alert rules, and setting dynamic thresholds).

As businesses started creating more complex applications, we saw a chance to make discovery easier and more intuitive. Different teams tend to approach their systems in unique ways; some focus on the application itself, while others look at infrastructure or telemetry. By broadening the discovery options, we empower customers to create health models using familiar frameworks, making it simpler to adjust these models as applications and architectures evolve.

Azure Monitor health models now offer various discovery options:

• Application Insights-based discovery for application-focused modelling.
• Azure Resource Graph (ARG) discovery for scalable, query-based resource selection.
• Support for Service Groups, including nested Service Groups, as part of a comprehensive set of discovery options.

This shift towards loosely coupled modelling allows customers to define health based on application architecture rather than solely on infrastructure groups. Discover more about Discovery options

Discovery Rule

Our aim has been to assist customers in achieving meaningful health insights more swiftly and with less manual effort. By providing platform defaults and highlighting recommended signals, we simplify the alignment of health models with established Azure best practices from the outset. We also maintain support for existing alert strategies and investments, allowing customers to build upon what they already have. These enhancements strike a balance between simplicity, guidance, and flexibility as systems grow.

Health Models now include the following health signal features:

• Resource Health as a default signal, ensuring every model starts with a solid, platform-provided baseline.
• Recommended signals, automatically surfaced in line with Azure service best practices, enhanced through Azure Monitor Baseline Alerts (AMBA) integration.
• Reuse of existing signals, with the ability to import Azure Monitor alert rules as health signals.

 Explore more about Signals

Signals – Recommended and Import from Alert Rules for Azure Resources

Modern cloud applications are designed for resilience, redundancy, and tolerance to partial failure. Health Models reflect this reality by allowing customers to specify what “healthy” looks like for their architecture. Flexible aggregation rules enable teams to model intent rather than just the states of individual components, leading to health views that more accurately align with operational priorities and business impact.

Health Models now support advanced aggregation logic, allowing for the following scenarios:

• Regional resiliency aggregation using numeric thresholds (e.g., 2 out of 4 regions must remain healthy).
• Cluster and fleet health aggregation using percentage thresholds (e.g., 60% of VMs in a cluster must be healthy).

This enables the modeling of resilience patterns, partial failures, and graceful degradation, providing a clearer picture of actual business impact.

Health is most beneficial when it reflects both system performance and application context. By allowing custom health inputs, customers can incorporate signals closely aligned with their business logic and application state. Contextual annotations further enhance analysis, making health timelines simpler to interpret and correlate with change events.

To facilitate this, Health Models now accommodate:

• Custom health report ingestion for health signals from external applications and systems.
• Data annotations that overlay deployments, incidents, and configuration changes on the health state.

To stay ahead of health state changes, you can create Alert rules and set up action groups to trigger automated responses, like notifying users. Now, you can view all alerts on a Health Model and begin troubleshooting.

Note: To take advantage of these new features, upgrade your health models to the latest API version using the built-in migration wizard in the Azure portal for a straightforward guided experience.

Alerts in Health Model

Note: To take advantage of these new capabilities, upgrade your health models to the latest API version using the built-in migration wizard in the Azure portal for a simple, guided experience.