Table of Contents

Azure Monitor

Azure Monitor is Microsoft's fully managed, cloud-native observability platform. It is the alternate monitoring stack in this section — where Prometheus + Grafana is self-hosted open source you run and scale yourself, Azure Monitor is a managed service that collects, stores, analyzes, and alerts on telemetry with no infrastructure for you to operate.

Note

This section documents the two stacks side by side. See the Monitoring overview for the self-hosted Prometheus + Grafana stack, and choosing a stack below for when each one fits. The two are not mutually exclusive — a common hybrid keeps Prometheus for Kubernetes workloads and uses Azure Monitor for Azure platform resources.

What the Azure-native stack provides

Component Role Prometheus-stack equivalent
Azure Monitor Metrics Near-real-time, dimensional platform and custom metrics Prometheus TSDB
Log Analytics Log and event store queried with KQL Loki / Prometheus (partial)
Application Insights Application performance monitoring (APM) — requests, dependencies, traces, exceptions Prometheus client libraries + Tempo
Azure Monitor Agent (AMA) Collects guest metrics and logs from VMs and servers via Data Collection Rules Node Exporter + scrape configs
Azure Alerts Metric, log, and activity-log alert rules Prometheus alert rules
Action Groups Routes and delivers alert notifications (email, SMS, webhook, ITSM, Functions) Alertmanager receivers
Workbooks / Dashboards / Managed Grafana Visualization and reporting Grafana
Azure Monitor Managed Service for Prometheus Managed, Prometheus-compatible metrics with PromQL Self-managed Prometheus

What Azure Monitor collects

  • Platform metrics — emitted automatically by Azure resources (VMs, App Service, storage, SQL, AKS) at no ingestion cost.
  • Resource logs — diagnostic logs from Azure resources, routed to Log Analytics via diagnostic settings.
  • Activity log — control-plane events (who did what to which resource).
  • Guest OS telemetry — CPU, memory, disk, and log files from inside VMs, collected by the Azure Monitor Agent through Data Collection Rules.
  • Application telemetry — requests, dependencies, exceptions, and distributed traces from Application Insights SDKs / auto-instrumentation.

Choosing a stack

Neither stack is strictly better — they optimize for different environments.

Consideration Prometheus + Grafana Azure Monitor
Hosting Self-hosted; you run and scale it Fully managed SaaS
Cost model Infrastructure + operational effort (no licensing) Pay per GB ingested / metrics / alert rules
Best for Kubernetes, multi-cloud, on-prem, full control Azure-native workloads, hybrid with Arc, low ops overhead
Query language PromQL KQL (logs) + Metrics Explorer (metrics)
Data model Pull-based scraping, dimensional labels Push/agent collection; metrics + logs stores
Long-term retention Add Thanos/Cortex/VictoriaMetrics Built-in; configurable retention per table/workspace
Vendor lock-in Portable, open standards Azure-coupled (mitigated by Managed Prometheus + Grafana)
Tip

If you already run in Azure, the fastest path to coverage is diagnostic settings + a handful of Azure Alerts — no agents, no servers. Reach for Prometheus when you need Kubernetes-native scraping, multi-cloud portability, or want to avoid per-GB ingestion costs at high volume.

Architecture

 Azure resources ──(platform metrics)──────────────┐
     │                                              ▼
     └──(diagnostic settings)──► Log Analytics ◄── Azure Monitor Agent ◄── VMs / Arc servers
                                   workspace              (Data Collection Rules)
 Application Insights SDK ─────────► (logs + metrics)
                                          │
                    ┌─────────────────────┼──────────────────────┐
                    ▼                     ▼                      ▼
              Azure Alerts          Workbooks /            Azure Managed
              (rules)               Dashboards             Grafana
                    │
                    ▼
              Action Groups ──► email · SMS · webhook · ITSM · Logic App · Function

Quick start

Bring an existing Azure resource under monitoring without deploying anything:

Note

Examples throughout this section show both the az CLI (quick, ad-hoc changes) and Terraform (durable, version-controlled configuration). Terraform snippets use the AzureRM provider and reference a shared azurerm_log_analytics_workspace.core and azurerm_resource_group.monitoring defined once for the stack.

  1. Route resource logs. On the resource, open Diagnostic settings and send logs and metrics to a Log Analytics workspace.

    az monitor diagnostic-settings create \
  --name to-law \
  --resource "$RESOURCE_ID" \
  --workspace "$WORKSPACE_ID" \
  --logs    '[{"categoryGroup":"allLogs","enabled":true}]' \
  --metrics '[{"category":"AllMetrics","enabled":true}]'

    Or as code with Terraform:

    resource "azurerm_monitor_diagnostic_setting" "to_law" {
  name                       = "to-law"
  target_resource_id         = var.resource_id
  log_analytics_workspace_id = azurerm_log_analytics_workspace.core.id

  enabled_log {
    category_group = "allLogs"
  }

  metric {
    category = "AllMetrics"
  }
}

  2. Query the telemetry in Log Analytics with KQL — see the infrastructure KQL examples.

  3. Create an alert on a metric or log signal — see Azure Alerts.

  4. Visualize it in a Workbook or Azure Managed Grafana — see Visualization.

In this section

  • Data Collection — metrics vs logs, the Azure Monitor Agent, Data Collection Rules, Log Analytics, and Application Insights.
  • Azure Alerts — metric, log, and activity-log alert rules, dynamic thresholds, action groups, and alert processing rules.
  • Visualization — Metrics Explorer, Workbooks, Azure dashboards, and Azure Managed Grafana.
  • Monitoring overview — the self-hosted Prometheus + Grafana stack and the section map.
  • Prometheus — the open-source metrics engine, including the managed Prometheus option in Azure.
  • KQL — the query language for Log Analytics and Application Insights, with infrastructure examples.

References

Edit this page