Skip to main content

Inventory and Visibility Considerations

As your organization designs and implements your Microsoft Fabric-based cloud environment, robust inventory and visibility practices are critical for stable operations, governance, and business continuity. This design area defines guidance for telemetry, observability, and access models to achieve consistent, actionable insights across the platform and workload layers.

Motivation

The cloud operating model decisions made early in a landing zone architecture—such as centralized vs. decentralized operations—directly influence your inventory and monitoring strategy. Microsoft Fabric introduces new resource types, metrics, and logs that must be accounted for in platform operations.

Inventory Considerations

Inventory focuses on collecting, storing, and managing metadata, metrics, and logs. Key aspects include:

  • Central Log Analytics Workspace: Use a centralized Log Analytics workspace for platform-level telemetry collection (Fabric Capacities, Workspaces, Pipelines, Dataflows).
  • Resource Inventory Tools:
    • Microsoft Purview (for structured data inventory)
    • Azure Resource Graph (for subscription-wide queries)
    • Azure Resource Manager (for resource and tag tracking)
  • Log Types to Ingest:
    • Fabric Activity Logs (Workspace operations, Pipeline errors)
    • Capacity Metrics (utilization, throttling)
    • Microsoft Entra Audit Logs
    • Azure Diagnostic settings from connected services (e.g., Key Vault, Storage)
  • Retention Requirements:
    • Fabric Activity Logs: 90 days default
    • Azure Monitor Logs: Up to 2 years analytic, 7 years archive
    • Export long-term logs to immutable storage

Visibility Considerations

Visibility defines how insights are surfaced to stakeholders and integrated with operational processes:

  • Alert Routing:
    • Use Azure Monitor action groups to notify specific platform or workload teams.
    • Segment action groups by Fabric Capacity or Subscription.
  • Dashboards:
    • Use Azure Workbooks to create customized dashboards for Fabric observability.
    • Monitor workspace refreshes, error rates, refresh durations, and success ratios.
    • Publish dashboards via Power BI for non-technical audiences.
  • Governance Integration:
    • Use Azure Policy to audit diagnostic settings and enforce logging compliance.
    • Apply policies that ensure Fabric Capacity and Workspace diagnostics are sent to a Log Analytics workspace.
  • Security Monitoring:
    • Ingest Entra ID sign-in logs and Defender for Cloud signals.
    • Use Microsoft Sentinel for behavioral analysis and alert correlation across services.

Advanced Visibility Scenarios

ScenarioApproach
Application team logs without access to central workspaceDeploy dedicated Log Analytics workspace
Long-term auditability (7+ years)Export logs to Azure Storage with immutable retention
Alert noise reductionUse Alert processing rules to suppress non-critical signals
Fabric Capacity resource-level accessEnable Resource-Centric RBAC in Log Analytics

Platform Recommendations

  • Use resource locks on shared platform components (e.g., central Log Analytics workspace).
  • Implement deny policies alongside RBAC to prevent misconfiguration.
  • Avoid exporting raw logs to on-prem systems. Use alerting and dashboards instead.
  • Test alert policies in staging before production rollout.
  • Treat Fabric as a monitored application platform: include refresh pipelines, data activations, semantic model deployments, and workspace lifecycle events in your monitoring strategy.

Azure Landing Zone Accelerator Integration

When using the Azure Landing Zone accelerator, the following monitoring configurations are deployed automatically:

  • Central Log Analytics Workspace
  • Diagnostic settings for Fabric and other services
  • Microsoft Defender for Cloud enabled
  • Baseline alerts and action groups

This model supports shared platform telemetry while enabling application teams to route or duplicate their diagnostics into their own workspace if needed.


Contributors