Skip to main content

create-your-initial-capacity

Create your initial Fabric Capacity​

When getting started with Microsoft Fabric, you need to provision your initial Capacity—the foundational unit that enables you to run Fabric experiences like Lakehouses, Warehouses, Real-Time Analytics, and Data Activator. A well-structured initial capacity strategy lays the groundwork for scalable, secure, and cost-efficient adoption.

Start with capacity policies​

Before creating your first capacity, define how you intend to manage Fabric capacities across your tenant. Consider:

  • Workspace Separation: Use separate capacities (and workspaces) for Production, Development/Test, and Sandbox scenarios to maintain isolation, reduce risk, and manage cost attribution.
  • Data Governance Domains: Align capacities with business domains following Data Mesh principles to encourage ownership and autonomy.
  • SKU Selection: Choose the appropriate SKU (F2–F2048) depending on your concurrency needs, cost targets, and workload profile.

Avoid noisy neighbors with multi-capacity strategies​

Fabric resources in the same capacity compete for CPU and memory. To prevent noisy neighbor issues (e.g. when intensive Lakehouse jobs affect Real-Time dashboards):

  • Allocate high-intensity workloads to dedicated capacities
  • Assign Real-Time Intelligence workloads to their own capacity for predictable performance
  • Use cost centers or budgets to map team/business-unit-specific capacities

Create your initial capacity​

We recommend starting with three core capacities:

  1. Production Capacity: For certified reports, data pipelines, and workloads with business-critical SLAs.
  2. Non-Production Capacity: For development, integration, and testing of Fabric assets.
  3. Sandbox Capacity: For experimentation, innovation, training, and evaluation of new features.

You can create capacities in the Microsoft 365 Admin Center or using the Power BI Admin APIs.

Azure Subscription requirements​

Before provisioning a Fabric Capacity, ensure you have access to an Azure Subscription. Microsoft Fabric capacities are provisioned within the Power BI Service, but are logically tied to the Azure resource hierarchy. This relationship is important for governance, cost management, and organizational alignment.

Here's how the hierarchy is structured:

This model allows you to align Fabric workloads with your enterprise governance strategy, including tagging, access control, and budget management across Azure and Microsoft 365.

Automate provisioning and governance​

Use scripting (PowerShell, REST APIs, or Azure CLI) to automate:

  • Capacity creation
  • Workspace assignment
  • Resource tagging
  • Budget alerts

Maintain workload isolation, automated governance, and consistency by managing Fabric as code.

Monitor usage and scale​

Use the Fabric Capacity Metrics App to monitor utilization. Track:

  • CPU and memory saturation
  • Dataset refresh failures
  • Query performance
  • Resource usage by workspace or domain

Proactively plan to scale up (SKU tier) or out (additional capacities) based on demand.

This addition brings the Fabric setup guidance into alignment with the Cloud Adoption Framework’s subscription strategy while addressing Fabric-specific architecture patterns like multi-capacity usage and performance isolation.

Understand the dependency on the Power BI Service​

Microsoft Fabric is built on top of the Power BI Service and fully integrates into its architecture. All Fabric artifacts—including Lakehouses, Warehouses, Pipelines, Real-Time Analytics, and Reports—are managed inside Power BI Workspaces.

This means that Fabric leverages Power BI’s:

  • Workspace model for organizing content
  • Role-based access control (RBAC) for managing permissions
  • Tenant-level and workspace-level settings from the Power BI Admin Portal
  • Licensing and identity integration with Microsoft Entra ID

Understanding this dependency is crucial for setting up roles, permissions, and governance. We will explore this further in the roles and access control sections of the Fabric Adoption Framework.

Contributors