Understanding Microsoft Fabric Capacity SKUs

Microsoft Fabric provides a variety of capacity SKUs (Stock Keeping Units) to meet diverse performance needs and budget limitations. Selecting the appropriate SKU is essential for maximizing cost-effectiveness and ensuring that your analytics workloads run efficiently. This blog will explore the specifics of Fabric capacity SKUs to assist you in making knowledgeable choices.

What is a Capacity?

Capacity refers to a dedicated pool of resources, measurable in Capacity Units (CU), within the Microsoft Azure environment. This pool may comprise various resources such as CPU and memory, providing the necessary computing power to process Fabric services.

What are Fabric Capacity SKUs?

Fabric capacity SKUs represent predefined configurations of computational resources that define the processing power, memory, and storage capabilities within a Microsoft Fabric environment. These SKUs cater to a range of data processing requirements, from modest analytics to large-scale big data operations.

Understanding the Fabric Capacity SKUs

Resources are categorized into various levels, each distinguished by different SKUs.

Microsoft Fabric F SKUs

The CU can be used to understand the computing power of each SKU. For example, F32 has twice the power of F16 and is half of the power of F64.

The prices listed in the tables are in USD, and the F SKUs are not bound by annual contracts. They can be activated or deactivated at any time. In the coming months, there will be an announcement regarding annual licenses (RI) that will offer reduced pricing due to their minimum annual commitment.

For any SKUs below F64, individual Power BI licenses (Pro or PPU) are necessary for sharing Power BI content.

Storage Cost

OneLake’s storage costs are not included in the capacity licensing. You will need to pay for that separately. The OneLake cost is similar to ADLS (Azure Data Lake Storage) pricing. For example, US West 2 storage costs $0.023 per GB monthly.

Choosing the Right Fabric Capacity SKU

When selecting a Fabric capacity SKU, consider the following factors:

• Workload Requirements: Evaluate the nature and scale of your analytics workloads. Determine whether your applications require more computational power, memory capacity, or a balanced mix of both.
• Performance Expectations: Assess the expected performance metrics such as query speed, data processing throughput, and concurrency levels. Choose a SKU that aligns with your performance expectations without over-provisioning resources.
• Cost Efficiency: Balance performance requirements with cost considerations. Opt for the SKU that offers the best value for your budget while meeting your operational needs. Consider using Azure Cost Management tools to monitor and optimize spending.
• Scalability: Ensure the chosen SKU allows for scalability as your data and analytics requirements grow over time. Fabric capacity SKUs should be scalable both vertically (upgrading to higher SKUs) and horizontally (adding more instances if needed).

The optimal way to gauge its effectiveness is by hands-on engagement with your data. Start by provisioning an F2 capacity, establish a Lakehouse or Warehouses, populate them using Data Pipelines and Dataflows, and proceed to create additional objects. Should the entire process take up to three hours, the cost would be approximately $1. This nominal investment allows you to assess its performance capabilities directly. I am confident that you will be impressed with the performance, even on an F2 capacity.

For Provisioning and Managing Fabric Capacity SKUs, here is the link to my blog - Getting Started with Microsoft Fabric: Creating a Fabric Capacity and Assigning it to a Workspace