Fabric Capacity Planning: How to Size Your OneLake Workload

Fabric Capacity Planning

Microsoft Fabric capacity planning determines how much compute and storage your organization needs — and what it costs. Fabric uses capacity units (CUs) measured in F-SKUs: F2 ($262/month) for dev/test, F64 ($17K/month) for production analytics, F128+ ($34K+/month) for enterprise-scale. The sizing factors: data volume (OneLake storage is $0.023/GB/month), concurrent users (each DirectLake query consumes CUs), Spark job complexity (notebook execution burns CUs fast), and pipeline frequency. Under-provisioning = query timeouts and pipeline failures. Over-provisioning = wasted budget. See: Fabric Consulting and Fabric Demand Report'.

How It Works in Practice

Enterprise implementations follow structured progression: assessment (2-4 weeks), design (2-4 weeks), build (4-12 weeks), stabilization (2-4 weeks). Total: 10-24 weeks. Organizations that skip assessment spend 40-60% more due to mid-build misalignment.

What Does This Cost?

Consulting rates: $120-350/hr. Typical engagement: $75K-300K over 8-20 weeks. Through Xylity, rates are 20-35% below traditional consulting firms — 4.3-day deployment, 92% first-match acceptance rate. Related: Data Engineering Consulting Cost.

How Do You Get Started?

Start with a 2-week paid assessment ($15K-30K) that tells you exactly what to build, how long it takes, and what it costs. Then deploy pre-qualified specialists through Xylity — 4.3 days to first profile, 92% acceptance rate, 200+ delivery partners across 20+ domains and 22 industry verticals. Related: The True Cost of a Vacant Seat — every week of delay costs $12K-24K.

Go Deeper

Continue building your understanding.