Choosing the right Microsoft Fabric architecture for your business size
With Microsoft Fabric, one size doesn’t fit all. How you lay out workspaces, deployment pipelines and capacity has a big effect on cost, performance and what your report users experience.
Here’s what we recommend, depending on the size of your business.
Small businesses — keep it lean
If you’re a small business just getting started with Fabric, keep things lean.
- One Fabric workspace for all the ETL: dataflows, lakehouses, pipelines and the rest.
- Power BI Pro workspaces for the semantic models and reports in production.
Why this works: only the data movement runs on Fabric capacity. Your production reports run in Pro workspaces, so you can stay on a small SKU and still deliver fast reports.
Medium businesses — add deployment pipelines for control
As you grow, so does the need for structured change management.
- Keep a single ETL workspace for simplicity.
- Move the reporting workspaces onto the Fabric capacity too. Deployment pipelines only work on workspaces backed by a capacity or Premium Per User, not on Pro workspaces.
- Set up two deployment pipelines:
- An architecture pipeline, at least Dev → Prod, for ETL and data model changes.
- A reporting pipeline to match, so report changes are tested before they reach production.
Why this works: you decide when changes reach your users. Report queries now use some of the capacity, so keep an eye on utilization, but a modest SKU usually covers it. Below F64, everyone who views a report still needs a Pro license.
Large organizations — full Fabric for scale
For an enterprise-scale analytics environment:
- F64 capacity or above. From F64, people with a free license can view reports, so you no longer buy a Pro license for every viewer. With a large audience, that is usually what makes F64 pay off.
- Every workspace on Fabric capacity.
- Deployment pipelines and Git integration to manage application lifecycle management (ALM) for both architecture and reporting.
Why this works: one platform carries everything from raw data to reports, and governance and automation keep the environment from turning chaotic as it grows.
A note on Copilot
You don’t need F64 for Copilot. Copilot in Fabric runs on any paid capacity from F2 up, so it isn’t a reason on its own to size up. Trial capacities don’t support it.
The bottom line
Small businesses should plan for cost, medium businesses for governance and large organizations for scale. Whatever your size, starting with the right architecture saves you headaches — and money — down the road.