FMD Framework for Microsoft Fabric Explained: A Video Guide by Matthias Falland

FMD Framework for Microsoft Fabric Explained: A Video Guide by Matthias Falland

Fabric Metadata‑Driven Framework (FMD)

Erwin

by Erwin | Nov 19, 2026

Table of Contents

"From the architecture to your first table — and when a different approach may be the better fit."

A practical introduction to FMD, from someone else’s perspective

Explaining a framework you build yourself is one thing. Seeing someone else unpack the ideas, follow the data and ask where it fits is something different. Matthias Falland has created a video series about my Fabric Metadata-Driven Framework (FMD), covering the architecture, the loading process, onboarding a table and the decision between FMD and Fabric Copy job.

A big thank-you to Matthias for putting this material together. If you have been curious about FMD but would rather start with a guided explanation than dive straight into the repository, these videos give you a useful route into the framework.

 

Start here: what is FMD?

The starting point is a familiar data engineering problem: loading another table should not mean creating and maintaining another near-identical pipeline. FMD takes a metadata-driven approach. You describe what needs to be loaded in a configuration database, and reusable pipelines and notebooks carry out the work.

The first video introduces that idea through the core loading journey: source data arrives as files in the Landing Zone, Bronze represents the current state, and Silver maintains history. It also draws an important boundary: the ingestion journey covered here ends at Silver. Your business-specific Gold models, metrics and reporting logic remain yours to design.

That distinction matters. A reusable loading framework should give you a consistent foundation without pretending it knows how your organisation defines revenue, customers or business performance.

 

Follow one table from Landing to Silver

Architecture becomes easier to understand when you follow a concrete example. In the second video, Matthias follows an Orders table through the layers, showing how the orchestrator runs Landing, Bronze and Silver in sequence.

The useful detail is not just where the data ends up, but how the work is coordinated. Views turn configuration into the queries needed for execution, while queues connect the processing stages. The video also highlights the ordering of the queue step and watermark update: the watermark is stored after the queue step.

For anyone assessing the framework, this is the video to watch when you want to move beyond the architecture diagram and understand the mechanics of a load.

 

Add a table through configuration, not another pipeline

The third video brings the metadata-driven approach into focus. Rather than editing a pipeline for each new source table, you register the table in the configuration database. The example covers a connection, a data source and the entity registrations for Landing, Bronze and Silver.

Matthias explains the separation between the database schemas: integration holds the definitions you configure; execution determines the work to perform; logging records what happened. Credentials stay in Fabric rather than becoming part of those configuration rows.

The practical takeaway is simple: onboarding becomes configuration work. That does not make configuration unimportant. Treat it with the same care as code, because those rows determine what the framework will execute.

FMD or Fabric Copy job? Start with the requirement

I am glad this series also includes a discussion about when not to use FMD. Choosing an approach should start with your requirements, not with a preference for a particular framework.

In the final video, Matthias makes the case for considering Fabric Copy job first where its supported sources and change-handling capabilities meet your needs. He positions FMD as an alternative when you want a configuration-based table registry and history-building patterns without relying on source CDC.

The limitations deserve equal attention. The incremental loading approach discussed cannot detect source deletes simply from a watermark, and FMD is a framework you deploy and operate yourself. History in Silver should not be mistaken for a guarantee that every source-side change has been captured.

Check the behaviour you need for your actual source, especially deletes and historical accuracy. That is a more useful starting point than asking which option wins in every situation.

 

Thank you, Matthias — and over to you

Open source is not only about publishing code. Clear explanations help people understand the design, challenge assumptions and decide whether a project fits their own platform. Thank you, Matthias, for making FMD more approachable through this series.

If you are new to FMD, start with the overview, follow the Orders example, and then look at table onboarding. Finish with the Copy job comparison before deciding what belongs in your own architecture.

FMD is available as an open-source project under the MIT licence. Explore the repository, try it against your own requirements, and share your questions, issues and contributions. What would you like to see explained next?

New to the Microsoft Fabric Metadata Driven Framework?

Learn more about the framework architecture, deployment model, and features here:

The Microsoft Fabric Metadata Driven Framework continues to help organizations standardize data platform delivery, automate Microsoft Fabric deployments, improve governance, and accelerate AI-ready architectures. With every release, the framework becomes more reliable, secure, and easier to adopt for enterprise-scale Microsoft Fabric implementations.

GitHub Repository
FMD_FRAMEWORK

Hit Subscribe to get future release notes delivered straight to your inbox!

Feel free to leave a comment

Reassigning Microsoft Fabric Capacities in Bulk: Don’t Let an Expiring Trial Catch You by Surprise

Reassigning Microsoft Fabric Capacities in Bulk: Don’t Let an Expiring Trial Catch You by Surprise

Erwin

by Erwin | Jul 14, 2026

Table of Contents

Avoid Losing Fabric Items When a Trial Expires: Bulk Reassign Workspaces Between Capacities Using Fabric CLI

With the summer holiday season approaching, many organizations are running Microsoft Fabric workloads on trial capacities that were originally created for proof-of-concepts, workshops, or development activities. What often gets overlooked is that Fabric trial capacities eventually expire, and when they do, the consequences can be significant. According to Microsoft Fabric community guidance, you have only seven days after trial expiration to reassign your workspaces to an active capacity. After that retention period, non-Power BI Fabric items are removed according to the retention policy.

For a tenant with a handful of workspaces, manually reassigning capacities might be manageable. But what if you need to migrate dozens or even hundreds of workspaces from Capacity A to Capacity B?
Fortunately, there is an easy way to automate this process.

Why Reassign Capacities?

There are several common scenarios:

  • A Fabric Trial capacity is about to expire.
  • You purchased a new Fabric capacity and want to move workloads.
  • You're consolidating capacities to reduce costs.
  • You're separating development, test, and production workloads.
  • You're moving workspaces to a larger SKU due to increased usage.

Whatever the reason, workspace reassignment is a straightforward administrative task that can be automated through the Fabric CLI and Fabric APIs.

The Hidden Risk of Trial Capacities

Many organizations start their Fabric journey on a trial capacity. It works great for evaluations, hackathons, and pilot projects. However, when the trial expires, the clock starts ticking.
Fabric Trail Expire
Microsoft states that after trial expiration, you have a seven-day window to save your Fabric content by assigning the affected workspaces to an active Fabric capacity. If you don't migrate the workspaces assigned to your Fabric trial capacity before your free Fabric trial capacity expires, the workspaces will be moved to Power BI with license mode Pro and you will not be able to use Fabric (non-Power BI) items. Fabric items not moved to a Power BI Premium or Fabric capacity within 7 days after your trial ends may be permanently deleted. If you have Power BI items, they will continue to exist in the workspace. Items that you created in other workspaces not assigned to your Fabric trial capacity will be unaffected.

I've seen customers develop:

  • Lakehouses
  • Notebooks
  • Data Pipelines
  • Dataflows Gen2
  • Semantic Models
  • Warehouse solutions

...only to discover during a holiday period that their trial is expiring.

My advice: don't wait until the last day.

Bulk Migration with Fabric CLI

Instead of opening every workspace manually, you can automate the entire process.

The approach is simple:

  1. Enumerate all Fabric workspaces.
  2. Identify the current capacity.
  3. Optionally filter specific workspaces.
  4. Assign them to the target capacity.
  5. Create a migration report.
  6. Helps you avoid manual workspace-by-workspace administration

tasks = []

filter_set = (
    {w.lower() for w in filter_workspaces}
    if filter_workspaces
    else None
)

workspaces = get_workspaces()

for ws in workspaces["result"]["data"]:

    workspace_name = ws.get("name", "").removesuffix(".Workspace")
    capacity_name = ws.get("capacityName", "")

    # Optional workspace filter
    if filter_set and workspace_name.lower() not in filter_set:
        continue

    # Skip unassigned workspaces
    if capacity_name == "N/A":
        continue

    if assign_capacity_to_all_fabric_workspaces:

        if capacity_name != target_capacity:

            print(f"Moving {workspace_name}  from {capacity_name} to {target_capacity}" )
            result = assign_capacity(workspace_name,target_capacity )
            tasks.append({"Workspace": workspace_name,"Source Capacity": capacity_name,"Target Capacity": target_capacity,"Remarks": result})

        else:
            print(f"{workspace_name} no action needed, capacity already assigned")
            tasks.append({"Workspace": workspace_name,"Source Capacity": capacity_name,"Target Capacity": target_capacity,"Remarks": "Capacity already assigned"})
 
    else:

        if capacity_name == source_capacity:
                        
            print(f"Moving {workspace_name}  from {source_capacity} to {target_capacity}" )
            result = assign_capacity(workspace_name,target_capacity )
            tasks.append({"Workspace": workspace_name,"Source Capacity": capacity_name,"Target Capacity": target_capacity,"Remarks": result})

        else:

            print(f"{workspace_name} skipped current capacity: {capacity_name}" )
            tasks.append({"Workspace": workspace_name,"Source Capacity": capacity_name,"Target Capacity": target_capacity,"Remarks": "Skipped - not on source capacity"})

Check out the script on GitHub and adapt it to your environment:

Check out my GitHub repository for the complete migration script.

Fabric_Tooling/NB_WORKSPACE_CHANGE_CAPACITY

Whether you're cleaning up trial capacities, consolidating environments or moving to a larger Fabric SKU, automation can save a lot of time and reduce the risk of missing a critical workspace.

Why Automation Matters

I've noticed many Fabric environments that started as small experiments and quickly grew into business-critical platforms.

A workspace that started as:

"Let's just test Fabric for a few days"

Can suddenly contain:

  • Production pipelines
  • Business reporting
  • Data engineering workloads
  • Governance processes

When a trial capacity approaches expiration, manually managing dozens of workspaces becomes error-prone.

Automating the migration process provides:

  • Repeatability
  • Traceability
  • Faster execution
  • Lower risk
  • Full reporting

Final Thoughts

If you're using Microsoft Fabric Trial capacities today, take a few minutes to check:

  • Which workspaces are assigned to the trial.
  • When the capacity expires.
  • Whether a long holiday period overlaps with the expiration date.

Remember that Microsoft provides only a limited grace period after trial expiration to save your Fabric items by moving the workspace to an active capacity.

The good news is that moving workspaces from Capacity A to Capacity B can be fully automated. A simple Fabric CLI or REST API script can handle the migration in minutes and help you avoid an unpleasant surprise when you return from vacation.

Tip: Even if you're not currently running a trial, keeping an automated capacity migration script in your Fabric toolbox is a great operational practice. Sooner or later, you'll need it for a capacity upgrade, consolidation, or tenant migration.

Feel free to leave a comment