Dataverse to Fabric: Link to Fabric or Azure Synapse Link

There are two built-in ways to get Dataverse data into analytics: Link to Fabric, which makes your tables available in Microsoft Fabric without you running any storage, and Azure Synapse Link for Dataverse, which exports them to an Azure storage account you own. The choice matters more now, because Azure Synapse Link's Delta Lake (Parquet) export stops being available to new customers on 15 October 2026. For most teams starting today, Link to Fabric is the answer; Azure Synapse Link is for when the data has to sit in your own storage.

The difference at a glance

Both start from the same place: Link data in Power Apps, where Fabric links and the other links (Azure Synapse Link, now labelled Link to Data lake) sit side by side.

Choose Link to Fabric when

• You want Dataverse data in Fabric or Power BI without building anything. The link creates a lakehouse, a SQL analytics endpoint and a Power BI semantic model, and keeps them up to date as Dataverse changes.
• You'd rather not run Azure storage or pipelines. There's no storage account, Synapse workspace or Spark pool to provision.
• You already have a Fabric capacity, or Power BI Premium, in the same Azure region as the environment. A free Fabric trial capacity works for trying it.
• You're coming from the Delta Lake export. Microsoft names Link to Fabric as the way to keep Delta Parquet output, with lower sync delay.

Choose Azure Synapse Link when

• The data must land in storage you own, for example because other tools or teams read it from your data lake, or your policies require it.
• You need tight control over which tables go out and where the files are kept.
• CSV files are what your downstream jobs expect. CSV output stays available.

One trap: an existing Azure Synapse Link profile that writes CSV can't be connected to Fabric later. Only profiles with Delta Parquet output can, and not those secured with a managed identity.

The Delta Lake export is closing

Azure Synapse Link can convert its export to Delta Lake format with a Spark pool in your Synapse workspace (also called Synapse Parquet or bring-your-own-Synapse Parquet). Microsoft has deprecated that option:

• From 15 October 2026, it isn't available to new customers.
• Existing customers must move by December 2027, either to CSV output in Azure Synapse Link or to Link to Fabric. After that the option is gone.
• Eligibility is per environment, not per tenant. An existing environment that used it can create such a link again through an opt-in address on the Link data page. A new environment can't, even in the same tenant, and the opt-in doesn't move the December 2027 deadline.

If you use it today, list what reads the deltalake folder (Synapse views, notebooks, Power BI models) and plan the move to Link to Fabric before the deadline, not after a failed refresh.

Set up Link to Fabric

You need the System Administrator role in the environment, admin rights on the Fabric workspace, and a capacity in the same region. Your Fabric admin must also allow users to create Fabric items and workspaces.

1. In Power Apps, open Link data, select + New link, then Link data via Fabric. (Or, from Tables, select Analyze > Analyze in Fabric.)
2. Getting started checks your capacity and prerequisites.
3. Setup configuration: choose the workspace and create the connection. Microsoft recommends workspace identity, which avoids managing secrets; the identity must also be added to the environment as an application user with a suitable role.
4. Select tables: clear any you don't need. Only the selected tables use storage.
5. Review and create, then Finish.

The first sync can take a while for large tables, and later changes can take up to an hour to show in OneLake. A faster, low-latency sync engine is rolling out gradually; new links use it automatically once it's available for your environment.

Things that catch people out

• It uses Dataverse storage. The copy in Delta Parquet counts against your environment's database capacity; check it in the Power Platform admin center.
• More than 2,000 active tables in the environment can make the link fail.
• Don't build on the lakehouse's name. Microsoft says its naming pattern can change; refer to it by workspace and lakehouse ID.
• Unlinking deletes the lakehouse, and it fails while other Fabric items depend on it. Relinking (for example to move an older link to low-latency sync) runs a full initial sync again.
• Timestamps change type with low-latency sync, from INT96 to INT64. Check queries and pipelines that read them.

Sources

• Link your Dataverse environment to Microsoft Fabric, Microsoft Learn (the comparison with Azure Synapse Link)
• Link to Microsoft Fabric, Microsoft Learn (prerequisites, set-up and low-latency sync)
• Export Dataverse data in Delta Lake format, Microsoft Learn (the deprecation dates and eligibility)