- Data from F&O reaches Fabric through Dataverse. Link to Fabric keeps a copy of the data in Dataverse storage, Azure Synapse Link writes it to your own storage account in Azure.
- Link to Fabric syncs on Dataverse storage and compute, and the copy counts toward Dataverse database capacity. With Synapse Link you pay for storage in Azure and write the data merge in Fabric yourself.
- Export to Azure Data Lake, the earlier export straight from F&O, stops for good on November 30, 2026. Microsoft offers Synapse Link and Link to Fabric in its place.
Contents 7 sections
- 01Microsoft Fabric and Dynamics 365, how the data gets there
- 02Fabric Link for D365 F&O vs Synapse Link
- 03Where are Fabric Link and Synapse Link billed?
- 04How fresh is F&O data in Fabric?
- 05Dynamics 365 Finance in Fabric, entities or tables
- 06Export to Azure Data Lake is ending
- 07FAQ
Microsoft Fabric and Dynamics 365, how the data gets there
Finance and Operations sends data for analytics through Dataverse. You pick F&O tables in Power Apps on the Link data page, the same way as Dataverse tables. Each path sends the bill to a different meter (F&O data in Synapse Link).
Paths from F&O to Fabric
| Path | Where the data lives | What you need for it |
|---|---|---|
| Link to Fabric, also called Fabric Link | a Delta-format copy in Dataverse storage, Fabric points to it with shortcuts | a workspace on a Fabric capacity in the same Azure geography as the Dataverse environment |
| Azure Synapse Link, incremental updates in CSV | timestamped folders in your ADLS Gen2, Fabric reads them through a shortcut | a storage account in Azure, no Synapse workspace and no Spark pool |
| Azure Synapse Link, tables in Delta format | Delta tables in your ADLS Gen2 | a storage account, a Synapse workspace and a Spark pool that computes the conversion to Delta format |
| F&O OData endpoint | wherever you write the data, for example with a notebook or Dataflow Gen2 | your own loading and your own incremental logic |
Neither Dataverse nor Dynamics 365 is among the sources of Mirroring in Fabric (mirroring overview). Microsoft routes Dynamics 365 data into OneLake through Link to Fabric.
Microsoft describes the F&O OData endpoint for real-time integrations over smaller volumes and doesn't recommend it for Power BI reports (F&O integration). OData calls are also subject to service protection limits, and once they're exceeded, F&O rejects requests with error 429 (API limits).
Fabric Link for D365 F&O vs Synapse Link
You turn on Link to Fabric in Power Apps through Link data, New link and Fabric link. It needs no storage account of your own and no Synapse workspace. Dataverse creates a Delta-format copy of the data in its own storage and, in the workspace you choose, creates a lakehouse with shortcuts, a SQL analytics endpoint and a semantic model (Link to Fabric).
F&O tables aren't selected automatically when you create the link, you add them in the wizard or later in Manage tables. The Fabric capacity must be in the same Azure geography as the Dataverse environment. The copy of the data stays in the region of the Dataverse environment (Link to Fabric and Synapse Link).
Azure Synapse Link writes changes to your ADLS Gen2. In incremental update mode it saves CSV files into timestamped folders, one for each interval in which the data changed. You set the interval when you create the link and can change it later in Manage tables, the shortest is 5 minutes (incremental updates in Synapse Link). Fabric points to the folders with a shortcut, and you write the merge into Delta tables yourself, for example with a notebook.
The merge reads only closed folders. The folder Synapse Link is currently writing to is listed in the `Changelog/changelog.info` file. Deleted rows carry the `IsDelete` flag and are empty apart from the system fields, the rest of the record is looked up by `Id`. Labels for the numeric values of F&O enums come in the `GlobalOptionsetMetadata` file (F&O data and metadata).
You can add F&O tables only to a new Synapse Link profile, not to an existing profile with Dataverse tables. The new profile then accepts tables from both sources. Microsoft also lists limitations: long text fields are truncated to 2,000 characters, binary fields and `TimeZoneID` aren't exported, and derived tables don't carry the base table's columns.
Where are Fabric Link and Synapse Link billed?
Link to Fabric syncs on Dataverse storage and compute, not on the Fabric capacity. The Delta-format copy counts toward Dataverse database capacity. In the Power Platform admin center you find it in the environment's capacity details as tables with the `-Analytics` suffix (Fabric Link FAQ). So F&O tables start taking up Dataverse capacity they didn't take up before.
A tenant gets Dataverse database capacity as a default entitlement, and more accrues with user licenses. On top of that you can buy more in 1 GB increments (Dataverse capacity). Only the selected tables use capacity, so the selection in Manage tables also controls consumption.
Azure Synapse Link writes to your storage account, and the sync uses no Fabric capacity. Azure bills storage and transactions in ADLS. For tables in Delta format, add the Spark pool in Synapse on top, and the shorter the conversion interval you set, the more compute it uses.
In Fabric, capacity pays only for the notebooks, queries and models that read and transform the data. Fabric doesn't count transactions for reads through a shortcut to ADLS in CU, Azure bills them directly (OneLake consumption). When the storage account sits in a different region than the capacity, every query pulls data across the region boundary and you pay for transfer between regions (shortcuts and regions).
Who pays for which part
| Meter | Synapse Link to your own ADLS | Link to Fabric |
|---|---|---|
| Fabric capacity | notebooks, queries and models over the data | queries, notebooks and models over the data |
| Dataverse capacity | nothing extra, the data sits in your storage account | a copy of the selected tables in database capacity |
| Azure | storage, transactions, transfer between regions, the Spark pool for Delta format | nothing, you don't need a storage account |
How fresh is F&O data in Fabric?
According to the setup page, Link to Fabric refreshes data in OneLake within 60 minutes, longer at peak times. The newer low-latency engine writes from the Dataverse database straight to Delta Parquet, with no intermediate CSV step (low latency).
For typical CRM and ERP application workloads, Microsoft says most changes arrive within a few minutes and the vast majority within about 10 minutes. Longer delays come with the volume of changes. For 95% of changes, Microsoft gives about 13 minutes for an ERP transaction table, around 47 minutes for hundreds of millions of changes a day and around 90 minutes during bulk invoicing (latency planning).
Microsoft turns on the low-latency engine automatically, one stamp at a time, and there's no separate setting for it. A new link gets it automatically. An existing profile switches over only after you unlink and link again, which starts a full initial sync. The F&O environment must run at least the build the page lists for each version.
Azure Synapse Link publishes changes in intervals. A row reaches Fabric at the earliest once the interval in which it changed has closed and your processing has run again. So two terms set the freshness: the Synapse Link interval and the notebook schedule in Fabric. For Delta format, Microsoft notes that actual freshness can be longer than the conversion interval you set, depending on the Spark pool size and the volume of changes.
We look at how freshness and consumption behaved on a real F&O environment on the blog.
Dynamics 365 Finance in Fabric, entities or tables
Both Synapse Link and Link to Fabric carry F&O tables. For standard F&O tables you don't need to turn on change tracking. For custom tables you turn it on through row version change tracking, otherwise they don't show up in the table list.
Synapse Link can also carry F&O data entities. First you make the entity available in Dataverse as a virtual table, which gets the `mserp_` prefix. Then you turn on change tracking for the entity. The option is in the F&O validation report under System administration > Setup > Row version change tracking, or as Track changes in the table properties in Power Apps. The entity then shows up among the Dataverse tables in Manage tables.
Change tracking can't be turned on for every entity. Entities built for data migration, for example, fail validation. For entities that fail validation, Microsoft points to the EntityUtil solution from the FastTrack team, which assembles the entity's shape from tables (F&O entities in Synapse Link).
Export to Azure Data Lake is ending
Export to Azure Data Lake is an F&O add-in from Lifecycle Services that writes changes as CSV to your data lake. Microsoft deprecated it on November 1, 2024 and will stop it for good on November 30, 2026. After that it no longer exports new or changed rows, and Microsoft won't grant an extension. Files already exported stay in the lake (transition FAQ).
The replacement is Synapse Link or Link to Fabric. Incremental CSV updates from Synapse Link correspond to the Change feeds feature of Export to Data Lake, and Microsoft offers sample tools for adapting existing pipelines (moving from legacy services).
FAQ
Does the sync from Dynamics 365 use Fabric capacity?
Not the sync itself. Link to Fabric runs on Dataverse storage and compute, Synapse Link writes to your storage account. Fabric capacity is used only by the notebooks, queries and semantic models that read and transform the data.
How fast does a change in F&O show up in Fabric?
With Link to Fabric on the low-latency engine, Microsoft says most changes arrive within a few minutes, later with a large volume of changes. For the older engine the stated refresh is within 60 minutes. With Synapse Link, freshness is set by the export interval, 5 minutes at minimum, plus the schedule of your processing in Fabric.
Does Link to Fabric count toward Dataverse storage?
Yes. The Delta-format copy of the selected tables counts toward Dataverse database capacity, and you see it in the Power Platform admin center in the environment's capacity details. Tables you remove from the link stop taking up capacity.
Can we take data entities from F&O, or only raw tables?
Both. You pick F&O tables directly in both Synapse Link and Link to Fabric. For an entity in Synapse Link, you first make it available as a virtual table and turn on change tracking for it, then you find it among the Dataverse tables with the `mserp_` prefix.
What if we use Export to Data Lake today?
Export to Azure Data Lake stops for good on November 30, 2026, and Microsoft won't grant an extension. The move is to Azure Synapse Link or Link to Fabric. Files already exported stay in your lake.
Where do you stand?
Environment and Cost Audit
€2,100excl. VAT
We go through your path from Dynamics 365 to Fabric, read the actual meters in Fabric, Dataverse and Azure, and tell you where you pay and how much. The audit can be followed by a Fabric implementation at your company.
