Microsoft Fabric and Dynamics 365

Microsoft Fabric is Microsoft's unified analytics platform — a single SaaS product covering lakehouse, data warehouse, real-time analytics, data science, and Power BI on top of a shared storage layer called OneLake. For Dynamics 365 customers, Fabric is becoming the destination for analytics that need cross-system data, large historical volumes, or complex transformations beyond what Power BI's own model can handle.

Synapse Link for Dataverse

A managed near-real-time pipe from Dataverse into a Fabric lakehouse (or, historically, a customer-owned Azure Data Lake). When enabled per environment, Dataverse continuously writes Delta-Parquet snapshots of selected tables to OneLake. Downstream Fabric workloads — notebooks, SQL endpoints, Power BI semantic models — read from OneLake without touching Dataverse.

Synapse Link for F&O

The equivalent for Finance and Supply Chain. Data entities or specific tables stream into OneLake. Replaces the older Export to Data Lake feature in F&O and the Bring Your Own Database (BYOD) pattern, both of which Microsoft is deprecating.

Business Central data lake export

Business Central also publishes its data into Fabric via a similar mechanism (still evolving as a generally-available feature). The end state is a single OneLake estate containing CRM (Dataverse), F&O, and BC tables side by side, queryable as one logical data set.

DirectLake

A new Power BI mode that reads parquet directly from OneLake without import or DirectQuery. Performance close to import, no refresh schedules. The right pattern for most Dynamics 365 reporting at scale.

Use cases.

• Cross-system reports. Sales pipeline (Dataverse) joined to finance actuals (F&O) joined to web analytics (Adobe Analytics) — all in one Fabric lakehouse, with one semantic model in Power BI.
• History beyond the source. Dataverse and F&O retain data per their own retention policies; OneLake retention is independent and cheap.
• Data science. Train ML models against the lakehouse rather than hitting the live ERP.
• Real-time analytics. Stream from operational events into Fabric's KQL database for dashboards.

Operational reality

Fabric capacity is sized in F-SKUs (or Power BI Premium capacity). Plan capacity for the actual workload — peak Power BI users, lakehouse compute, semantic model size. Capacity oversubscription is the most common Fabric pain point.

Governance

Microsoft Purview integrates with Fabric for cataloguing, classification, and lineage. Use it from day one if compliance matters.

Choosing what feeds Fabric, and how

Source	Mechanism	Latency	Best for
Dataverse	Synapse Link for Dataverse	Near-real-time (minutes)	Sales, service, marketing data joined with other sources
Finance and SCM	Synapse Link for F&O	Near-real-time	Financial and supply-chain reporting at scale, replacing BYOD
Business Central	Data lake export	Scheduled / evolving to near-real-time	SMB customers wanting the same OneLake estate as their CRM/ERP siblings
Any REST/OData source	Fabric Data Factory pipelines	Batch, scheduled	Third-party systems with no native Fabric connector

The right first move for most Dynamics 365 customers is Synapse Link for whichever app holds the data a specific report needs — not an all-at-once Fabric migration. Turn on the pipe for the tables a concrete reporting need requires, validate the resulting lakehouse model, then expand.

Cost shape

Fabric capacity (F-SKUs) is billed by compute-seconds consumed across every workload sharing that capacity — lakehouse ingestion, notebook runs, DirectLake queries, and Power BI refreshes all draw from the same pool. This is different from classic Power BI Premium, where BI was the only consumer. A Fabric estate that also runs notebooks and Data Factory pipelines needs capacity planning that accounts for all of them, not just the reports; teams that size capacity purely off "how many Power BI users" routinely under-provision once data engineering workloads land on the same capacity.

Migrating off Export to Data Lake / BYOD

Customers still running the legacy Export to Data Lake (F&O) or Bring Your Own Database patterns should treat the Fabric migration as a project with its own testing cycle, not a flag flip: semantic models built against the old schema (BYOD's SQL views, or Data Lake's folder structure) typically need rebuilding against the OneLake/Delta-Parquet shape, and any downstream consumer — a Power BI report, a data warehouse load, a partner integration — needs to be repointed and validated before the legacy pipe is switched off. Run both in parallel for at least one full reporting cycle before decommissioning the old export.

Where it stops

Highly specialised analytics (statistical forecasting, complex MDX cubes, certain GIS workloads) sometimes still need a separate purpose-built platform. For mainstream BI and data integration on Dynamics 365, Fabric is now the recommended home. Organisations already committed to a different lakehouse platform still commonly feed it from F&O rather than migrating wholesale — see F&O with Databricks and F&O with Snowflake for that pattern.

Where to go next

The feeds into Fabric each have a guide: Synapse Link for Dataverse, Fabric Link for Dataverse, and data lake export for Finance. On top sits Power BI for Dynamics 365. Where a lakehouse ends and a customer data platform begins is CDP vs data warehouse.