Data entities and the Data Management Framework

The Data Management Framework (DMF) is Dynamics 365 Finance and Supply Chain's tool for bulk import and export. Where Dual-write handles synchronous, transactional sync, DMF handles batch — initial migrations, large recurring loads, and exports for downstream systems.

Data entities

A data entity is a denormalised view of one or more underlying tables, presented as a single shape with stable field names. Microsoft ships hundreds — customer, vendor, item, sales order header, sales order line, sales tax, fixed asset, GL journal voucher, project transaction, and so on. Entities are versioned so consuming integrations don't break when the underlying database evolves.

Data Management workspace

The Data Management workspace in F&O is where users define data projects — collections of entities with mappings, filters, and source format (CSV, XML, Excel, ODBC) — and run import or export jobs. A project can be a one-off load or a recurring scheduled run.

Import

An import project maps incoming source columns to entity fields, validates the data, and either commits in bulk or stages for review. Composite entities import related tables together (e.g. sales order header + lines + charges as one transactional unit). The framework provides error handling: rows that fail validation are quarantined for correction without blocking the rest of the batch.

Export

An export project pulls entity data out as CSV, XML, or to an Azure SQL endpoint. Exports can be filtered, scheduled, and incremental (delta-only based on a change timestamp).

Recurring integrations

The Recurring Integrations API exposes data projects as HTTP endpoints, so external systems can POST import payloads or GET export results on a schedule. The pattern: external app pushes a CSV to the F&O endpoint every hour; F&O queues the import for processing.

The BYOD pattern (deprecated)

Bring Your Own Database exported entity data into a customer-owned Azure SQL Database. BYOD is being phased out in favour of Synapse Link and Microsoft Fabric, which stream F&O tables into a lakehouse without requiring data entities as the export shape.

Performance

DMF jobs can run in parallel threads, configurable per project. Large initial loads (millions of rows) are tuned with parallel processing, narrow batch sizes, and selective indexes. A reasonable initial customer migration loads in hours, not days, when tuned.

Best practice

Use Dual-write for synchronous CRM ↔ ERP sync, DMF for batch ETL and high-volume migrations, and Synapse Link or Fabric for analytics export.