ERP Integration Patterns

As we architect our D365 implementation, integration patterns are top of mind. We're keeping our BFC Dakota WMS, our Adobe Commerce B2B platform, our Shopify Plus B2C site, and our Paycom HR system. All of these need to talk to D365 reliably.

Real-Time vs. Batch

The first question for any integration: does it need to be real-time, or can it batch? Real-time is harder and more expensive to build and maintain. Batch is more resilient to temporary failures. Choose wisely.

For us: Order entry from e-commerce needs near-real-time integration - customers expect immediate confirmation and inventory visibility. But inventory sync from WMS to e-commerce can batch every 15 minutes without customer impact. Financial data can batch nightly for reporting purposes.

D365 Integration Options

D365 F&O offers several integration approaches:

Data Management Framework (DMF): Excel-like import/export for batch data. Good for migrations and periodic syncs. Not designed for real-time scenarios.

Data Entities: REST APIs that expose D365 data and operations. This is the primary integration surface for most scenarios. Well-documented, reasonable performance, supports both query and transaction operations.

Dual-write: Real-time sync between D365 F&O and Dataverse. Useful if you're building Power Platform applications that need F&O data. Adds complexity but enables interesting scenarios.

Virtual Entities: Query external data from within D365 without copying it. Good for reference data scenarios where you don't want duplication.

Our Architecture

BFC Dakota WMS integration: Bidirectional, near-real-time. Orders flow from D365 to WMS for fulfillment. Shipment confirmations flow back. Inventory adjustments sync both directions. We're using the Data Entity APIs with a middleware layer for transformation and error handling.

Adobe Commerce integration: Customer and product master data from D365. Order capture from Commerce to D365. Pricing and inventory visibility from D365 to Commerce. Mix of real-time (pricing lookups) and batch (catalog sync) patterns.

Paycom integration: Employee master data flows to D365 for expense approvals and purchasing authorization. Batch sync, nightly, low complexity but critical for access control.

Integration Middleware

We're using Azure Integration Services (Logic Apps, Service Bus, API Management) as our middleware layer. This provides transformation capability, error handling, retry logic, and monitoring without building everything custom.

The middleware layer also provides a buffer between systems. If D365 is down for maintenance, the middleware queues messages and processes them when the system returns. Better than direct point-to-point integrations that fail when any endpoint is unavailable.

Lessons Learned

Integration is where ERP implementations often struggle. Budget time for it. Test thoroughly with production-like data volumes. Plan for failure scenarios - what happens when a message doesn't process? How do you recover?

Our partner's integration experience was a key selection factor. They've done this before with similar system combinations.