Planning Optimization in Dynamics 365 Supply Chain

Planning Optimization is the modern master-planning engine in Dynamics 365 Supply Chain Management. It replaced the legacy AX-era MRP engine that struggled with large data volumes and long-running planning runs. The new engine is in-memory, parallelised, and dramatically faster — but it has scope differences worth knowing.

Architecture

Planning Optimization runs as a separate Microsoft-managed Azure service, not in the F&O application tier. F&O streams the planning-relevant data (items, BOMs, routings, on-hand, open orders, forecasts, parameters) to the service; the service computes; the results flow back as planned orders, action messages, and reschedules. Because the engine is in-memory and runs in parallel, it can process plans for hundreds of sites and millions of items in minutes rather than hours.

Net change and regenerative

The service supports both net change plans (only items affected by a recent transaction are replanned) and regenerative plans (everything from scratch). Net change makes near-real-time MRP feasible for very large catalogues.

Filtering

Plans can be filtered by site, product, or any combination — useful for running a fast plan on a single site while keeping the global plan unchanged.

Forecast consumption

Forecast plans feed Planning Optimization through forecast time fences and consumption rules. The engine honours both global and per-coverage-group parameters.

Action messages

Output mirrors the legacy engine: New, Increase, Decrease, Reschedule earlier, Reschedule later, Cancel. Planners review and accept through the planning workbench in F&O.

What's not yet covered

Planning Optimization has parity for most AX MRP scenarios, but some specialised features remain on the legacy engine in some implementations: certain co-product / by-product configurations, intercompany advanced features, and some country-specific localizations. Microsoft publishes a feature-parity matrix; check before assuming everything ports cleanly.

Co-existence

During migration, customers can run both engines in parallel against the same data, compare results, and switch once parity is confirmed. This is the standard transition pattern from legacy MRP to Planning Optimization.

Manufacturing scheduling

Planning Optimization handles infinite-capacity material plans. Finite capacity scheduling (gantt-style scheduling on constrained work centres) remains a separate concern — typically run with the Scheduling engine or a partner scheduling product on top of the material plan.

Net result

A four-hour overnight MRP becomes a fifteen-minute job. Planning frequency goes from nightly to multiple times a day, and planners spend less time waiting and more time analysing.