Applying Domain‑Driven Design to Refactor Keep's E‑commerce Supply Chain System

In the rapidly evolving e‑commerce environment, legacy supply‑chain (inventory) systems often suffer from unclear boundaries, tangled layers, and inaccurate stock data, which hinder business growth. The author, a senior architect at Keep, outlines the pain points of the existing system, including unclear system boundaries, mixed responsibilities, and unreliable inventory logs.

The refactor strategy is organized into three main parts: (1) clarifying inventory business scenarios, (2) defining bounded contexts, and (3) modeling inventory entities such as occupied, usable, physical, in‑transit, and frozen stock.

To support the new design, a Domain‑Driven Design approach is adopted, emphasizing a hexagonal architecture that separates the core domain from infrastructure, and introducing CQRS to isolate command and query models. Event sourcing is chosen as the persistence mechanism, with EventStore handling domain events, their publication, and subscription.

Implementation details include publishing domain events (code omitted for brevity), registering subscription groups, and declaring event types. Unit testing is emphasized: domain tests, command‑execution tests, Mockito for mocking, H2 in‑memory database for isolation, and CI integration to run tests automatically.

The outcomes of the transformation are fivefold: accurate inventory counts, easier feature extensions (e.g., finance integration), rapid issue diagnosis thanks to clear code structure, a reusable EventStore component across Keep's services, and a mature DDD best‑practice guide for future projects.

Finally, the article invites readers to discuss DDD and share experiences, emphasizing its natural synergy with micro‑service architectures.