# Domain-Driven Design Architecture Blueprint ## PHASE 1: DOMAIN ANALYSIS & BOUNDED CONTEXTS **Core Domain:** Food Delivery Platform. To decouple the Node.js monolith, we break the system into four distinct Bounded Contexts, ensuring high cohesion within each service and loose coupling between them: 1. **Catalog Context:** Manages restaurant menus, item availability, and pricing. 2. **Order Context:** Handles the customer cart, checkout process, and order state machine. 3. **Delivery Context:** Manages courier dispatch, GPS tracking, and delivery ETAs. 4. **Billing Context:** Handles payment processing, invoices, and restaurant payouts. ## PHASE 2: DATA OWNERSHIP & AGGREGATES To eliminate the shared PostgreSQL database bottleneck, we implement the "Database-per-Service" pattern. - **Order Context (Target DB: Amazon Aurora PostgreSQL):** - Aggregate Root: Order - Entities: OrderItem, DeliveryAddress - Constraint: Cannot query the Catalog database for prices. Prices are copied into the OrderItem at the time of checkout for immutability. - **Catalog Context (Target DB: MongoDB or DynamoDB):** - Aggregate Root: RestaurantMenu - Entities: MenuItem, Modifier - Constraint: Optimized for heavy reads. - **Delivery Context (Target DB: Redis + PostgreSQL):** - Aggregate Root: DeliveryTrip - Entities: Courier, Route ## PHASE 3: MICROSERVICE COMMUNICATION STRATEGY We will heavily utilize Event-Driven Architecture (Choreography) to avoid tight synchronous coupling. **Synchronous Communication (gRPC/REST):** - The API Gateway synchronously queries the Catalog Service to display menus to the user. **Asynchronous Communication (Apache Kafka):** - *Event: OrderPlacedEvent* -> Emitted by the Order Context. Consumed by the Billing Context to initiate payment. - *Event: PaymentSucceededEvent* -> Emitted by Billing Context. Consumed by Order Context to change status to "Preparing", and by Delivery Context to start searching for a nearby courier. - *Event: CourierAssignedEvent* -> Emitted by Delivery Context. Consumed by Order Context to notify the customer. ## PHASE 4: INCREMENTAL MIGRATION ROADMAP **Strangler Fig Pattern Implementation:** 1. **Step 1: Event Interception:** Implement Change Data Capture (CDC) using Debezium on the existing monolithic database to broadcast legacy state changes to Kafka. 2. **Step 2: Edge Routing:** Deploy an API Gateway (e.g., Kong) in front of the monolith. Route 100% of traffic to the monolith initially. 3. **Step 3: Extract Catalog:** Build the new Catalog Service in Go. Sync data from the monolith to the new Catalog DB via Kafka. Update the API Gateway to route /api/menus traffic to the new Go service. 4. **Step 4: Extract Delivery:** Build the Delivery Service. It listens to monolith DB events to know when to dispatch drivers. 5. **Step 5: Decommission:** Once all contexts are extracted, the monolith is retired.