microservices

What I do

I provide guidance on building microservices architecture - independently deployable services with bounded contexts.

When to use me

Load this skill when:

• •Multiple independent teams (> 20 developers)
• •Clear bounded contexts
• •Need independent scaling of services
• •Polyglot persistence requirements
• •Can tolerate eventual consistency
• •Mature DevOps practices

Core Principles

1. Service Per Bounded Context

Each service owns a bounded context with its own data.

2. Database Per Service

No shared databases between services.

3. API-First Communication

Services communicate via APIs (REST/gRPC) or messaging.

4. Independent Deployment

Deploy services independently without coordinating releases.

5. Decentralized Governance

Teams own their service tech stack choices.

Service Structure

services/
├── user-service/
│   ├── src/
│   ├── Dockerfile
│   └── package.json
├── product-service/
│   ├── src/
│   ├── Dockerfile
│   └── package.json
└── order-service/
    ├── src/
    ├── Dockerfile
    └── package.json

Communication Patterns

Synchronous (REST/gRPC)

// Service-to-service HTTP call
class OrderService {
  async createOrder(userId: string, productIds: string[]) {
    // Call user service
    const user = await fetch(`${USER_SERVICE_URL}/users/${userId}`);
    
    // Call product service
    const products = await fetch(`${PRODUCT_SERVICE_URL}/products`, {
      method: 'POST',
      body: JSON.stringify({ ids: productIds }),
    });
    
    // Create order with fetched data
  }
}

Asynchronous (Message Queue)

// Publish event to message queue
await messageQueue.publish('order.created', {
  orderId: order.id,
  userId: order.userId,
  items: order.items,
});

// Other services subscribe
messageQueue.subscribe('order.created', async (event) => {
  await inventoryService.reserveItems(event.items);
});

Advantages

1. •Independent Scaling: Scale services based on load
2. •Technology Diversity: Choose best tool per service
3. •Team Autonomy: Teams own services end-to-end
4. •Fault Isolation: Service failure doesn't crash system
5. •Parallel Development: Teams work independently

Challenges

1. •Distributed Complexity: Network calls, latency
2. •Data Consistency: Eventual consistency, distributed transactions
3. •Operational Overhead: More services to deploy/monitor
4. •Testing Complexity: Integration testing across services
5. •Debugging: Tracing requests across services

Patterns

API Gateway

Client → API Gateway → [User Service, Product Service, Order Service]

Service Mesh

For service-to-service communication (Istio, Linkerd)

Saga Pattern

For distributed transactions across services

CQRS

Separate read and write models

Best Practices

1. •Design for failure: Circuit breakers, retries, timeouts
2. •Observability: Distributed tracing, centralized logging
3. •API versioning: Don't break consumers
4. •Service discovery: Dynamic service registration
5. •Configuration management: Centralized config
6. •Automated deployment: CI/CD for each service