microservices-patterns

name	microservices-patterns
description	Microservices Patterns: Service decomposition, communication, data management, resilience.
triggers	{"keywords":["microservice","service mesh","API gateway","circuit breaker","service discovery","saga"]}
auto_load_when	Designing or debugging microservices
agent	architect
tools	["Read","Write","Bash"]

Microservices Patterns

Focus: Service boundaries, communication, distributed systems

1. Decomposition Strategies

How to split services:
├── By business capability (orders, payments, users)
├── By subdomain (DDD bounded contexts)
├── By team ( Conway's Law)
└── By operational need (scale, deployment)

Avoid:
├── By technical layer (all services share DB)
├── Too fine-grained (micromanagement)
└── Too coarse (distributed monolith)

---

2. Communication Patterns

Synchronous (request/response):
├── REST - CRUD, simple queries
├── gRPC - performance, contracts
└── GraphQL - flexible queries

Asynchronous (fire-and-forget):
├── Message queues (Kafka, RabbitMQ)
├── Pub/Sub patterns
└── Event-driven

---

3. Data Management

Database per service:
├── Each service owns its data
├── No shared databases
├── Polyglot persistence allowed

Patterns:
├── API composition (query across services)
├── CQRS (separate read/write models)
└── Saga pattern (distributed transactions)

---

4. Service Discovery

How services find each other:
├── Client-side (Eureka, Consul)
├── Server-side (API Gateway)
└── DNS-based (Cloud providers)

Key: Dynamic registration

---

5. Resilience Patterns

Circuit Breaker:
├── Fail fast after threshold
├── Fallback response
└── Auto-recovery

Retry with Backoff:
├── Exponential backoff
├── Jitter (randomization)
└── Dead letter queue

Bulkhead:
├── Isolate failures
├── Resource pools
└── Fail in isolation

---

6. API Gateway

What it does:
├── Routing
├── Authentication
├── Rate limiting
├── Request/response transformation
└── Protocol translation

Consider: Backend for Frontend (BFF)

---

7. Observability

The Three Pillars:
├── Logging - structured, correlated
├── Metrics - RED metrics (rate, errors, duration)
└── Tracing - request flow across services

Essential for debugging distributed systems

---

8. When Microservices

Use when:
├── Team size > 50 developers
├── Independent deployment needed
├── Different scaling requirements
└── Polyglot needed

Avoid when:
├── Team < 10
├── Tight deadline
├── Simple domain
└── No DevOps maturity

---

Key Patterns

1. API Gateway - single entry point
2. Circuit Breaker - resilience
3. Saga - distributed transactions
4. Service Discovery - dynamic routing
5. Observability - debug distributed

(End of file - 84 lines)

---

Anti-Patterns

❌ Distributed monolith — services that must deploy together
✅ True loose coupling: each service deploys independently

❌ Synchronous request chains (service A → B → C → D)
✅ Async event-driven for non-critical paths; aggregate at gateway

❌ No idempotency on message consumers
✅ Every consumer deduplicates by message ID

❌ Schema changes without backward compatibility
✅ Additive changes only; use schema registry for events

❌ No distributed tracing across service calls
✅ Propagate trace-id header; instrument with OpenTelemetry

---

Quick Reference

Pattern	Problem solved	Trade-off
API Gateway	Single entry point + auth	Extra hop
Circuit breaker	Cascade failures	Stale data
Saga	Distributed transactions	Complexity
CQRS	Read/write optimization	Two models to maintain
Event sourcing	Audit trail + replay	Storage growth
Sidecar	Cross-cutting concerns	Resource overhead
Service mesh	mTLS + observability	Ops complexity

🌍 Universal Language Support

• Turkish Native: This skill natively supports Turkish. If the user prompt is in Turkish, all analysis, formatting, and output MUST be entirely in Turkish. You do not need explicit "write in Turkish" instructions.