microservices-patterns

Decompose a monolith by identifying business capabilities or subdomains, then break functionality into independently deployable services with clear boundaries. Assign data ownership to each service, define API contracts for synchronous communication, and use event-driven messaging for asynchronous coordination to minimize coupling.

Use synchronous communication (REST, gRPC) for request-response flows requiring immediate replies, and asynchronous event-driven messaging for workflows where services publish and consume events. Event-driven patterns reduce tight coupling and improve resilience by decoupling service dependencies.

Implement the Saga pattern to coordinate transactions across multiple services without distributed locks. Sagas break operations into local transactions with compensating actions, achieving eventual consistency while maintaining data integrity across service boundaries.

Circuit breakers prevent cascade failures by stopping requests to failing services, while retry with backoff and bulkheads isolate faults. These resilience patterns protect distributed systems from cascading outages when services degrade or experience latency.

Yes. Database-per-service enforces service boundaries and data ownership, ensuring each service controls its schema. Pair it with event-driven communication and Saga pattern to achieve eventual consistency across services without shared databases.

API Gateway provides a single entry point, handles cross-cutting concerns like routing and rate limiting, and shields clients from service topology changes. Direct calls suit internal service communication; gateways suit client-facing APIs and polyglot ecosystems.