microservices-pattern-architect_skill

Overview

This skill helps architects and engineers choose and apply microservices patterns (Saga, CQRS, Event Sourcing, Circuit Breaker, API Gateway, Service Discovery) and provides implementation guidance, templates, and trade-offs. It targets migrations, distributed transactions, resilience, and read/write scalability problems and outputs concrete recommendations and starter code scaffolds for common frameworks.

How this skill works

I analyze the provided scenario and architecture_context to map problems to pattern families, then produce a concise pattern recommendation (primary pattern, variant, justification, trade-offs, complexity). For extended requests I generate deep-dive guidance: design notes, failure handling, observability, and framework-specific scaffolds (Spring Boot, Node.js, .NET, Go, Python). I validate input and stop if service boundaries or ACID requirements are missing.

When to use it

• When a distributed transaction spans multiple services and 2PC/XA is not viable
• When read and write workloads diverge significantly or audit/replay is required
• When external dependencies are unreliable and graceful degradation is needed
• When multiple client types require a unified entry point or protocol translation
• When service instances are dynamic in cloud or container environments
• When migrating a monolith to microservices and pattern composition is required

Best practices

• Map the primary concern (transactions, consistency, resilience, routing, discovery) before choosing a pattern
• Design compensating actions and idempotent operations for Sagas; prefer orchestration for complex workflows
• Use CQRS only when read/write separation yields clear benefits; pair with Event Sourcing for auditability
• Apply Circuit Breaker with sensible thresholds, timeouts, and fallback strategies; monitor state transitions
• Keep API Gateway slim; use BFFs for client-specific optimizations to avoid a monolithic gateway
• Use server-side discovery for language-agnostic routing; prefer client-side when low latency and SDK support exist

Example use cases

• E-commerce checkout spanning order, payment, inventory services → Saga (orchestration for complex compensations)
• High-read product catalog with heavy search and analytics → CQRS with Elasticsearch read model
• Audit trail and temporal queries for billing → Event Sourcing with snapshotting
• Third-party payment gateway with intermittent failures → Circuit Breaker + fallback cached response
• Mobile and web clients requiring different payload shapes → API Gateway + BFFs
• Kubernetes microservices with autoscaling → Service Discovery via Kubernetes DNS and health checks