architecture-paradigm-microservices: Скилл для проектирования микросервисов

--- name: architecture-paradigm-microservices description: 'Consult this skill when designing or evolving microservices architectures. Use when teams need high autonomy and independent releases, different capabilities have distinct scaling needs, strong DevOps/SRE maturity exists, polyglot tech stacks needed. Do not use when selecting from multiple paradigms - use architecture-paradigms first. DO NOT use when: small team with low organizational complexity. DO NOT use when: lack of DevOps maturity or limited platform engineering resources. DO NOT use when: strong transactional consistency required across operations.' version: 1.7.1 alwaysApply: false category: architectural-pattern tags: - architecture - microservices - distributed-systems - team-autonomy - scalability dependencies: [] tools: - service-boundary-analyzer - api-contract-generator - resilience-patterns usage_patterns: - paradigm-implementation - distributed-system-design - team-scaling - api-gateway-planning complexity: high model_hint: deep estimated_tokens: 900 --- ## Table of Contents - [When to Employ This Paradigm](#when-to-employ-this-paradigm) - [When NOT to Use This Paradigm](#when-not-to-use-this-paradigm) - [Adoption Steps](#adoption-steps) - [Key Deliverables](#key-deliverables) - [Technology Guidance](#technology-guidance) - [Risks & Mitigations](#risks-mitigations) - [Troubleshooting](#troubleshooting) # The Microservices Architecture Paradigm ## When to Employ This Paradigm - When the organizational structure requires high levels of team autonomy and independent release cycles. - When different business capabilities (bounded contexts) have distinct scaling requirements or would benefit from different technology stacks. - When there is a significant organizational commitment to investing in DevOps and SRE maturity, including advanced observability, CI/CD, and incident response capabilities. ## When NOT To Use This Paradigm - When team size is small and organizational complexity is low - When lack of DevOps maturity or limited platform engineering resources - When system requires strong transactional consistency across operations - When early-stage startup with rapidly evolving requirements - When regulatory constraints make distributed data management challenging ## Adoption Steps 1. **Define Bounded Contexts**: Map each microservice to a clear business capability and establish unambiguous data ownership. 2. **validate Service Data Autonomy**: Each service must own and control its own database or persistence mechanism. All data sharing between services must occur via APIs or events, not shared tables. 3. **Build a production-grade Platform**: Before deploying services, establish foundational infrastructure for service discovery, distributed tracing, centralized logging, CI/CD templates, and automated contract testing. 4. **Design for Resilience**: Implement resilience patterns such as timeouts, retries, circuit breakers, and bulkheads for all inter-service communication. Formally document Service Level Indicators (SLIs) and Objectives (SLOs). 5. **Automate Governance**: Implement automated processes to enforce security scanning, dependency management policies, and consistent versioning strategies across all services. ## Key Deliverables - An Architecture Decision Record (ADR) cataloging all service boundaries, their corresponding data stores, and their communication patterns (e.g., synchronous API vs. asynchronous events). - A set of "golden path" templates and runbooks for creating and operating new services on the platform. - A detailed testing strategy that includes unit, contract, integration, and chaos/resilience tests. ## Technology Guidance **API Communication**: - **REST APIs**: Spring Boot (Java), Express.js (Node.js), FastAPI (Python) - **GraphQL**: Apollo Server (Node.js), Hasura (PostgreSQL) - **gRPC**: gRPC frameworks for high-performance internal communication **Service Discovery & Configuration**: - **Service Registry**: Consul, Eureka, etcd - **Configuration**: Spring Cloud Config, HashiCorp Vault, AWS Parameter Store **Message Broking & Events**: - **Message Brokers**: Apache Kafka, RabbitMQ, AWS SQS/SNS - **Event Streaming**: Apache Kafka, Apache Pulsar, AWS Kinesis **Observability**: - **Distributed Tracing**: Jaeger, Zipkin, AWS X-Ray - **Metrics**: Prometheus, Datadog, CloudWatch - **Logging**: ELK Stack, Fluentd, Splunk ## Real-World Examples **Netflix**: Video streaming platform with hundreds of microservices handling different aspects like playback, recommendation, billing, and user authentication. Each team can deploy independently without affecting others. **Amazon**: E-commerce platform with separate services for product catalog, order processing, payment, inventory, and shipping. Enables independent scaling during high-traffic events like Prime Day. **Uber**: Ride-sharing platform with microservices for rider matching, driver dispatch, pricing, payment processing, and notifications, allowing rapid feature development and deployment. ## Risks & Mitigations - **Distributed System Complexity**: - **Mitigation**: The operational overhead for a microservices architecture is substantial. Invest in dedicated platform teams and shared tooling to manage this complexity and provide support for service teams. - **Data Consistency Challenges**: - **Mitigation**: Maintaining data consistency across services is a primary challenge. Employ patterns like Sagas for orchestrating transactions, validate message-based communication is idempotent, and use reconciliation jobs to handle eventual consistency. - **Incorrect Service Granularity ("Over-splitting")**: - **Mitigation**: If services are too small, the communication overhead can outweigh the benefits of distribution. validate each service owns a meaningful and substantial piece of functionality. Monitor change coupling between services to identify candidates for merging. ## Troubleshooting ### Common Issues **Command not found** Ensure all dependencies are installed and in PATH **Permission errors** Check file permissions and run with appropriate privileges **Unexpected behavior** Enable verbose logging with `--verbose` flag