Master Spring Cloud: From Service Discovery to Resilience and Dynamic Routing

Why Spring Cloud

Spring Cloud is an ordered collection of frameworks that leverages Spring Boot’s convenience to simplify distributed system infrastructure such as service discovery, configuration center, message bus, load balancing, circuit breaker, and data monitoring, all with one‑click startup.

Spring Boot

Spring Boot solves the redundancy of traditional Maven XML configurations, enabling rapid creation of single microservices. It provides auto‑configuration, embedded containers, and the Actuator for monitoring application health, metrics, and custom endpoints.

Spring Boot Advantages

Automatic bean initialization reduces repetitive code.

Embedded containers allow running applications as executable JARs.

Actuator offers built‑in monitoring of beans, environment, logs, threads, etc.

spring-boot-starter-web          // Web MVC
spring-boot-starter-data-redis   // Redis integration
spring-boot-starter-data-mongodb // MongoDB integration
spring-boot-starter-data-jpa     // MySQL/JPA integration

Custom Starter Creation

Create a starter project and define configuration properties.

Write an auto‑configuration class that loads beans based on those properties.

Register the class in spring.factories so Spring scans it at startup.

(Optional) Add spring-configuration-metadata.json for IDE assistance.

Include the starter as a Maven dependency in other projects.

Spring Boot Admin

Visualizes Actuator data, showing JVM info, thread details, logs, and request tracing.

Gateway / Zuul

Spring Cloud Gateway replaces Netflix Zuul, offering unified routing (reverse proxy) and filter chains for authentication, rate limiting, circuit breaking, path rewriting, and logging.

Route : basic routing unit with ID, target URL, predicates, and filters.

Predicate : matches HTTP requests (similar to Nginx locations).

Filter : pre‑ and post‑filters for parameter validation, auth, logging, etc.

Eureka / Zookeeper

Service registry decouples providers and consumers, supporting elastic scaling.

Eureka uses heartbeat, client caching, and a two‑level cache (read‑only ConcurrentHashMap refreshed every 30 s, and read‑write Guava cache refreshed every 180 s) to improve response speed while risking stale data.

Self‑Protection & Health Check

Shorten cache refresh intervals or disable read‑only cache for timely discovery.

Configure self‑protection thresholds based on expected heartbeat rates.

Feign / Ribbon

Feign simplifies HTTP client creation, integrates with Ribbon for load balancing and Hystrix for circuit breaking, and can use Protobuf for fast RPC.

InvocationHandlerFactory : JDK dynamic proxy for remote HTTP calls.

Contract : defines request method and URI.

Encoder/Decoder : handle request/response serialization.

Logger : configurable logging levels.

Client : replace default HttpURLConnection with Apache HttpClient for better performance.

Retryer : retries on I/O failures.

RequestInterceptor : add headers or other context before sending.

Load‑Balancing Algorithms

RoundRobinRule – round‑robin.

RandomRule – random selection.

BestAvailableRule – choose server with fewest concurrent requests.

Hystrix

Provides circuit breaking, resource isolation (thread pool or semaphore), fallback logic, and detailed metrics. Common pitfalls include configuration overload, ThreadLocal loss in semaphore mode, and the need for signal‑semaphore tuning.

Sentinel

Cloud‑native flow‑control and circuit‑breaker component that replaces Hystrix for Alibaba’s high‑traffic scenarios.

Offers real‑time monitoring, rule persistence, and integration with Spring Cloud or Dubbo.

Config / Nacos

Nacos is Alibaba’s open‑source solution for service discovery, configuration management, and metadata handling (Eureka + Config + Bus).

Key concepts: Namespace (environment), Group (project), Service (specific service), DataId (configuration file). Use @RefreshScope for dynamic updates.

Bus / Stream

Spring Cloud Stream abstracts underlying MQ differences (RabbitMQ, Kafka) to simplify message‑driven microservices.

Sleuth / Zipkin

Generates a unique Trace ID for each request and Span IDs for each processing unit, enabling latency analysis, service‑call tracing, and aggregated visualization in Zipkin.

Security Authentication

Session – simple but needs sticky sessions or Redis for clustering.

HTTP Basic – base64 Authorization header validation.

Token/JWT – stateless token containing user info, signed and verified by the server.

Token Best Practices

Set reasonable expiration times.

Invalidate revoked tokens via Redis blacklist.

Monitor usage frequency to detect abuse.

Use dynamic verification (CAPTCHA) for sensitive actions.

Rotate encryption keys during low‑traffic periods.

Gray Release & Multi‑Version Isolation

Gradual rollout reduces risk; route a subset of users to new versions, rollback quickly if issues arise. Use Ribbon’s load‑balancing strategies and Discovery zones to direct traffic to specific regions or zones.

Component Tuning

Tomcat – adjust accept-count, max-threads, max-connections.

Hystrix – increase semaphore limit (default 100) or switch to thread‑pool isolation.

Gateway – configure per‑route and total connection limits.

Ribbon – set MaxConnectionsPerHost and MaxTotalConnections.

Feign – replace default client with HttpClient for higher throughput.

Dynamic Routing & Logging

Leverage Gateway + Config Center for runtime route changes and Feign + Config Center for on‑the‑fly log level adjustments.