Building a Cloud‑Native Microservice Platform with Spring Cloud and Docker

The goal of a PaaS cloud platform based on microservice architecture and Docker containers is to provide developers with a fast‑development, deployment, operation, and continuous integration workflow. The platform supplies infrastructure, middleware, data services, and cloud servers, allowing developers to focus on business code while the system automatically builds and deploys applications.

Microservice Access Path

The typical request flow is: external request → load balancer → service gateway (GateWay) → microservice → data/message service. Both the gateway and microservices use service registration and discovery, and all clusters obtain configuration from a configuration center.

Service Gateway (GateWay)

The gateway sits between external clients (browsers, mobile devices) and internal services. It is deployed in a cluster behind an Elastic Load Balancer (ELB) on Amazon EC2, with SSL termination handled by Nginx. Core capabilities include dynamic routing, rate limiting and fault tolerance, authentication and security control, monitoring, and access logging. The implementation uses Spring Cloud Netflix’s Zuul component, which relies on a chain of filters to provide flexible gateway functions.

Service Registration and Discovery

Microservices register themselves with Eureka. Each instance sends heartbeats (default every 30 seconds); if a service misses three consecutive heartbeats (default 90 seconds), Eureka marks it as dead. Eureka runs in a cluster and synchronizes registration information via DNS. All service calls first query Eureka to locate target instances.

Microservice Deployment

Each microservice runs as a Docker container image. Docker simplifies deployment by packaging the JDK and application into a single image that can be run on any host. Containers are managed as a Swarm cluster, and a private image repository stores both base and final images.

Service Fault Tolerance

Hystrix (a Netflix open‑source library) provides circuit breaking, thread isolation, fallback, and rate limiting. Circuit breaking stops calls to unhealthy services, isolation prevents a failing service from exhausting shared resources, and fallback logic returns default responses. Hystrix commands are implemented by extending HystrixCommand and overriding run() and getFallback() .

Dynamic Configuration Center

Spring Cloud Config Server stores configuration files in a private Git repository. When a developer pushes changes, a Git hook notifies the Config Server, which refreshes the affected services. This enables runtime updates of parameters such as circuit‑breaker thresholds without rebuilding images.

In summary, the article demonstrates how Spring Cloud Netflix components (Zuul, Eureka, Hystrix, Ribbon) together with Docker containers form a complete cloud‑native microservice solution, covering gateway routing, service discovery, containerized deployment, fault tolerance, and dynamic configuration.