How to Build a Microservice‑Based PaaS Platform with Spring Cloud Netflix and Docker

Overview of the PaaS Microservice Platform

The goal of the PaaS cloud platform built on microservices and Docker is to give developers a fast, automated workflow for coding, building, deploying, and operating applications. Developers only need to write business code, push it to a code repository, and configure a few settings; the system then automatically builds and deploys the service, enabling agile development and rapid iteration.

Service Gateway (GateWay)

The gateway sits between external clients (browsers, mobile devices) and internal services. It is deployed in a cluster behind an Amazon EC2 instance pool with Elastic Load Balancing (ELB) to handle high concurrency. ELB scales automatically, and Nginx performs SSL termination. The gateway provides:

Dynamic routing to backend service clusters.

Rate limiting and circuit‑breaker‑style fault isolation.

Authentication, security controls, and anti‑scraping features.

Monitoring and metrics collection.

Access logging for request tracing and performance analysis.

Implementation uses Spring Cloud Netflix's Zuul with custom filters to realize these capabilities.

Service Registration and Discovery

Microservices register themselves with an Eureka server, sending periodic heartbeats (default every 30 seconds). Eureka marks a service as dead after 90 seconds of missed heartbeats unless self‑protection is disabled. Eureka clusters synchronize registration data via DNS‑based discovery, ensuring consistent service information across nodes.

Microservice Deployment

Each microservice is packaged as a Docker image and run in Docker containers managed by Docker Swarm. Images (including base JDK layers) are stored in a private image repository. Deploying a new version simply involves building a new image and updating the Swarm service, eliminating the need to install runtimes on every host.

Service Fault Tolerance with Hystrix

Hystrix, a Netflix open‑source library, provides circuit breaking, thread isolation, fallback handling, and rate limiting:

Circuit breaking: Triggers when error rate exceeds 50 %, returning a fallback response instead of cascading failures.

Thread isolation: Each service can use a dedicated thread pool (e.g., TestThreadPool) to prevent one faulty service from exhausting shared resources.

Fallbacks: Implemented by extending HystrixCommand and overriding getFallback() to return default values, cached data, or propagate a fast‑fail exception.

Rate limiting: Limits concurrent requests per service, rejecting excess traffic to protect downstream systems.

Dynamic Configuration Center

Configuration files that need runtime changes (e.g., circuit‑breaker thresholds) are stored in a Git repository and served by Spring Cloud Config Server. When a developer pushes updates, a server‑side Git hook notifies Config Server via a message queue, prompting it to refresh the affected services without rebuilding images.

Core Open‑Source Components Used

The implementation relies on the following Spring Cloud Netflix modules:

Zuul – API gateway.

Eureka – Service registration and discovery.

Hystrix – Fault tolerance and circuit breaking.

Ribbon – Client‑side load balancing.

Additional optional components such as logging and messaging services can be added as needed.

Illustrative Architecture Diagram

Conclusion

The article outlines the essential building blocks for a microservice‑based PaaS platform, emphasizing how Spring Cloud Netflix components, Docker containerization, and a dynamic configuration center together enable scalable, resilient, and easily maintainable cloud services.