Docker‑Based Microservice Architecture: Service Discovery, IPC, API Gateway, CI/CD and Operational Practices

The author shares over two years of experience with Docker and microservices, summarising how startups can move from monolithic applications to a container‑based microservice architecture that supports rapid iteration while controlling hidden costs.

Service discovery can be implemented via two main patterns. In the client‑side model, services query a registry such as Netflix Eureka, register themselves with @EnableEurekaClient, and use Ribbon for load‑balancing. In the server‑side model, a load‑balancer forwards requests to a registry like Consul, Etcd or Zookeeper, which provide health checks and KV storage.

Typical registration approaches include self‑registration (e.g., Eureka) and third‑party registration (e.g., Registrator with Consul). Example Eureka registration logs are shown below:

com.netflix.discovery.DiscoveryClient    : DiscoveryClient_SERVICE-USER/{your_ip}:service-user:{port}:cc9f93c54a0820c7a845422f9ecc73fb: registering service...

2018-01-04 20:41:37.290 INFO 49244 --- [Thread-8] c.n.e.EurekaDiscoveryClientConfiguration : Unregistering application service-user with eureka with status DOWN

Inter‑process communication (IPC) can be synchronous (REST/HTTP, Dubbo, Thrift) or asynchronous (message queues such as RabbitMQ, Kafka, ActiveMQ, RocketMQ). The article recommends using asynchronous messaging for domain‑driven designs to avoid tight coupling.

API Gateway acts as a façade for all client requests, handling request aggregation, load‑balancing, caching, authentication, error handling (Hystrix), and reactive programming (RxJava) to achieve non‑blocking I/O. Diagrams illustrate request flow from client → gateway → multiple microservices.

Continuous Integration / Continuous Deployment (CI/CD) is achieved by building Docker images with Maven or Gradle, storing them in a private registry, and orchestrating deployments with Docker Compose, Docker Swarm, or Kubernetes. The article also mentions using GitLab for source control and Jenkins (or GitLab CI) for pipeline automation.

Storage and caching recommendations include MySQL (InnoDB) for relational data, Redis for key‑value caching/persistence, MongoDB for document storage, Neo4j for graph data, and Elasticsearch for full‑text search. Each technology is briefly described with integration notes for Spring Boot.

Security considerations cover network protection (HTTPS, firewalls, high‑availability IP), service‑level authentication (Spring Security, token handling), and internal access controls for databases and registries.

Logging and monitoring should be non‑intrusive; the author suggests using a request filter to add trace IDs, forwarding logs to an ELK stack, and optionally routing log events through a message queue for high‑throughput scenarios.

The article concludes that a well‑designed Docker‑based microservice stack—combining service discovery, API gateway, reactive programming, CI/CD, container orchestration, appropriate storage, security, and observability—provides high performance, availability, scalability, extensibility and security for modern cloud‑native applications.