Mastering Service Discovery with Spring Cloud Eureka: A Practical Guide

This article originates from the ADDOPS team; the author Li Dong is responsible for the Wonder monitoring platform and the Furion code release platform. It introduces how to use Spring Cloud Eureka for service governance.

1. What is Eureka

Eureka is Netflix’s open‑source RESTful service for service registration and discovery, consisting of an Eureka server and Eureka client. The server acts as a registration hub, while the Java client simplifies interaction, provides load‑balancing, and supports fault‑tolerance.

Open‑source

Reliable

Feature‑rich

Java‑based

Integrated with Spring Cloud

2. Important Concepts

Register (service registration)

When a Eureka client registers with the server, it supplies metadata such as IP, port, health status, URLs, and homepage.

Renew (service renewal)

The client sends a heartbeat every 30 seconds. If the server does not receive a renewal within 90 seconds, it removes the instance from the registry. This interval should not be changed.

Get Registry (fetch registry information)

The client retrieves the registry from the server, caches it locally, and communicates using JSON/XML (default compressed JSON).

Cancel (service deregistration)

When a client shuts down, it sends a cancel request, after which its instance information is removed from the server’s registry.

3. High‑Availability Architecture

The official Eureka architecture diagram is shown below:

Eureka Server : provides Register, Get, Renew APIs for service registration and discovery.

Application Service : registers itself to the Eureka Server.

Application Client : obtains service instances from the Eureka Server and calls the Application Service.

4. Processing Flow

Each region has an Eureka cluster with at least one server per availability zone.

Application Services register as Eureka clients and send heartbeats; missing heartbeats for 90 seconds cause removal.

When a server’s data changes, it synchronizes the data to other Eureka servers.

Application Clients fetch instance information via the Get API.

Clients use internal load‑balancing to call services, supporting cross‑region calls.

Clients cache the registry locally and periodically fetch incremental updates (approximately every 30 seconds). If inconsistencies are detected, the full registry is re‑fetched. Propagation of changes across servers may take up to two minutes, but the local cache enables fast, resilient service calls.

5. Mode Introduction

Eureka Server uses a peer‑to‑peer (P2P) architecture without a master/slave distinction; each server acts both as a server and a client, allowing flexible clustering.

Clients use a local copy of the registry, so service calls and load‑balancing continue even if the server is unavailable. However, client caches may delay detection of a failed service by one to three heartbeat cycles, which can be mitigated with circuit‑breaker and fallback mechanisms such as Netflix Hystrix.

6. Deployment and Installation

Environment Preparation

VMWare virtual machine

CentOS 7

Git

Gradle

Tomcat

Steps

Install Git, Gradle, and Tomcat.

Build Eureka.

git clone https://github.com/Netflix/eureka.git
cd eureka/
./gradlew clean build

Deploy the Eureka Server:

/usr/local/tomcat/bin/shutdown.sh
cp eureka-server-1.1.134.war /usr/local/tomcat/webapps/eureka.war
/usr/local/tomcat/bin/startup.sh

Access the server via its URL to view registration information.

7. Conclusion

In practice, direct usage of Eureka is relatively rare, but it is widely adopted within microservice architectures through Spring Cloud, which provides out‑of‑the‑box integration and simplifies its deployment.