1. Introduction

Netflix Ribbon is an Inter Process Communication (IPC) cloud library. Ribbon primarily provides client-side load balancing algorithms.

Apart from the client-side load balancing algorithms, Ribbon provides also other features:

• Service Discovery Integration – Ribbon load balancers provide service discovery in dynamic environments like a cloud. Integration with Eureka and Netflix service discovery component is included in the ribbon library
• Fault Tolerance – the Ribbon API can dynamically determine whether the servers are up and running in a live environment and can detect those servers that are down
• Configurable load-balancing rules – Ribbon supports RoundRobinRule, AvailabilityFilteringRule, WeightedResponseTimeRule out of the box and also supports defining custom rules

Ribbon API works based on the concept called “Named Client”. While configuring Ribbon in our application configuration file we provide a name for the list of servers included for the load balancing.

Let’s take it for a spin.

2. Dependency Management

The Netflix Ribbon API can be added to our project by adding the below dependency to our pom.xml:

<dependency>
 <groupId>org.springframework.cloud</groupId>
 <artifactId>spring-cloud-starter-netflix-ribbon</artifactId>
</dependency>

The latest libraries can be found here.

3. Example Application

In order to see the working of Ribbon API, we build a sample microservice application with Spring RestTemplate and we enhance it with Netflix Ribbon API along with Spring Cloud Netflix API.

We’ll use one of Ribbon’s load-balancing strategies, WeightedResponseTimeRule, to enable the client side load balancing between 2 servers, which are defined under a named client in the configuration file, in our application.

4. Ribbon Configuration

Ribbon API enables us to configure the following components of the load balancer:

• Rule – Logic component which specifies the load balancing rule we are using in our application
• Ping – A Component which specifies the mechanism we use to determine the server’s availability in real-time
• ServerList – can be dynamic or static. In our case, we are using a static list of servers and hence we are defining them in the application configuration file directly

Let write a simple configuration for the library:

public class RibbonConfiguration {

 @Autowired
 IClientConfig ribbonClientConfig;

 @Bean
 public IPing ribbonPing(IClientConfig config) {
 return new PingUrl();
 }

 @Bean
 public IRule ribbonRule(IClientConfig config) {
 return new WeightedResponseTimeRule();
 }
}

Notice how we used the WeightedResponseTimeRule rule to determine the server and PingUrl mechanism to determine the server’s availability in real-time.

According to this rule, each server is given a weight according to its average response time, lesser the response time gives lesser the weight. This rule randomly selects a server where the possibility is determined by server’s weight.

And the PingUrl will ping every URL to determine the server’s availability.

5. application.yml

Below is the application.yml configuration file we created for this sample application:

spring:
 application:
 name: spring-cloud-ribbon

server:
 port: 8888

ping-server:
 ribbon:
 eureka:
 enabled: false
 listOfServers: localhost:9092,localhost:9999
 ServerListRefreshInterval: 15000

In the above file, we specified:

• Application name
• Port number of the application
• Named client for the list of servers: “ping-server”
• Disabled Eureka service discovery component, by setting eureka: enabled to false
• Defined the list of servers available for load balancing, in this case, 2 servers
• Configured the server refresh rate with ServerListRefreshInterval

6. RibbonClient

Let’s now set up the main application component snippet – where we use the RibbonClient to enable the load balancing instead of the plain RestTemplate:

@SpringBootApplication
@RestController
@RibbonClient(
 name = "ping-a-server",
 configuration = RibbonConfiguration.class)
public class ServerLocationApp {

 @Autowired
 RestTemplate restTemplate;

 @RequestMapping("/server-location")
 public String serverLocation() {
 return this.restTemplate.getForObject(
 "http://ping-server/locaus", String.class);
 }

 public static void main(String[] args) {
 SpringApplication.run(ServerLocationApp.class, args);
 }
}

And here’s the RestTemplate configuration:

@Configuration
public class RestTemplateConfiguration{
 @LoadBalanced
 @Bean
 RestTemplate getRestTemplate() {
 return new RestTemplate();
 }
}

We defined a controller class with the annotation @RestController; we also annotated the class with @RibbonClient with a name and a configuration class.

The configuration class we defined here is the same class that we defined before in which we provided the desired Ribbon API configuration for this application.

Notice we annotated the RestTemplate with @LoadBalanced which suggests that we want this to be load balanced and in this case with Ribbon.

7. Failure Resiliency in Ribbon

As we discussed earlier in this article, Ribbon API not only provides client side load balancing algorithms but also it has built in failure resiliency.

As stated before, Ribbon API can determine the server’s availability through the constant pinging of servers at regular intervals and has a capability of skipping the servers which are not live.

In addition to that, it also implements Circuit Breaker pattern to filter out the servers based on specified criteria.

The Circuit Breaker pattern minimizes the impact of a server failure on performance by swiftly rejecting a request to that server that is failing without waiting for a time-out. We can disable this Circuit Breaker feature by setting the property niws.loadbalancer.availabilityFilteringRule.filterCircuitTripped to false.

When all servers are down, thus no server is available to serve the request, the pingUrl() will fail and we receive an exception java.lang.IllegalStateException with a message “No instances are available to serve the request”.

8. Conclusion

In this article, we discussed Netflix Ribbon API and its implementation in a simple sample application.