SpringBoot Microservices tracing with Zipkin and Sleuth

Vishwanath SinhaMarch 14th, 2018Last Updated: March 14th, 2018

5 694 4 minutes read

In this article, we are going to understand the microservices distributed tracing using Zipkin and Spring cloud sleuth framework.

Microservices are great architecture, though its comes with its own challenges. Complex microservices architecture have downstream and upstream dependencies with other microservices and everything is interconnected. Most of the time individual team just focus on their own services, so in microservices distributed environment it’s very difficult to find any latency or trace the actual issues at runtime.

1. Distributed Tracing

As the complexity of microservices architecture increases, uniform observability across the services and server instances become critical. Each service requests passes through multiple services layer, identifying latency or any run-time issues becomes increasingly difficult using traditional debugging techniques. Distributed tracing provides a holistic view of requests transiting through multiple services, allowing for immediate identification of issues.

1.1 Distributed Tracing Vocabulary

Span – A span is an individual operation.
Trace – A Trace is an end-to-end latency graph, composed of spans.
Tracers – Tracers records spans and passes context required to connect them into a trace.

2. Zipkin

Zipkin is an open source distributed system that provides mechanisms for sending, receiving, storing, and visualizing traces details. This helps team to correlate activities between server instances and get a much deeper understanding of exactly what is going on in our services. Zipkin provide the UI interface to analyze the traces and call graph details between services.

2.1 Zipkin Installation

There are three different ways zipkin server can be installed.

Java – We can execute the below wget command to download the latest zipkin server. In case we don’t have wget installed, simply we can copy the below url and paste into our browser to download the latest zipkin server.

wget -O zipkin.jar 'https://search.maven.org/remote_content?g=io.zipkin.java&a=zipkin-server&v=LATEST&c=exec'

Zipkin source – We can download the zipkin source code from below github url and run the zipkin application in your local machine: https://github.com/openzipkin/zipkin
Docker – We can download the docker-zipkin container from https://github.com/openzipkin/docker-zipkin

For this sample application I have installed the Zipkin server using the Java method.

2.2 Zipkin UI

Zipkin UI provides following features to analyze the distributed tracing details across services:

Search traces by service name, span name etc
Trace dependency graph showing trace requests flow across services
Provides the details of total trace time for each individual span

By analyzing the details provided by Zipkin UI, it becomes easier to find the latency or any particular services issues across the interconnected microservices architecture.

3. Spring Cloud Sleuth

Spring cloud sleuth add traces and span ids to the Slf4j MDC (Mapped Diagnostic context), to extract logs from a given trace or span. Sleuth provide abstraction over common distributed tracing data models like traces, spans, annotations, key-value annotations. Sleuth instruments common ingress and egress points from Spring applications (servlet filter, rest template, scheduled actions, message channels, zuul filters, feign client).

3.1 Sleuth and Zipkin configuration at SpringBoot

Add following dependencies to the Springboot pom.xml to add the zipkin and spring cloud sleuth into the SpringBoot application.

pom.xml

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

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

4. Sample Application

For this article I have created a sample application which consist of three different microservices called UserGreetingService, UserNameService, UserAddressService.

UserGreetingService interns makes a call to UserNameService and UserAddressService services to constructs the complete greeting message for a user along with name and address details.

UserGreetingService.java

@SpringBootApplication
@RestController
@CrossOrigin
public class UserGreetingService {
	
	@RequestMapping("/api/user/greet")
	public String greet() {
		String greetingMsg = "Hello";
		String userName = restTemplate().getForObject("http://localhost:3001/api/user/name", String.class);
		String userAddress = restTemplate().getForObject("http://localhost:3002/api/user/address", String.class);
		
		return greetingMsg + " " + userName + "!\n\n" + userAddress;
	}
	
	@Bean
	RestTemplate restTemplate() {
		return new RestTemplate();
	}
	
	public static void main(String[] args) {
		SpringApplication.run(UserGreetingService.class,
				"--spring.application.name=user-greeting-service",
				"--server.port=3000"
		);
	}
}

To introduce the services latency, I have added Thread.sleep(1000) and Thread.sleep(2000) intentionally to UserNameService and UserAddressService services.

You can download the sample code from the github link – https://github.com/VishSinha/springbootzipkindemo

4.1 Start Zipkin Server

To start zipkin server in our local env, go to the directory where we have downloaded the zipkin server and open the command prompt or terminal and execute the following command:

java -jar zipkin-server-2.5.0-exec.jar

Once we start the zipkin server, Zipkin UI can be browsed using the url – http://localhost:9411/zipkin/

4.2 Run the Sample

Once you download the sample code from the github, start all the three microservices using your favorite IDE like eclipse, Intellij or using the command prompt.

Once all the three springboot microservices have been started, hit the UserGreetingService endpoint url http://localhost:3000/api/user/greet using browser or postman.

Overall you will notice greeting service took more than 3 secs to render the response. Now, let’s open the Zipkin UI to analyze what’s happening behind the scene and why overall response time is more than 3 secs.

Open a browser window and render the Zipkin UI using the default url – http://localhost:9411/zipkin. This will render zipkin default search page.

Let’s select the user-greeting-service from the service name and click on the Find Traces button.

You will see there are total 3 spans of the greeting services along with total response time.

Click on the “3 spans” link, this will render details of each services response time and other details as below.

Further by clicking on each services span, we can get the details of trace ids, span ids, parent trace ids, server instance info and other tracing details for each services request. This can easily help us to troubleshoot any faulty microservices or latency issues.

Once you open the Dependencies section of Zipkin UI, this will render the services dependency graph across the services as below:

5. Conclusion

Overall zipkin is very easy to integrate cloud based microservices application and this provide lot of insights and information across services to analyze the services overall health. Using Zipkin, we can easily identify any faulty services and corrective action can be taken.