การติดตามแบบกระจายด้วย Zipkin

In a microservices architecture, a single user request might travel through many different services. When something goes wrong, it's like finding a needle in a haystack!

Traditional logging only shows what happened within one service. We need a way to see the entire journey.

Distributed tracing is a technique to monitor and observe requests as they flow through different components of a distributed system.

• It helps you understand how services interact.
• It pinpoints performance bottlenecks.
• It makes debugging across service boundaries much easier.

Zipkin is an open-source distributed tracing system.

It helps you gather timing data needed to troubleshoot latency problems in microservice architectures. It manages both the collection and lookup of this data.

Two fundamental concepts in distributed tracing are Traces and Spans:

• A Trace represents a single request or workflow as it flows through your entire system.
• A Span is a single operation within a trace. It has a name, a unique ID, and a parent ID (unless it's the root span).

Think of a Trace as a story, and Spans as the individual chapters.

Zipkin works by collecting data from your instrumented services.

Here's the basic flow:

1. Your services send tracing data (spans) to the Zipkin Collector.
2. The Collector stores this data (e.g., in memory, MySQL, Elasticsearch).
3. You use the Zipkin UI to visualize and analyze the traces.

For Spring Boot applications, Spring Cloud Sleuth makes integration with Zipkin incredibly easy.

Sleuth automatically adds tracing information to your logs and HTTP headers, and sends it to Zipkin without much manual coding from you.

The easiest way to run a Zipkin server locally is using Docker. If you have Docker installed, just run this command in your terminal:

docker run -p 9411:9411 openzipkin/zipkin

Once running, you can access the Zipkin UI at http://localhost:9411.

To enable tracing, add Sleuth and Zipkin dependencies (spring-cloud-starter-sleuth and spring-cloud-sleuth-zipkin). Then, configure your application.properties to point to the Zipkin server:

spring.application.name=my-traced-service
spring.zipkin.base-url=http://localhost:9411
spring.sleuth.sampler.probability=1.0 (trace all requests)

Now, let's look at a simple service that sends traces!

Here's a basic Spring Boot service. With the properties from the last scene and Sleuth dependencies, it will automatically send trace data to Zipkin when you hit its /hello endpoint.

Try running this and then visiting http://localhost:8080/hello. Check your Zipkin UI!

After making requests to your traced service, open http://localhost:9411 in your browser.

You can search for traces by service name or trace ID. The UI will show you a visual timeline of each trace, including its spans, duration, and dependencies between services.

Great job! You've learned how distributed tracing helps debug microservices, and how Zipkin and Spring Cloud Sleuth make it easy.

• Distributed tracing visualizes request flow.
• Zipkin collects and displays trace data.
• Sleuth automates instrumentation for Spring Boot.

This helps you quickly find issues and optimize performance across your services.