Distributed Tracing using Spring Cloud Sleuth and Zipkin

Transcript

1. Distributed tracing with Spring Cloud and Zipkin Zoltan Altfatter @altfatterz
2. About me • Zoltan Altfatter • Software Architect @triforkams • Twitter: @altfatterz • Blog: http://zoltanaltfatter.com
3. Agenda • Introduction to the problem • Log aggregation • Distributed tracing • Spring Cloud Sleuth • Zipkin • Demo
4. Typical microservice architecture
5. microservice routes
6. The Network tab in Chrome debugger
7. The call three is the microservice route
8. Where the time spent most?
9. Agenda • Introduction to the problem • Log aggregation • Distributed Tracing • Spring Cloud Sleuth • Zipkin • Demo
10. Log aggregation • add a correlation id into the logs and propagate it across processes • the correlation id identifies the business activity • application instances send instances to a central service • tools: ELK, Papertrail, Loggregator, graylog, Splunk
11. Agenda • Introduction to the problem • Log aggregation • Distributed tracing • Spring Cloud Sleuth • Zipkin • Demo
12. Distributed tracing • distributed context propagation • it is like centralised logging but with structure • view end-to-end latency graphs (traces) in near real time • academic paper Dapper (from Google) • span • trace
13. Span • an individual operation that took place, contains timestamped events and tags • example: sending an RPC call or sending an RPC response. • have an ID, and another ID which identifies the trace • other data: description, timestamped events, key-value annotations (tags)
14. Trace • is an end-to-end latency graph, composed of spans.
15. Agenda • Introduction to the problem • Log aggregation • Distributed tracing • Spring Cloud Sleuth • Zipkin • Demo
16. Spring Cloud Sleuth • passes tracing information automatically by instrumenting common communication channels (HTTP client integration, Asynchronous communication, Messaging, etc…) • adds the trace and span ids to the Slf4J MDC (Mapped Diagnostic Context) • records timing information • sampling • propagates the IDs (trace, span) in-band and the rest (tags, annotations) out-of band
17. Spring Cloud Sleuth
18. Spring Cloud Sleuth
19. Spring Cloud Sleuth with Rabbit or Kafka binding
20. Spring Cloud Sleuth
22. Agenda • Introduction to the problem • Log aggregation • Distributed tracing • Spring Cloud Sleuth • Zipkin • Demo
23. Zipkin • created byTwitter (open-sourced in 2012) • https://github.com/openzipkin • distributed tracing system • OpenZipkin is a GitHub organisation • aggregates timing data that can be used to track down latency issues. • the community is starting to standardise around Zipkin
24. Running Zipkin • java -jar zipkin.jar • docker run -d -p 9411:9411 openzipkin/zipkin • spring cloud zipkin
25. How does Zipkin work? • application instances send spans to Zipkin collectors. • via HTTP (spring-cloud-sleuth-zipkin) • via Spring Cloud Stream (spring-cloud-sleuth-stream) • the spans are stored into a database (mysql / cassandra) • a UI is provided which is using an REST API to query the database
26. Agenda • Introduction to the problem • Log aggregation • Distributed tracing • Spring Cloud Sleuth • Zipkin • Demo
27. Demo
28. Follow
29. Thank you. https://github.com/altfatterz/ microservices-meetup-amsterdam-2016