Distributed Tracing with Jaeger on Kubernetes

With the support of Kubernetes, highly efficient and scalable applications can be developed using a microservices architecture. Unlike monoliths, applications can be divided into a collection of small, autonomous services, each covering a specific business area or function. These services (microservices) can be scaled as needed. However, microservices also bring new challenges that must be mastered. One of these challenges is the distributed tracing of events (REST, Kafka, etc.) processed by multiple microservices. Jaeger offers a solution that allows us to trace each event and see how each service processes it.

Transcript

1. Distributed Tracing with Jaeger on Kubernetes Onur Kasimlar, Lead Cloud Engineer at Posedio VldgWebDev Meetup 17.05.2023
2. 2 POSEDIO GMBH Company founded: • 2020 Locations: • Vienna • Vorarlberg Co-workers: • 17 Cloud experts
3. 3 SERVICES Cloud Platform Cloud-Native Consulting Data Platform Application Modernization Multicloud FinOps Automated self- service Platform Data Governance Big Data Data Accessibility Security and Governance Scalability & Cost optimization DevOps Security Review Cloud readiness Review Architecture Review
4. AGENDA 1. Challenge Distributed Tracing 2. Solutions 3. Showcase Minikube 4.Architecture 5. Real World Example
5. CHALLENGE DISTRIBUTED TRACING 1
6. 7 CHALLENGE DISTRIBUTED TRACING Example Architecture
7. 8 CHALLENGE DISTRIBUTED TRACING What happens if an Error occurs?
8. 9 CHALLENGE DISTRIBUTED TRACING What makes it difficult to find the error? • Error is not logged • Logs are not accessible • Request can’t be tracked • Error != Performance Issue • Where is the bottleneck • …..
9. 10 CHALLENGE DISTRIBUTED TRACING Solution: Monitoring of Systems • Logs: Textual records generated by software applications • Metrics: CPU, Memory, Error rates, … • Alerts: Notifications generated by monitoring systems • Dashboards: visual representation of monitoring data • Traces
10. 11 CHALLENGE DISTRIBUTED TRACING What is a Trace? • Capture execution flow • Request / Transaction path with time • End-to-end visibility
11. 12 CHALLENGE DISTRIBUTED TRACING Example Architecture
12. SOLUTION 2
13. 14 SOLUTION What is a Trace-ID? • Often referred as X-Correlation-Id • Unique Identifier assigned to a trace • Correlate and track the flow of a request or transaction • HTTP-Header in case of HTTP-requests
14. 15 SOLUTION Where/When create a Trace-ID? • ASAP you have the possibility • Check Trace-Id available • (y) – use the existing one • (n) – create one and propagate it
15. 16 SOLUTION Source of error • Every component must stick to the rules • Every component = potential single point of failure • Solution: OpenTelemetry / Jaeger
16. 17 SOLUTION What is OpenTelemetry • Project under Cloud Native Foundation (CNCF) • Standardization of telemetry data • Telemetry data: logs, metrics, traces • APIs and SDKs to generate data • A lot of programming languages supported
17. 18 SOLUTION What is Jaeger • Tool for analyisis of storage and visualization of trace data • Uses OpenTelemetry data
18. 19 SOLUTION How does a Trace look like • A trace represents data / execution path through system • Acyclic graph of spans
19. EXAMPLE IN MINIKUBE AND QUARKUS SERVICES 3
20. ARCHITECTURE 4
21. 22 SOLUTION Architecture
22. 23 SOLUTION Architecture
23. REAL WORLD EXAMPLE 5
24. THANK YOU! POSEDIO GMBH +43 660 8484500 onur.kasimlar@posedio.com Weyringergasse 1-3, 1040 Wien, Millenium Park 4, 6980 Lustenau www.posedio.com