Control and monitor_microservices_with_microprofile
0 likes162 views
The document discusses the use of MicroProfile for deploying, monitoring, and managing microservices, focusing on observability and application performance management. It details the integration of metrics, tracing, and fault tolerance within MicroProfile, as well as the use of Prometheus and Grafana for data visualization. Additionally, it addresses Kubernetes orchestration and monitoring for maintaining application health.
Control and monitor_microservices_with_microprofile
- 1. Deploy, monitor, and take control of your micro-services with MicroProfile
- 2. • How do you monitor this? • Observability Octo Blog
- 3. What You’ll Learn Today • Metrics Background • MicroProfile: Metrics - Fault Tolerance • Application Metrics - Prometheus - Grafana
- 4. Rudy De Busscher • Payara • Service team • JSR-375 • MicroProfile committer • Committer in Eclipse EE4J groups - Jakarta EE • Java EE Security API Expert group member @rdebusscher https://blog.payara.fish/ https://www.atbash.be
- 5. Observability • Measure of the internal states of a system • Monitor system(s) • Metrics • Tracing • Health • Data on the behaviour of systems • Proactive • Automation • Application Performance Management (APM)
- 6. Metrics • Telemetry data • Gather data and collected centrally for monitoring • Just collecting data • No decisions made • Collect upfront • Data need to be available when investigation needed • Start when there are already issues? • Base line
- 7. Tracing • Understanding your application’s behaviour • Trace single client request • Details of request flowing though different micro-services • Debugging purposes • Finding bottlenecks • Monitoring
- 8. Health • Mostly based on metrics • Kill the server? • “Treat your servers like cattle not pets”
- 9. MicroProfile • MicroProfile can make your Java EE application more MicroService-ish • Specifications for • Configuration • TypeSafe Rest Client • Security (JWT Auth) • Metrics • OpenTracing • Fault tolerance • Health • OpenAPI • …
- 10. Test system Payara Server Prometheus Grafana Time series DB Graphics Dashboards Expose metrics Using Docker containers
- 11. Setup demo environment • Creating containers • Connecting them together • ‘service’ simulation • Create load through ‘bad’ parallel Prime values calculation.
- 12. Demo
- 13. MP Metrics • Expose metrics with Rest endpoint • Built on CDI only • Included in MicroProfile 1.2+ • Automatic instrumentation • No developer code required • Integration with other MicroProfile specs • Like Fault Tolerance • Custom developer metrics
- 14. MicroProfile Metrics • /metrics endpoint • Prometheus or JSON format • Different categories 1. System based (default) 2. On events indicated by developer 3. On Fault Tolerance events. 4. Custom defined (business related)
- 15. • Why Not JMX? • Polyglot nature of Micro-services • JMX difficult for non-java systems. MicroProfile metrics
- 16. Default MP Metrics • Required by specification • System based • CPU • Memory - Garbage Collection • Threads • Class count Freepik 1
- 17. Prometheus • Open Source • Started by SoundCloud in 2012 • Now Cloud Native Computing Foundation. • Multi-dimensional data model with time series • Query Language • Pull data • Alert Manager
- 18. Prometheus alternatives • Graphite • Need to push metrics to it • Influx DB • Needs rest of stack (Telegraf - Chronograf - Capacitor)
- 19. Grafana • Open Source tool for visualization • query, visualize, alert on • For time series
- 20. • Kibana • Recent improved support for time-series • Works only on top of ElasticSearch Grafana alternatives
- 21. On events • Invocation Count • @Counted • Execution time information • @Timed • @Metered • More detailed information with Tracing Freepik 2
- 22. Fault Tolerance • Number of retries • @Retry • @CircuitBreaker • Open, Half-Open, Closed • @BulkHead • Accepted, Rejected • @Callback • # calls Freepik 3
- 24. Developer defined • Custom defined values • @Gauge • On field or getter • If business value, store them somewhere Freepik 4
- 25. Kubernetes • Orchestration • Docker-compose • Ideal for smaller ‘installations’ • Open-Source Platform to Manage containers • Not only docker containers • Core tasks • Deployment • Scale • Monitoring
- 27. Kubernetes Monitoring • Metrics • Metrics for k8s processes/actions • Metrics-server • Scaling • Horizontal Pod scaling (kubectl autoscale ) • Monitoring • Restart pods • Keep system ‘healthy'
- 28. K8s monitoring • Demo • Kill server • Automatic recovering
- 29. Demo K8s monitoring • K8S • Kill server/container/pod • Automatic recovering • Prometheus within k8s
- 30. Take aways • Get Metrics from your app/‘server’ • Connect to Prometheus • Show time series with Grafana • Different types of metrics • System • ‘Events’ • From other MP • Kubernetes • Keep app healthy
- 31. Code • MicroProfile • https://microprofile.io • Demo code • https://github.com/rdebusscher/mp-metrics-demo
- 32. Q & A
- 33. Thank You Not using the Payara Platform yet? Download the open source software: Payara Server or Payara Micro https://payara.fish/downloads Need support for the Payara Platform? https://payara.fish/support