Adding OpenTelemetry to Production Apps Lessons Learned
summary
Engineers add OpenTelemetry to an open source microservice app (Bank of Sirius) running on Kubernetes using OpenTelemetry Collector, Jaeger, Elastic Stack (Filebeat + Kibana), Prometheus + Grafana, Micrometer, Spring Cloud Sleuth, Python libs, nginx module, Locust load generator, Jenkins and Pulumi. The Collector proves indispensable; tracing works (GA maturity), metrics use Micrometer for Java, logs overwhelm elastic (one client hit 50 TB/hour), and Java often needs a shared telemetry module or manual instrumentation. Sampling, clock skew and log‑trace correlation are the main operational headaches.
AI-written summary · about
chapters (14)
- 0:00Start
- 0:20Dave Mcallister
- 2:32Performance Bottlenecks
- 3:09Modern Applications Reference Architecture
- 9:23Core Dumps
- 14:26Indexing
- 16:36Distributed Tracing
- 17:24Open Telemetry Collector
- 19:41Ntp
- 27:18Trace Name Interceptor
- 40:02Auto Configuration
- 41:29Propagating the Trace Idea amongst All the Different Services
- 42:46Use Tracing Information To Replace Logs
- 46:54Is It Important To Have Accurate Timestamps
related talks
35mOpenTelemetry for Busy People
23mModern Application Debugging: An Intro to OpenTelemetry
02h 02mHow To Troubleshoot and Monitor Applications using OpenTelemetry
41mModern Application Debugging: An Introduction To OpenTelemetry
22mOpenTelemetry – An Observability Framework for Cloud-Native Software
05mOpenTelemetry in production for a year plus: the good the bad the ugly
51mPractical introduction to OpenTelemetry tracing
50mBootiful OpenTelemetry