Summary
Adobe has implemented a three-tier OpenTelemetry pipeline to achieve scalable observability across its diverse infrastructure, prioritizing simplicity for its service teams. The system utilizes a user-facing Helm chart with sidecar and deployment collectors, a centrally managed namespace with signal-specific collectors, and various observability backends – allowing for independent scaling and fault isolation. Leveraging auto-instrumentation and custom components (like a circuit breaker for backend errors), Adobe has successfully adopted OpenTelemetry with minimal effort from individual teams, while maintaining flexibility for advanced use cases.