For eight years, Lifemote Networks has analyzed Wi-Fi data for European ISPs, processing billions of events to preempt customer issues. Yet internally, their observability stack was a fragmented mess. Metrics lived in Prometheus and Grafana, logs were scattered between OpenSearch and CloudWatch, and distributed tracing was non-existent in production.

Debugging meant hours of manually stitching data across dashboards. Searching for a unified solution, the team chose OpenTelemetry. Adopting OTel Collector as a sidecar pattern for ECS services ensured failure isolation and compatibility with their AWS App Mesh.

The biggest hurdle was bridging existing logging libraries—Zap in Go and Loguru in Python—to emit OTel-compatible logs with trace context. This correlation became the game-changer. By linking traces, logs, and metrics via trace_id and span_id, Lifemote slashed debugging time, moving from scattered metrics to pinpointing bottlenecks instantly.

They even kept Prometheus for cost efficiency while using Grafana Cloud for visualization. The next stop? Integrating continuous profiling to complete the full observability picture.