moai-observability-advanced

Distributed tracing visualizes request flows through microservices by instrumenting services with OpenTelemetry SDKs and exporting traces to Jaeger. This reveals latency bottlenecks, service dependencies, and error propagation across your architecture.

Metrics aggregate numeric data over time (throughput, latency), logs capture discrete events with context, and traces map request journeys across services. Together they form the three pillars of production observability.

OpenTelemetry instrumentation for Python involves adding SDK packages, configuring exporters for Jaeger and Prometheus, and instrumenting Flask or request libraries. Automatic instrumentation captures spans and metrics without code changes.

Yes. OpenTelemetry collectors run as sidecars or agents in Kubernetes to gather traces, metrics, and logs from containerized services. This enables cloud-native observability across development, staging, and production workloads.

SLIs (Service Level Indicators) measure performance aspects like latency or availability; SLOs (Service Level Objectives) set acceptable thresholds. Define SLIs from trace and metric data, then enforce SLOs through alerting and dashboards.

Log correlation embeds trace IDs and span IDs into structured logs so you can jump between logs and traces for the same request. OpenTelemetry propagates context automatically across services.