02 – Foundations of Observability & Distributed Tracing

Learning Objectives

• Explain logs, metrics, and traces and how they complement each other.
• Describe the anatomy of a trace and span, and context propagation.
• Understand common trace formats: OTLP, Jaeger, Zipkin.

Lecture Summary

• Logs are detailed event records; metrics are aggregated time series; traces show request causality across services.
• A trace contains spans (operations) with attributes, events, links, and parent/child relationships.
• Propagation uses W3C Trace Context headers: traceparent and tracestate.
• OTLP is the OpenTelemetry-native protocol; Tempo supports OTLP and Jaeger ingestion.

Visual Mental Model

• Client → API Gateway → Service A → Service B → DB
• Each arrow adds a span; the waterfall view exposes latency per hop.

Hands-on Lab: Inspect a Trace Payload

1. Copy the JSON below to trace-sample.json.
2. Identify the traceId, spanId, attributes, and parent relationships.

{
  "resourceSpans": [
    {
      "resource": {"attributes": [{"key": "service.name", "value": {"stringValue": "orders"}}]},
      "scopeSpans": [
        {
          "spans": [
            {
              "traceId": "0af7651916cd43dd8448eb211c80319c",
              "spanId": "b9c7c989f97918e1",
              "name": "GET /orders",
              "kind": 2,
              "attributes": [
                {"key": "http.method", "value": {"stringValue": "GET"}},
                {"key": "http.status_code", "value": {"intValue": 200}}
              ]
            }
          ]
        }
      ]
    }
  ]
}

Deliverables

• Short write-up describing the key parts of a trace.

Quiz (Self-check)

• What headers carry trace context across service boundaries?
• Which pillar would you use to track request per second? To debug a single slow request?

Resources

• W3C Trace Context
• OpenTelemetry Specifications

---

Visuals

Trace Lifecycle (sequence)

Observability Pillars