Lesson series · 6 lessons
OpenTelemetry Concepts: From Traces to Resilient Collectors
Short narrated video lessons. Pause on any moment and ask about it. Anything beyond the series' sources is tagged “General knowledge”.
~3:00Lesson 1 · 3 min
What is a distributed trace, and why does it matter?
Explain observability and its three telemetry signals, describe OpenTelemetry's role as the instrumentation layer rather than a backend, and define a trace as one request's path through many services.
~3:00Lesson 2 · 3 min
How do spans form a trace in OpenTelemetry?
Trace how a shared trace_id and parent_id links turn separate spans into a parent-child tree, read it as a waterfall, and name what else a span carries.
~3:55Lesson 3 · 4 min
How does OpenTelemetry context cross service boundaries?
Trace how a trace ID and parent span ID move from Frontend to Product Catalog in the traceparent header, and decide what baggage should and should not carry.
~3:02Lesson 4 · 3 min
Head vs. tail sampling in OpenTelemetry: when to use each
Explain sampled versus not-sampled terminology and choose head sampling, tail sampling, or both for a given tracing scenario, citing what each sampler can know.
~3:00Lesson 5 · 3 min
Agent vs. gateway Collectors: how telemetry flows in each pattern
Explain where a Collector sits in agent and gateway patterns, predict why trace-ID routing lets tail sampling see whole traces, and recognize a single-writer violation.
- ~4:00
Lesson 6 · 4 min
When the backend goes down: how OpenTelemetry Collectors avoid losing traces
Trace where a batch of spans waits or is lost when the backend or a Collector fails. Predict whether a sending queue, write-ahead log, or Kafka protects that hop.