Observability—Metrics, Logs and Traces

By the end

What you'll build

• Choose between counters, gauges and histograms for a given measurement and explain the consequences
• Predict and control label cardinality before it becomes a cost or reliability problem
• Write structured logs with correlation identifiers and a defensible retention and sampling policy
• Instrument a service boundary so trace context propagates and latency can be attributed
• Correlate a trace with the metrics and logs that explain it
• Reduce telemetry cost while keeping the questions you need to answer answerable

Curriculum

What's inside

7 modules · 35 lessons

1. 01

   Observability compared with monitoring

   5 lessons
   • Known unknowns: answering questions you did not plan for
   • Signals, telemetry and the cardinality problem
   • Instrumenting for questions rather than for dashboards
   • Telemetry cost as a design constraint from day one
   • Signals, questions and what each one costsmatch pairs
2. 02

   Metrics

   5 lessons
   • Choosing counters, gauges and histograms correctly
   • Labels, cardinality and the explosion problem
   • Aggregation, quantiles and their common pitfalls
   • Naming and unit conventions that survive reuse
   • Pick the right metric, then read it rightfill blank
3. 03

   Logs

   5 lessons
   • Structured logging that is worth the storage it costs
   • Levels, sampling and retention decisions
   • Correlation identifiers across service boundaries
   • Keeping sensitive and personal data out of logs
   • Making one request findablesequence order
4. 04

   Traces

   5 lessons
   • Reading a distributed trace to find where the latency lives
   • Spans, context propagation and sampling strategies
   • Instrumenting a service boundary
   • Linking traces to logs and metrics
   • Where the latency livesscenario
5. 05

   Vendor-neutral instrumentation

   5 lessons
   • Instrumenting once against an open, vendor-neutral telemetry specification
   • Automatic and manual instrumentation compared
   • Collectors, pipelines and in-flight processing
   • Migrating an existing estate incrementally
   • Pipeline pieces and where they runmatch pairs
6. 06

   Making telemetry usable

   5 lessons
   • Designing a service dashboard someone can read under pressure
   • Drilldown paths from summary to example
   • Telemetry ownership and lifecycle
   • Cutting telemetry cost without losing visibility
   • Dashboards someone else can readfill blank
7. 07

   Practice and check

   5 lessons
   • Signals and what they meanmatch pairs
   • Working a latency alert at 03:00sequence order
   • Getting the terms exactly rightfill blank
   • High latency, unfamiliar servicescenario
   • Course quizquiz

The shape of it

How this course works