Last Minute Notes
  • Github
  • Docs
  • Learn
    @sayanstwt
Lessons
0/37 done
  • Timeouts, Retries & Breakers
  • Multi-Region Design
  • Cells & Shuffle Sharding
  • SLIs, SLOs & Error Budgets
  • Observability
  • Deployment Strategies
  • Backups & Disaster Recovery
Learn/System Design/Reliability & Operations

Observability

7 min Intermediate 7 steps

Metrics, logs and traces, and how a trace follows a request.

Metrics
none
Logs
scattered
Traces
none
ClientGatewayOrdersPaymentsDatabasecheckoutcheckoutcheckoutcheckout
request response data
In this lessonStep 1 of 7
  1. The problem
  2. Four signals
  3. OpenTelemetry
  4. Tracing
  5. What to measure
  6. Sampling
  7. Recap
Step 1 / 7problem

Checkout fails, and nobody can see why

A checkout request crosses four components and comes back as a 500.

Without telemetry every hop is a question mark. Observability is the ability to understand internal state from external outputs.

Key idea

You need observability before the incident, because you cannot add it during one.

←→ step·R replay·Space pause

You'll learn

  • Tell metrics, logs, traces and profiles apart by the question each answers
  • Why OpenTelemetry decouples instrumentation from backends
  • Follow one request through a trace with a shared id
  • Choose RED, USE and business metrics, and sample traces sensibly
Read the full notesReliability and Operations, 2