Sam Okonkwo
Site Reliability Engineer
Sam has carried the pager for systems on both AWS and GCP and writes the runbook they wish had existed at 3 a.m.
- DRM-free PDF and EPUB, yours to keep
- Instant download the moment you check out
- Free updates for the life of the edition
Observability That Tells the Truth
Metrics, Traces, and Logs Without the Cargo Cult
By Sam Okonkwo
Multi-cloudPractitionerPlatform Engineering
Most observability stacks are expensive and unhelpful in the same incident. They collect enormous volume and still cannot answer 'which change caused this'.
Sam Okonkwo works backwards from the questions you ask under pressure and derives the instrumentation that answers them: cardinality budgets that hold, trace sampling that keeps the interesting traces, and alerting tied to user-visible symptoms rather than machine states.
Vendor-neutral, with concrete guidance for CloudWatch and Google Cloud Observability.
What you'll learn
- Derive instrumentation from the questions you ask during incidents
- Hold a cardinality budget without losing diagnostic power
- Sample traces so the interesting ones survive
- Alert on user-visible symptoms instead of machine states
Table of contents
- 01Working Backwards From the Incident28 pp
- 02Cardinality Budgets36 pp
- 03Trace Sampling34 pp
- 04Logs as a Last Resort30 pp
- 05Symptom-Based Alerting38 pp
- 06CloudWatch and Cloud Observability32 pp
6 chapters · 322 pages total
Details
- ISBN
- 978-1-959321-13-7
- Edition
- 1st
- Pages
- 322
- Published
- Formats
- PDF, EPUB
- Platform
- Multi-cloud
About the author
Also available in The Cloud Security Track — save $23Identity, networking, and observability together — the three things an attacker uses and a defender needs.View bundle →Also available in The Platform Engineering Track — save $37Everything needed to build and run an internal platform on managed Kubernetes and Terraform without losing your users.View bundle →