InternshipKey terms and common mistakes
The words you will hear in Site Reliability Engineering work and the mistakes that most often cost interns approvals.
Key terms
| Term | Meaning |
|---|---|
| SLI | A measured number such as success rate or latency |
| SLO | The target for an SLI, like 99.9% successful requests |
| Error budget | How much failure the SLO allows |
| Golden signals | Latency, traffic, errors and saturation |
| Runbook | Step-by-step instructions for an alert |
| Postmortem | A blameless written review of an incident |
| Toil | Manual, repetitive work that should be automated |
| Canary release | Sending a new version to a small share of traffic first |
Common mistakes to avoid
- Alerting on every CPU spike instead of user-facing symptoms.
- Watching only average latency instead of p95 or p99.
- Writing postmortems that blame people instead of fixing causes.
- Never testing that backups can be restored.
- Making manual changes on servers that are not in code.
Want to go deeper? The free Kubernetes learning path has short modules, games and a certificate for each module.