Linux Basic — What to Automate
What to Automate
Not every task benefits from automation. This page gives you a decision framework.
Good Automation Candidates
| Task Type | Why Automate |
|---|---|
| Daily backups | high frequency, critical, tedious |
| Log rotation | prevents disk-full disasters |
| Report generation | scheduled, data-driven |
| Scheduled deployments | off-hours, repeatable |
| Health checks / monitoring | constant vigilance |
| Cleanup of temp files | prevents creep over time |
Poor Automation Candidates
- One-offs that will never run again — just document them
- Tasks requiring judgment (approving a production deploy)
- Complex workflows with many edge cases — a human catches subtleties
- Anything you don’t fully understand yet — automate once it’s stable
The 5-Minute Rule
If the task takes < 5 minutes and runs < once per week, doing it manually is often cheaper than maintaining automation.
If it’s daily or hourly, even a 30-second task is worth automating — that’s 3 hours per year saved.
Automation ROI Calculation
time_saved = task_duration × frequency × (automation_lifespan_months / period)
cost = script_dev_time + maintenance_time
ROI = time_saved - cost
Positive ROI after 3 months? Automate it.
Next → cron vs at vs systemd
