On-Call Fundamentals and Rotation Models
What you'll learn: How to structure engineer availability for incident response through sustainable rotation models that balance system reliability with human well-being.
What Is On-Call?
On-call duty means being available to respond to production incidents outside regular working hours. When alerts fire—like those multi-window, multi-burn-rate alerts you've configured—someone needs to be ready to investigate and mitigate. On-call rotations distribute this responsibility across a team so no single person carries the burden continuously.
Think of it like lifeguards at a beach: the water doesn't care what time it is, so you need scheduled coverage with clear handoffs.
Common Rotation Models
Primary/Secondary Rotation
One engineer is the primary responder; a secondary backs them up if the primary doesn't acknowledge within a time limit (typically 5-15 minutes). This provides redundancy without immediate double-paging.
Follow-the-Sun Model
Teams in different time zones hand off on-call duties so coverage happens during business hours for each region. An engineer in London covers daytime Europe, then passes to someone in California for their daytime—minimizing "middle of the night" pages.
Weekly vs Daily Rotations
Longer rotations (one week) reduce context-switching and handoff overhead. Shorter rotations (24 hours) spread disruption more thinly but require more handoffs.
Balancing Coverage with Well-Being
- Compensation: On-call work deserves extra pay or time off
- Alert quality matters: Actionable, low-noise alerts (remember alert fatigue?) reduce unnecessary wake-ups
- Runbooks: Every alert needs clear remediation steps
- Rotation fairness: Distribute difficult shifts equitably; avoid perpetually assigning nights to junior engineers
Key Takeaway: On-call rotations ensure 24/7 incident response while preserving engineer health through structured schedules, backup coverage, and a relentless focus on reducing toil and false alarms.