What is on-call?
On-call means that a named person is responsible, for a set period, for responding to alerts and incidents, including outside working hours. Teams share the duty through a rotation, for example one week each, with a secondary as backup, overrides for holidays, and an escalation policy if the primary does not respond.
Why on-call matters
Someone has to answer when production breaks at night. Without a clear rota, alerts go to a channel everyone assumes someone else is watching, or to the one person who always answers until they burn out.
A healthy rota keeps the load fair and bounded. Google's SRE book suggests no more than two incidents per 12-hour on-call shift, so that each one can be handled properly and followed up.
How to measure on-call
weeks per year as primary (weekly rota) = 52 ÷ people in the rota pages per shift = pages ÷ shifts
Worked example
Five engineers share a weekly primary rotation. Last quarter (13 weeks) the primary handled 52 pages.
- Each engineer is primary for 52 ÷ 5 = 10.4 weeks a year.
- Pages per weekly shift = 52 ÷ 13 = 4.
- Four pages a week is sustainable if they are real and well documented; if the number grows, fix noisy alerts before adding people.
Result: Track pages per shift and out-of-hours pages over time; they are the best early warning that on-call is becoming a burnout risk.
Common mistakes
- No backup. A single on-call person with no secondary means a flat phone battery becomes an outage.
- On-call without authority. The person paged must be able to roll back, restart and scale without waiting for approval.
- Paging for non-urgent alerts. Night pages should need action now. Everything else waits for working hours.
- No handover. Ongoing issues and recent changes should be handed to the next person at the change of shift.
How Vigil handles on-call
- Vigil's on-call module has rotations, overrides and an ICS feed for calendars; the on-call channel pages whoever is on duty by email, SMS or phone call.
- Escalation sends an incident to a second set of channels if it is not acknowledged within the minutes you choose.
- Acknowledge or resolve from Slack buttons, the chat bot or the app, which installs as a PWA with push notifications; urgent Pushover alerts repeat until acknowledged.
On-call: frequently asked questions
How long should an on-call shift be?
One week is common because it limits handovers; some teams use 12-hour shifts split across time zones so nobody is paged at night.
Should on-call be paid?
That depends on your employment terms and local law. Many companies pay an allowance or give time off in lieu, especially for out-of-hours pages.
How many people does a rotation need?
Enough that everyone gets long breaks between shifts and a separate secondary exists. With very few people, weekly duty comes round so often that it wears people down.