Most on-call advice assumes a team large enough to spread a rotation across eight or ten people. For a five-person engineering team, that advice just means everyone is on call every other week.
What actually works at small scale is being ruthless about what pages a human at 2am. If an alert doesn't require an action only a human can take, it shouldn't be a page, it should be a ticket for the morning.
We also rotate the person who owns the alerting rules themselves, not just who's on call. Someone has to be responsible for deleting the noisy alert, or the rotation quietly gets worse every quarter.
- DevOps
- Team Process
MH
Written by Mustafa Halvadwala
Engineering Lead at HASANI.
