[ SRE_ARENA ]
ALERTING & ON-CALL

Alert Fatigue Will Burn Out Your Best Engineers.

Visualization: 1. The Psychology of Alerting (Why We Alert)
DOWNLOADING CONCEPT: 1. THE PSYCHOLOGY OF ALERTING (WHY WE ALERT)...
  • ↳The fastest way to burn out an engineering team is a poorly configured alerting system.
  • ↳If an engineer's phone buzzes 50 times a night for non-actionable issues (e.g., a background job took 2 seconds longer than usual), psychological Alert Fatigue sets in.
  • ↳The human brain naturally learns to ignore the pager.
  • ↳When a real, catastrophic Sev-1 outage occurs the next week, the engineer will subconsciously assume it's just 'another noisy alert' and roll over to go back to sleep.
  • ↳Alerting is a psychological tool: every single page must represent an immediate, undeniable threat to the business that requires human intervention.
▲ LIVE INCIDENTAn engineer was paged 45 times last night for the same database issue.
sre@arena:~/alerting
sre@arena:~$ amtool alert --alertmanager.url=http://alertmanager:9093