Alerts Don't Suck YOUR Alerts Suck!
summary
An argument that alerts fail because they are poorly designed not inherently bad, with concrete fixes: avoid one-off CPU or vendor-example rules, treat FYI alerts are trash, threshold on baseline and duration, prefer user-experience signals, and bake in automation (example: automatic disk-clean reduced tickets by over 300 tickets and saved about $80,000). Real costs guide priorities, require runbooks, verify alerts happen in the wild (1,544 accidental tickets shows the cost of bad design), and review alerts regularly.
AI-written summary · about
chapters (12)
- 0:00Introduction and personal story
- 1:15Talk overview and speaker intro
- 2:37What alerts really are
- 3:20Four key points about alerts
- 6:14Five lessons on alerting
- 11:31Monitoring vs observability
- 12:56When is something an alert?
- 14:01Four common alert pitfalls
- 17:51The power of automation
- 21:47Talking to the right people
- 26:25Proving alert value to business
- 29:37Maintaining alerts over time
