Sending an alert from production to an operator, or a developer, or a customer support person is an expensive proposition. That person has to put down what they’re doing to try to figure out what’s going on and doing that can sometimes be really hard where distributed systems are concerned.
CASE is a minmum bar we set for all system alerts that helps in this regard. Note: This is a hard problem that never goes away and requires constant vigilance!
- Context heavy
- System based
Important note 1: Alerts that satisfy this criteria today, may fail tomorrow. Re-evaluate alerts periodically to ensure that they are adding value to your practice, and team