Alerting generates and delivers notifications when monitored functional or nonfunctional conditions indicate failure, degradation, threshold violations, or changing real-world behavior. section 9.5
Implement alerting as an operational component with explicit support responsibilities:
- Capture relevant operational events, including database connections, user logins, data updates, and model decisions. section 9.5
- Monitor both nonfunctional behavior and ML-functional behavior, because changing entities, economics, or technology can invalidate formerly useful predictions. section 9.5
- Route informative alarms to technical support groups, preferably before an operational failure becomes visible to users. section 9.5
- Balance notification coverage against metric-computation load, logging volume, storage cost, and retention requirements. section 9.5
Pipeline and database alerts
Connect data tests to monitoring so a database becoming unreachable, a pipeline step running out of order, or extreme values can trigger rapid notification before the ML team must intervene. section 5.4.3
