Alerts are the notification mechanism of Cloud Native Observability. They watch signals collected from metrics, logs, and traces and trigger when predefined conditions are met—such as a spike in error rate or CPU usage crossing a threshold. When an alert fires, it notifies on-call engineers or incident channels, often with context, severity, and links to runbooks, and can be routed through tools that manage who gets notified and when (for example, Slack, PagerDuty, or email). Proper alerting enables fast, targeted responses and helps avoid fatigue by tuning thresholds and using suppression windows and multi-signal correlation. Storing logs, configuring proxies, or simply measuring latency are separate activities. Logs are about recording events, proxies handle traffic routing, and latency measurement is part of data collection, not the alerting action itself.