A notification that a system health threshold has been crossed.
An alert is a trigger that a specific change in a system's health has occurred, indicating potential issues. When an alert fires, the system notifies operators so they can take action to remediate or fix the underlying problem.
Alerts are a foundational component of any observability strategy. They are typically configured against metrics, log patterns, or trace error rates that exceed a defined threshold. Good alerting practice balances sensitivity — catching real problems early — with specificity, avoiding alert fatigue from too many false positives. Modern alerting systems support multi-condition rules, deduplication, and automatic grouping to reduce noise. obseria.io provides AI-assisted alert configuration that learns from historical anomalies to suggest thresholds automatically.
See Alert in action
obseria.io gives you full-stack observability — logs, metrics, traces, and AI-powered root cause analysis.