Tracing & APM
distributed traces with full request context, retained 90 days by default
Logs
structured ingestion, sub-second search, no per-GB surprise bills
Metrics
Prometheus-compatible, native to dql, alerts in seconds not minutes
Replay & Debug
re-run any failing request in an isolated sandbox
Alerts & On
call — escalations, fatigue control, paid SLA on missed pages
SLOs
error budgets with weekly burn-rate forecasts and auto-generated reports