Metrics

Metrics provide the aggregate view of platform behaviour over time. They are useful for capacity planning, provider comparison, and alerting.

Core counters

Request volume, success rate, error rate, and fallback frequency are the baseline counters most teams should track. These values help explain whether a routing policy is stable in production.

Latency series

Latency should be tracked by percentile, category, and provider so routing tradeoffs are visible. Median latency alone rarely captures the spikes that matter to production traffic.

Evaluation aggregates

Meridias can expose benchmark rollups alongside runtime metrics so teams can compare planned expectations with live results. This creates a common view across product, platform, and operations teams.

Alerting posture

Alerts should be tied to meaningful thresholds and routed to the teams that can act on them. Useful alerts are specific enough to drive action without becoming background noise.

Next step

Use Provider Health and Logs & Tracing together when metrics show sustained degradation.