Skip to main content

Server Timing

Server Timing​

The Embrace SDK automatically reads backend performance metrics sent by your server via the Server-Timing HTTP response header on the navigation request. This gives you visibility into server-side latency contributors — database queries, cache lookups, CDN processing — alongside your frontend telemetry.

How Server Timing Instrumentation Works​

When the page finishes loading, the SDK reads performance.getEntriesByType('navigation')[0].serverTiming and emits one log per PerformanceServerTiming entry. If the navigation entry has no server timing data, the SDK no-ops silently.

No changes to your frontend code are required. Visibility depends on your server emitting the header:

Server-Timing: db;dur=78, cache;desc="HIT";dur=0, render;dur=163

Known Limitation​

serverTiming is only populated for same-origin navigations, or cross-origin navigations where the server includes the Timing-Allow-Origin response header. When neither condition is met, the array is empty and the SDK no-ops silently.

Data Captured​

For each Server-Timing entry, the SDK captures:

  • name — the metric name (e.g. db, cache, cfEdge)
  • description — the desc parameter value; in practice carries high-cardinality data such as cache status, country codes, and request IDs
  • duration — milliseconds; 0 when the server omits the dur parameter (common for metadata-only entries)

Integration with Other Features​

  • Server timing logs are associated with the current session and page navigation
  • Custom dashboards can be built to monitor duration percentiles per metric name, and filter by description values
  • Alerts can be set on server-side latencies exceeding specific thresholds