Home / Benchmarks / Latency

API Latency Benchmark

Per-day p50, p95, p99 latency for the FlashAlpha analytics REST API over the trailing 30 days, plus a summary row showing the arithmetic mean of those daily percentiles. Sourced directly from the production request log table - no cherry-picking, no synthetic load.

As of pending first refresh

Window 30 days

Source production ApiRequestLogs

Benchmark data refresh pending

The first nightly refresh has not yet populated wwwroot/data/latency-benchmark.json. Once the refresh job runs against /api/admin-dashboard/metrics/latency, this page will render a daily-aggregate table - one row per day in the trailing window, plus a summary row with the mean of the daily percentiles.

Why no seeded numbers? Publishing latency commits to keeping the numbers honest. Rather than seed with synthetic values that would have to be cleared on first real refresh, the page renders blank until real data arrives.