All docs

Edge Standards

Why we require sub-100ms latency for all Heritage sites.

Orchestration is the second pillar of the Vellum Standards. Edge Standards define why we require sub-100ms latency for all Heritage-tier sites and how we measure it.

The 100ms Rule

For a site to be classified as Heritage under Vellum & Code:

  • P95 latency from our edge to your origin must be under 100 ms.

This is not arbitrary. Sub-100ms response times:

  • Keep interactions feeling instant
  • Reduce the risk of timeouts and retries
  • Align with modern expectations for API and page delivery
  • Support the rest of the Sentinel Protocol (headers, SSL) without adding noticeable delay

How We Measure Latency

  1. Probes — Scribe Sentinel runs probes from multiple edge locations.
  2. Round-trip time — We measure the time from request send to first byte (or full response) received.
  3. Percentiles — We report P50, P95, and P99. The P95 value is used for Heritage grading.
  4. Target — P95 < 100 ms → within standard; P95 ≥ 100 ms → below Heritage tier until improved.

Meeting the Standard

  • Use a CDN or edge runtime — Serve static assets and, where possible, API routes from the edge.
  • Optimize cold starts — If you use serverless, keep cold start within budget or use always-on edge functions.
  • Choose regions wisely — Place origins and edge nodes close to your users.
  • Monitor continuously — Use The Sentinel Protocol so that latency regressions are caught early.

Related