Library / First Principles Framework (FPF) - Core Conceptual Specification
Jump to passage
In this reading

Link to current text

Published source confirmed at last check

Source changed 2026-10-03 10:39:28 UTC · snapshot created 2026-10-03 10:40:04 UTC · last check 2026-10-03 10:55:17 UTC

A.19.CN:8.2 - Software/SRE line — Latency CN‑frame (SRE_Prod_Cluster_EU_2026)

  • cs_basis: P50Latency[ms] (↓), P99Latency[ms] (↓), Load[req/s]
  • chart: client vantage, trace sampler v4; MethodDescription#HTTP_probe_v4
  • Before comparing a client end-to-end reading with a server processing reading, resolve their actual Characteristic and observation basis. Both may use ms yet answer different questions. Keep separate coordinates or obtain readings for the same declared characteristic and basis; a unit conversion alone cannot supply that match.
  • normalization: monotone time‑warp compensation for collector skew; invariant = percentile order
  • comparability: normalization‑based (UNM) with declared normalization
  • aggregation: use a max-of-mins latency fold only when the declared service model justifies it; a WLNK bound needs its own bottleneck model. Neither follows from latency percentiles alone.
  • RSG hook: DeployerRole.Active gated if P99 < declared SLO over the admission window.