Home / Docs-Technical WhitePaper / 20-EFT.WP.Metrology.TimeBase v1.0
Chapter 14 — Quality Metrics, SLOs & Dashboards
One-line objective: With a unified SLI/SLO vocabulary, quantify the accuracy, stability, and availability of the time-base service, and close the loop from paneling to error budgets in an auditable way.
I. Scope & Targets
- Covered
- Publication quality of internal tau_mono and external ts, health of synchronization chains, harmonization of arrival time, and jitter control.
- Statistical windows & aggregation: rolling/sliced windows Delta_t, quantiles pXX, error budgets and burn-down (burn rate).
- Visualization & audit: TS.sli.* dashboard fields, thresholds tol_*, alerts, and rollback integration.
- Inputs
- Base observations: offset, skew, J, offset_1pps, delta_form, adev(tau), mdev(tau), hdev(tau).
- Event streams: source_switch, holdover, leap_event, smear_active, failover, drift_alarm.
- Policy: tol_offset, tol_skew, tol_J, tol_Tarr, target SLOs, and alerting strategies.
- Outputs
- Quality reports: report.sli, report.slo, burn_rate.*, q_score.
- Dashboards: overview / accuracy / stability / source health / events & rollback / compliance.
- Manifest mapping: manifest.time.sli.* and manifest.time.slo.*.
II. Terms & Variables
- Accuracy-related: offset (s), skew (ppm), J (s), offset_1pps (s), p50/p95/p99(x).
- Stability-related: adev(tau), mdev(tau), hdev(tau) with tau ∈ {1, 10, 100, ...}(s).
- Arrival consistency: dual-form T_arr and delta_form; threshold tol_Tarr (s).
- Availability & budgets: good(t) ∈ {0,1}, SLO_target ∈ (0,1], error budget EB, burn_rate.
- Panel fields: TS.sli.offset.p99, TS.sli.skew.p99, TS.sli.J.p99, TS.sli.adev.tau_*, TS.sli.delta_form.max.
- Units & dimensions: unit(x), dim(x), check_dim(expr) — all in SI and [T] coherent.
III. Axioms P514- **
- P514-1 (Window consistency): Compute all SLIs on tau_mono, publish on ts; attach uncertainties for offset/skew/J.
- P514-2 (Invariant gauges): Quantile definitions and ADEV/MDEV/HDEV estimators are fixed; no in-window estimator switching.
- P514-3 (Additive error budgets): Error budgets are chainable across windows; burn rate scales with window multiplicity for multi-horizon alerts.
- P514-4 (Dual-form mandatory): Whenever T_arr appears, record both forms and delta_form.
- P514-5 (Monotonic & traceable): non_decreasing(ts) and signed traceability (Ch. 13); dashboards reference signed, persisted metrics only.
- P514-6 (Unit conservation): Run check_dim(expr) prior to publication; normalize units before cross-field arithmetic.
IV. Minimal Equations S514- **
- S514-1 (Quantile definition)
p_q(x) = inf{ z : ( ∑_{j=1..N} 1{ x_j ≤ z } ) / N ≥ q },q ∈ (0,1)。 - S514-2 (Availability & SLO evaluation)
Define pass predicate
g_i = 1{ |offset_i| ≤ tol_offset ∧ |skew_i| ≤ tol_skew ∧ J_i ≤ tol_J ∧ delta_form_i ≤ tol_Tarr };
window SLI: SLI = ( ∑ w_i * g_i ) / ( ∑ w_i );window SLO pass: pass = 1{ SLI ≥ SLO_target }。 - S514-3 (Error budget & burn rate)
EB = max( 0 , SLO_target - SLI );
burn_rate_W = ( bad_time_W / ( (1 - SLO_target) * W ) ) with window length W。
Multi-scale burn: burn_rate = max( burn_rate_W1 , burn_rate_W2 )。 - S514-4 (RMS jitter & Allan family)
J_rms = sqrt( (1/N) * ∑ (offset_i - mean(offset))^2 );
adev(tau), mdev(tau), hdev(tau) follow Chapter 7’s definitions (phase differencing & overlap estimators)。 - S514-5 (Dual-form arrival difference)
delta_form = | ( 1 / c_ref ) * ( ∫ n_eff d ell ) - ( ∫ ( n_eff / c_ref ) d ell ) | with delta_form ≤ tol_Tarr。
The integration path gamma(ell) and measure d ell must be explicit.
V. Measurement & Publication Flow M50-14 (Collect → Aggregate → Evaluate → Visualize → Persist)
- Collect
From sync/observation stacks, gather offset/skew/J/1pps/adev/* and events; simultaneously pull both T_arr forms. - Aggregate
On window Delta_t, compute p50/p95/p99, J_rms, adev(tau); produce TS.sli.*. - Evaluate
Apply S514-2 and S514-3 to compute SLI, EB, burn_rate; emit report.slo. - Visualize
Render dashboards: overview / accuracy / stability / source health / events / compliance; stamp release_tag and TraceID. - Persist
Write manifest.time.sli.* and manifest.time.slo.*; sign and chain per Chapter 13.
VI. Contracts & Assertions C50-14*
- C50-141 (Quantile gates): |offset|_p99 ≤ tol_offset ∧ |skew|_p99 ≤ tol_skew ∧ J_p99 ≤ tol_J.
- C50-142 (ADEV gates): adev(tau_k) ≤ adev_max(tau_k) for target grid tau_k.
- C50-143 (Dual-form difference): max(delta_form) ≤ tol_Tarr.
- C50-144 (Availability SLO): SLI ≥ SLO_target; otherwise degrade or roll back.
- C50-145 (Burn-rate alerts): burn_rate ≥ br_hi → high-priority alert; burn_rate ≥ br_mid → pre-alert.
- C50-146 (Panel integrity): required_fields ⊆ TS.sli.* ∧ check_dim(all) = pass ∧ signature_valid = true.
- C50-147 (Monotonicity & gaps): violations( non_decreasing(ts) ) = 0 ∧ missing_rate(TS.sli.*) ≤ m_thr.
VII. Implementation Bindings I50-14*
- collect_sli(ds, sensors, window) -> sli_raw
- aggregate_sli(sli_raw, window, quantiles, taus) -> TS.sli.*
- compute_allan(ds, taus, mode) -> {adev, mdev, hdev}
- evaluate_slo(TS.sli.*, targets) -> {SLI, EB, burn_rate, pass}
- render_time_panels(TS.sli.*, report, layout) -> dashboard
- emit_sli_manifest(TS.sli.*, report) -> manifest.time.sli
- slo_alerts(report, policy) -> actions
- Invariants: unique(TraceID); mean(weights)=1 (if weighted); unit/dimension checks pass; delta_form exists and has been evaluated.
VIII. Cross-References
- Sources for servo & sync panels: Chapter 5.
- Robust offset/skew/jitter estimation: Chapter 6.
- Allan-family definitions & estimators: Chapter 7.
- Arrival-time metrology & path harmonization: Chapter 9.
- Compliance, contracts & signed releases: Chapter 13.
- Streaming ops & backpressure indicators: Chapter 11.
IX. Panel Design & Operational Notes
- Panel sections
- Overview: offset_p50/p95/p99, J_p99, SLI, burn_rate, holdover_duration_s.
- Accuracy: time series and quantile bands for offset and offset_1pps.
- Stability: multi-tau adev/mdev/hdev curves with target bands.
- Source health: pdop, cn0, source_switch_count, smear_active_flag.
- Events & rollback: leap_event, failover, drift_alarm, rollback_to(tag) records.
- Compliance: contract_fail_rate, signature_valid, audit_gap_seconds.
- Ops rules
- Multi-window burn: monitor a short window W1 (5 min) and a long window W2 (1 h); trigger on the maximum burn rate.
- Rollback loop: if C50-144 fails or burn_rate ≥ br_hi, enact gray degradation and rollback_to(last_good).
- Change control: any threshold or gauge change requires a new release_tag and re-signing of the manifest.
Summary
.availability, and stability, accuracy. Using these dashboards and SLOs, the time-base service achieves measurable, traceable, and rollbackable compliance in observation → aggregation → evaluation → visualization → persistence → alerting* interfaces, this chapter closes the loop from I50-14* gates, and C50-14 flow, M50-14* formulae, S514-* baselines, P514-WithCopyright & License (CC BY 4.0)
Copyright: Unless otherwise noted, the copyright of “Energy Filament Theory” (text, charts, illustrations, symbols, and formulas) belongs to the author “Guanglin Tu”.
License: This work is licensed under the Creative Commons Attribution 4.0 International (CC BY 4.0). You may copy, redistribute, excerpt, adapt, and share for commercial or non‑commercial purposes with proper attribution.
Suggested attribution: Author: “Guanglin Tu”; Work: “Energy Filament Theory”; Source: energyfilament.org; License: CC BY 4.0.
First published: 2025-11-11|Current version:v5.1
License link:https://creativecommons.org/licenses/by/4.0/