docs(eval): Fig 7 query-latency CDF — warm p50 18.3ms/p99 20.0ms (real gct) - #493
Merged
Merged
Conversation
… (real gct) Fold the measured query-latency results into the eval plan: warm sketch-answered PromQL is production-usable (overall p50 18.25ms / p99 20.03ms across a 599-query mix @15 QPS on the real cold-OFF stack, guard-verified). Bimodal CDF — lossless sum ~1.6ms vs 691-series DDSketch quantile reconstruction ~18-21ms, tight tails. Cold-fallback arm honestly marked blocked (cold ship rides the disabled control channel); warm-only, single-node loopback (no network RTT). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Folds the measured query-latency (Fig 7) results into
docs/evaluation-plan-figures.md.Warm-tier sketch PromQL is production-usable. Real Google-2019 trace, cold-OFF stack, wall-clock-anchored, 599-query mix @15 QPS, guard-verified before timing (DDSketch read returned 691 real warm series;
sum=exact GT 216.3535; alldata_source=asap_query; 0 empties/errors).Bimodal CDF: lossless
sum~1.5–2.3 ms; 691-series DDSketch quantile reconstruction ~18–21 ms; tight tails.Honest caveats: cold-fallback arm attempted but blocked (cold ship rides the disabled control channel → MinIO empty, old-ts queries still served warm) → warm-only reported, not faked. Single-node loopback (server-side latency only, no network RTT).
count_over_timeexcluded (does not resolve on warm path).Eval artifacts on branch
feat/query-latency-cdf:datasets_eval/latency/.🤖 Generated with Claude Code