[milindsrivastava1997]

Summary

Regenerates rqe-optimizer/results/e2e-rate10/ (classic, multigroup and all at shared = 1, 10 samples/s) on main. That picks up #213's cost table and curves, #215's one-scrape instant RQEs, and #216's one RQE per schema query. These are the plans for the end-to-end run.

• Tables: export_autosketch_eval_table.py --synthetic --samples-per-sec 10, without --keep-duplicate-queries.
• Results: --runs 3 --no-chosen. Plans: --runs 1, then extract_autosketch_plans.py.
• README: RQE counts, costs, classic deployments, the change from the previous 10/s plans, ASAPQuery conversion, and Reproduce.

RQEs

Workload	RQEs at 200/s (§6.3)	Duplicate queries	Below the curves	RQEs at 10/s, all served
classic	42	0	0	42
multigroup	62	10	2	50
all	104	10	2	92

Every chosen accuracy is measured, and no Hydra grid is chosen. Every e2e query's RQEs are served by one deployment in every plan.

Cost (cheapest plan, ASAP, CPU weights), previous 10/s plans → now

Workload	CPU (vCPU)	Memory (GiB)	Fargate ($/h)	Deployments	Rolled up
classic	0.309 → 0.318	15.3 → 10.7	0.0787 → 0.0585	6 → 7	0 → 0
multigroup	0.0950 → 0.0911	1.34 → 1.33	0.00872 → 0.00846	9 → 9	41 → 33
all	0.404 → 0.409	16.7 → 12.0	0.0874 → 0.0670	15 → 16	41 → 33

• Memory falls ~30% and Fargate ~25%. With study: denser KLL/DDSketch grid and Hydra W 65536; §6.3 rerun #213's denser grid, plans pick DDSketch α = 0.1.
• CPU rises ~3% on classic and all. sum by (job) (sum_over_time) no longer shares the instant sum by (job) deployment, which is now a 100 ms window.
• The tightest feasible latency bounds are unchanged: classic 21 ms, multigroup and all 490 ms.

ASAPQuery

--planner saved needs --data-ingestion-interval-ms 100 on these tables, the scrape interval. With ASAPQuery main (ProjectASAP/ASAPQuery#866), every ASAP plan converts: classic PromQL 12/12, multigroup SQL 8/8, all SQL 8/8 and PromQL 8/8.

Verification

• An independent re-plan in another session gave identical tables (classic, all, multigroup) and identical plans in all 120 (method, weights, kind, SLA) combinations: classic 48, all 36, multigroup 36.
• Runs were on a 16-core desktop, not an idle cluster node, so planning_secs are indicative only.

🤖 Generated with Claude Code