Market-journal reconciliation benchmark

2026-10-21 10:06:56 -04:00
parent 1f11210b5b
commit 687b628424
4 changed files with 78 additions and 1 deletions
@@ -46,6 +46,8 @@ Observed in the development Docker environment:

 These numbers are a reproducible **baseline**, not acceptance SLOs: host CPU/storage and production concurrency were not selected, outputs include synthetic article payloads, warm caches apply, one transaction amortizes insertion fsync, and real validation/startup/HTTP/market workloads are excluded. They do not establish a supported production envelope.

+The synthetic market-journal harness [benchmark](../scripts/benchmark-market-reconciliation.ts), available as `npm run benchmark:market -- 10000`, separately measures bounded reconciliation of terminal order/listing records in its own temporary SQLite directory. On the same development Docker environment, 10,000 orders plus 10,000 listings completed in 157 reconciliation passes in 772 ms (about 25,903 records/s), with pass-latency p50/p95/max of 4.57/9.13/20.04 ms and a 7,180,288-byte SQLite file. This dataset intentionally avoids active-order delivery and listing validation branches; it measures local keyset/SQLite overhead only. It is not a target-hardware result or throughput/SLO claim and excludes concurrent HTTP traffic, event-loop interference from other work, PostgreSQL, chain validation and adversarial workloads.
+
 ## Remaining gates

 M2 still needs target-hardware/data/SLO decisions, populated long-journal/large-event benchmarks, bounded worker/event-loop latency under load, production financial database/schema and measured recovery/retention/storage growth. The new scheduler bounds work per pass, but does not yet provide queue isolation, multi-worker coordination, or a demonstrated production throughput envelope. Ledger PoW replay and DAG state memory still scale with retained history; indexed queries do not solve those costs. Startup validation checks undo-record counts but not every possible future undo mutation against a second reconstructed undo model; comprehensive corruption/fuzz coverage and independent review remain required.