Adapted OpenSearch Big5
An adapted search workload covering text, sorting, aggregations and ranges.
Readiness reviewed 2026-10-04. Local fixture and offline checks are development evidence; they do not establish benchmark performance.
Implemented
- Upstream operations and corpus preparation are pinned; all 50 query operations are registered.
- Tokenizer and query semantics are mapped, with initial Elasticsearch query translations.
- Native Elasticsearch aggregation and sorted-browse paths passed a live 5,000-document fixture check; the canonical Big5 workload is still unmeasured.
Before results can be published
- Freeze and verify the canonical dataset.
- Complete engine adapters and query translations, then qualify correctness against an independent oracle.
- Run the adapted workload on matched resources and review measurements before publication.
Measurements to report
Latency, throughput and correctness by query family
Coverage is targeted across all eight engine configurations. Adapters, correctness checks and comparable measurements must be qualified before this suite is published. Unsupported operations will remain explicit coverage gaps.
Upstream workload ↗ · Platform coverage and cold-start status →