SQL-Native Vector Search at Billion Scale in Presto

Summary: Presto Vector Search recasts partitioned vector search as relational algebra, making indexes pluggable aggregates and enabling SQL-native pre-filtering, joins, optimizer-driven zero-shuffle execution, and a unified declarative/ expert API. At billion scale, it delivers 96.6% Recall@1 in under 2.5 minutes, with 95–99% CPU and ~7,000× network reductions. (summarized by gpt-5.6-luna on Aug 28 2026)

Paper ID: hb48f40e880b34423
Venue: VLDB
Year: 2026
Pagerank: 4.9793485e-05
Overall Rank: 10,929 | 26.52%
DOI: 10.14778/3827998.3828025

Incoming Non-self Citations Over Time

No non-self incoming citations found for this paper in this database.

Authors

1. Zhichen Xu (Meta)
2. Ge Gao (Meta)
3. Ke Wang (Meta)
4. Aakash Deep (Meta)
5. Junjie Qi (Meta)
6. Jort Gemmeke (Meta)
7. Amit Dutta (Meta)
8. Xiaoxuan Meng (Meta)
9. Trang Nguyen (Meta)
10. Jiayin Yan (Meta)
11. Xiao Du (Meta)
12. Vivek Gaur (Meta)
13. Kaushik Ravichandran (Meta)
14. Vishal Gandhi (Meta)

BibTeX Citation

@article{xu_vldb26,
        title = {{SQL-Native Vector Search at Billion Scale in Presto}},
        author = {Xu, Zhichen and Gao, Ge and Wang, Ke and Deep, Aakash and Qi, Junjie and Gemmeke, Jort and Dutta, Amit and Meng, Xiaoxuan and Nguyen, Trang and Yan, Jiayin and Du, Xiao and Gaur, Vivek and Ravichandran, Kaushik and Gandhi, Vishal},
        journal = {PVLDB},
        series = {{VLDB} '26},
        volume = {19},
        number = {12},
        pages = {4182--4194},
        doi = {10.14778/3827998.3828025},
        url = {https://doi.org/10.14778/3827998.3828025},
        year = {2026}
}

Incoming Citations (Sorted by Pagerank)

Showing 1 of 1 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
10,944	From Presto to Prestissimo: A Velox-Powered Modernization Journey	2026	VLDB	4.9793485e-05

Outgoing Citations (Sorted by Pagerank)

Showing 11 of 11 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
194	Milvus: A Purpose-Built Vector Data Management System	2021	SIGMOD	0.00025636725
341	AnalyticDB-V: A Hybrid Analytical Engine Towards Query Fusion for Structured and Unstructured Data	2020	VLDB	0.00020539791
804	RaBitQ: Quantizing High-Dimensional Vectors with a Theoretical Error Bound for Approximate Nearest Neighbor Search	2024	SIGMOD	0.00013832333
916	PASE: PostgreSQL Ultra-High-Dimensional Approximate Nearest Neighbor Search Extension	2020	SIGMOD	0.00013094482
1,424	Velox: Meta's Unified Execution Engine	2022	VLDB	0.00010719232
2,085	SingleStore-V: An Integrated Vector Database System in SingleStore	2024	VLDB	9.0709364e-05
7,432	Cracking Vector Search Indexes	2025	VLDB	5.5292193e-05
7,564	SIEVE: Effective Filtered Vector Search with Collection of Indexes	2025	VLDB	5.4955249e-05
8,848	HARMONY: A Scalable Distributed Vector Database for High-Throughput Approximate Nearest Neighbor Search	2026	SIGMOD	5.2646236e-05
8,849	VStream: A Distributed Streaming Vector Search System	2025	VLDB	5.2646236e-05
10,944	From Presto to Prestissimo: A Velox-Powered Modernization Journey	2026	VLDB	4.9793485e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	12,346	Vectorizing an In Situ Query Engine	2016	SIGMOD
2	9,113	Presto’s History-based Query Optimizer	2024	VLDB
3	8,849	VStream: A Distributed Streaming Vector Search System	2025	VLDB
4	10,409	An In-Depth Study of Filter-Agnostic Vector Search on a PostgreSQL Database System: [Experiments & Analysis]	2026	SIGMOD
5	2,060	Similarity search in the blink of an eye with compressed indices	2023	VLDB
6	10,910	TEngineDB-V: An OLAP-Native Vector Search System for Large-k Workloads at Tencent	2026	VLDB
7	9,031	Cost-Effective, Low Latency Vector Search with Azure Cosmos DB	2025	VLDB
8	9,955	Turbocharging Vector Databases using Modern SSDs	2025	VLDB
9	2,085	SingleStore-V: An Integrated Vector Database System in SingleStore	2024	VLDB
10	7,953	Fast Vector Search in PostgreSQL: A Decoupled Approach	2026	CIDR