DOT: Dynamic Knob Selection and Online Sampling for Automated Database Tuning

Yifan Wang, Debabrota Basu, Pierre Bourhis, Romain Rouvoy, Patrick Royer · 2026

Read DOT: Dynamic Knob Selection and Online Sampling for Automated Database Tuning by Yifan Wang, Debabrota Basu, Pierre Bourhis, Romain Rouvoy, Patrick Royer (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-15540.txt.

Description

Wang and co-authors present DOT, an automated tuner for database systems that picks key knobs and samples settings online. Most tuning parameters matter little, but finding the vital few used to cost warm-up runs and expert time. DOT prunes weak knobs with recursive feature elimination and cross-validation. A likelihood ratio test balances exploration against exploitation during search. Bayesian optimization then tunes on the fly, with no warm-up phase. Tests match or beat top tuners while cutting tuning overhead sharply.

Catalog record

CREATORYifan Wang, Debabrota Basu, Pierre Bourhis, Romain Rouvoy, Patrick Royer
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE69733
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.15540

Browse the full library · About the Tower of Babel