Automatic database indexing using reinforcement learning
DOI: 10.21293/1818-0442-2026-29-2-116-128
Abstract: Relevance. Indexes have a significant impact on the performance of database management systems, and their automatic selection under changing workloads remains a challenging multicriteria problem. Purpose of the study. The purpose of the study is to develop a query-oriented method for automatic index management based on reinforcement learning for operation in a high-load environment. Methods. The study employs the following tools and approaches: formulation of the problem as a Markov decision process, comparative analysis of modern indexing approaches, the query optimizer's costing mechanism, DBMS operational telemetry, and validation of results on TPC-H workloads. Scientific novelty. An automatic indexing method based on reinforcement learning is proposed, in which the indexing action is selected in the local context of the query being optimized, while the optimizer's surrogate reward is separated from the actual reward calculated from the statistics of executed queries. Results. In sequential, concurrent, and synthetic scenarios, the reinforcement learning strategy outperformed baseline indexing in terms of query execution time and aggregate performance metrics. Practical significance. The proposed method can be used as a component of a safe automatic indexing loop in industrial DBMSs.
Keywords: reinforcement learning, automatic indexing, query-oriented index selection, dynamic index tuning, physical database design, query execution statistics, query cost estimation
For citation:
Kolomiets V. V. Automatic database indexing using reinforcement learning. Doklady Tomskogo gosudarstvennogo universiteta sistem upravleniya i radioelektroniki, 2026, vol. 29, no. 2, pp. 116–128. DOI: 10.21293/1818-0442-2026-29-2-116-128
Authors and copyright holders:
- Kolomiets V. V. , Siberian Transport University
- 1. Agrawal S., Bruno N., Chaudhuri S., Narasayya V. AutoAdmin: Self-Tuning Database Systems Technology. IEEE Data Engineering Bulletin, 2006, vol. 29, no. 3, pp. 7–15.
- 2. Chaudhuri S., Narasayya V. Self-Tuning Database Systems: A Decade of Progress. VLDB '07: Proceedings of the 33rd international conference on Very large data bases. Vienna Austria, ACM, 2007, pp. 3–14.
- 3. Wu Y., Zhou X., Zhang Y., Li G. Automatic Index Tuning: A Survey. IEEE Transactions on Knowledge and Data Engineering, 2024, vol. 36, no. 12, pp. 7657–7676.
- 4. Lan H., Bao Z., Peng Y. An Index Advisor Using Deep Reinforcement Learning. Proceedings of the 29th ACM International Conference on Information and Knowledge Management (CIKM), New York, ACM, 2020, pp. 2105–2108.
- 5. Shi J., Cong G., Li X.-L. Learned Index Benefits: Machine Learning Based Index Performance Estimation. Proceedings of the VLDB Endowment, 2022, vol. 15, no. 13, pp. 3950–3962.
- 6. Sutton R.S., Barto A.G. Reinforcement Learning: An Introduction. 2nd ed. Cambridge, MA, MIT Press, 2018, 552 p.
- 7. Mnih V., Kavukcuoglu K., Silver D. et al. Human-level Control through Deep Reinforcement Learning. Nature, 2015, vol. 518, pp. 529–533.
- 8. Kossmann J., Kastius A., Schlosser R. SWIRL: Selection of Workload-aware Indexes Using Reinforcement Learning. Proceedings of the 25th International Conference on Extending Database Technology (EDBT), Edinburgh, UK, 2022, pp. 155–168.
- 9. Wu W., Wang C., Siddiqui T. et al. Budget-aware Index Tuning with Reinforcement Learning. Proceedings of the 2022 International Conference on Management of Data (SIGMOD), USA, ACM, 2022, pp. 1528–1541.
- 10. Wang T., Yoneki E. IA2: Leveraging Instance-Aware Index Advisor with Reinforcement Learning for Diverse Workloads. Proceedings of the 4th Workshop on Machine Learning and Systems (EuroMLSys), New York, ACM, 2024, pp. 10–17.
- 11. Wang T., Yoneki E. AutoIndexer: A Reinforcement Learning-Enhanced Index Advisor Towards Scaling Workloads. arXiv e-prints, 2025, art. no. arXiv:2507.23084.
- 12. Basu D., Lin Q., Chen W. et al. Regularized Cost-Model Oblivious Database Tuning with Reinforcement Learning. Transactions on Large-Scale Data– and Knowledge-Centered Systems, Lecture Notes in Computer Science, Cham, Springer, 2016, vol. 28, pp. 96–132.
- 13. Sharma A., Schuhknecht F.M., Dittrich J. The Case for Automatic Database Administration using Deep Reinforcement Learning. arXiv e-prints, 2018, art. no. arXiv:1801.05643.
- 14. Licks G.P., Couto J.C., Miehe P. et al. SMARTIX: A Database Indexing Agent Based on Reinforcement Learning. Applied Intelligence, 2020, vol. 50, no. 8, pp. 2575–2588.
- 15. Sadri Z., Gruenwald L., Leal E. Online Index Selection Using Deep Reinforcement Learning for a Cluster Database. 2020 IEEE 36th International Conference on Data Engineering Workshops (ICDEW), Dallas, TX, IEEE, 2020, pp. 158–161.
- 16. Sadri Z., Gruenwald L., Leal E. DRLindex: Deep Reinforcement Learning Index Advisor for a Cluster Database. Proceedings of the 24th International Database Engineering & Applications Symposium (IDEAS), Seoul, Republic of Korea, ACM, 2020, article 11, pp. 1–8.
- 17. Sharma V., Dyreson C., Flann N. MANTIS: Multiple Type and Attribute Index Selection using Deep Reinforcement Learning. Proceedings of the 25th International Database Engineering & Applications Symposium (IDEAS), New York, USA, 2021, pp. 56–64.
- 18. Yan Y., Yao S., Wang H., Gao M. Index Selection for NoSQL Database with Deep Reinforcement Learning. Information Sciences, 2021, vol. 561, pp. 20–30.
- 19. Sharma V., Dyreson C. Indexer++: Workload-Aware Online Index Tuning with Transformers and Reinforcement Learning. Proceedings of the 37th ACM/SIGAPP Symposium on Applied Computing (SAC), New York, ACM, 2022, pp. 372–380.
- 20. Welborn J., Schaarschmidt M., Yoneki E. Learning Index Selection with Structured Action Spaces. arXiv e-prints, 2019, art. no. arXiv:1909.07440.
- 21. Wu Y., Zhang Y., Li N. Index Advisor via DQN with Invalid Action Mask in Tree-Structured Action Space. Advanced Data Mining and Applications. ADMA 2022. Lecture Notes in Computer Science, Cham, Springer, 2022, vol. 13781, pp. 434–445.
- 22. Zhou W., Lin C., Zhou X., Li G. Breaking It Down: An In-depth Study of Index Advisors. Proceedings of the VLDB Endowment, 2024, vol. 17, no. 10, pp. 2405–2418. DOI:10.14778/3675034.3675035.
Last issue
Editorial office address
Executive Secretary of the Editor’s Office
Editor’s Office: 40 Lenina Prospect, Tomsk, 634050, Russia
Phone / Fax: + 7 (3822) 701-582
Executive Secretary of the Editor’s Office
Editor’s Office: 40 Lenina Prospect, Tomsk, 634050, Russia
Phone / Fax: + 7 (3822) 51-21-21 / 51-43-02
Subscription for updates
