| 本文已被：浏览 1729次 下载 1420次 |
码上扫一扫！ |
|
|
| 智能查询优化算法研究综述 |
|
何家豪1,2, 王嘉辰1,2, 王晓1,2, 张喜盈3,2, 李翠平1,2, 陈红1,3
|
|
1.数据工程与知识工程教育部重点实验室(中国人民大学), 北京 100872;2.中国人民大学 信息学院, 北京 100872;3.数据库与商务智能教育部工程研究中心(中国人民大学), 北京 100872
|
|
| 摘要: |
| 查询优化是数据库系统中至关重要的环节, 查询优化器通过找出一条查询语句对应的最佳查询计划来减少查询执行的代价. 传统优化器依赖固定规则或简单启发式算法加工并筛选候选计划. 然而随着实际应用中关系模式和查询逐渐复杂, 传统的查询优化器已经难以满足应用需求. 智能查询优化算法将机器学习技术应用到查询优化领域, 通过学习查询计划与复杂关系模式的特征来协助传统优化器完成查询优化. 此类算法在代价模型、连接优化、计划生成和查询改写等方面都提出了创新有效的解决方案. 梳理上述4类智能查询优化算法近年来的研究成果和发展脉络, 并对智能查询优化未来的研究方向进行展望, 希望研究者可以全面了解智能查询优化算法的研究现状, 以助于其后续科研工作的开展. |
| 关键词: 查询优化 人工智能 强化学习 数据库系统 数据管理 |
| DOI：10.13328/j.cnki.jos.007449 |
| 分类号:TP311 |
| 基金项目:国家重点研发计划(2023YFB4503600); 国家自然科学基金(U23A20299, U24B20144, 62172424, 62276270, 62322214) |
|
| Survey on Learned Query Optimization Algorithms |
|
HE Jia-Hao1,2, WANG Jia-Chen1,2, WANG Xiao1,2, ZHANG Xi-Ying3,2, LI Cui-Ping1,2, CHEN Hong1,3
|
|
1.Key Laboratory of Data Engineering and Knowledge Engineering of the Ministry of Education (Renmin University of China), Beijing 100872, China;2.School of Information, Renmin University of China, Beijing 100872, China;3.Engineering Research Center of Database and Business Intelligence of the Ministry of Education (Renmin University of China), Beijing 100872, China
|
| Abstract: |
| Query optimization is a critical component in database systems, where execution costs are minimized by identifying the most efficient query execution plan. Traditional query optimizers typically rely on fixed rules or simple heuristic algorithms to refine or select candidate plans. However, with the growing complexity of relational schemas and queries in real-world applications, such optimizers struggle to meet the demands of modern applications. Learned query optimization algorithms integrate machine learning techniques into the optimization process. They capture features of query plans and complex schemas to assist traditional optimizers. These algorithms offer innovative and effective solutions in areas such as cost modeling, join optimization, plan generation, and query rewriting. This study reviews recent achievements and developments in four main categories of learned query optimization algorithms. Future research directions are also discussed, aiming to provide a comprehensive understanding of the current state of research and to support further investigation in this field. |
| Key words: query optimization artificial intelligence reinforcement learning database system data management |
