Large Language Models in Database Management System Optimization: A Survey
Sikha S. Bagui, Marta Malagutti, Riccardo Morelli, Michael Tamascelli · ACM Transactions on Intelligent Systems and Technology · 2026
Large language models have demonstrated remarkable capabilities in reasoning, code generation, and knowledge integration, making them increasingly relevant across a range of data-intensive applications. Their potential to interpret natural language, synthesize domain knowledge, and adapt to complex tasks has recently motivated research on their use within database management systems to enhance performance and automation. This survey reviews how these models have been employed to support core optimization tasks in modern databases, including query processing, system tuning, indexing, and performance diagnostics. We examine the methodological foundations, such as fine-tuning and RAG, and analyze how these approaches complement or extend traditional optimization components. The survey concludes by identifying key limitations and research gaps, outlining opportunities for advancing the integration of large language models into database system optimization.