SEFRQO: A Self-Evolving Fine-Tuned RAG-Based Query Optimizer

2025-08-24

AI Summary

To address the cold-start problem and poor adaptability to workload/schema changes in traditional learned query optimizers (LQOs), this paper proposes a retrieval-augmented generation (RAG)-based self-evolving query optimization framework. Our method integrates large language models’ (LLMs) in-context learning capability with real execution feedback via a dynamic prompting mechanism: RAG retrieves semantically similar queries and historical execution records to construct context-aware prompts, while supervised and reinforcement fine-tuning jointly train the LLM to generate high-quality optimization directives—enabling continuous evolution without frequent retraining. Evaluated on the CEB and Stack benchmarks, our approach reduces query latency by 65.05% and 93.57%, respectively, compared to PostgreSQL, significantly outperforming state-of-the-art learned optimizers. The key contribution lies in the first integration of contextual LLM reasoning with empirical execution signals within a RAG-driven, self-updating optimization pipeline.

Technology Category

• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP
• Data Mining & Knowledge Management: Intelligent Query Processing

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Querying, indexing, and retrieval in Web-related graphs

Abstract

Query optimization is a crucial problem in database systems that has been studied for decades. Learned query optimizers (LQOs) can improve performance over time by incorporating feedback; however, they suffer from cold-start issues and often require retraining when workloads shift or schemas change. Recent LLM-based query optimizers leverage pre-trained and fine-tuned LLMs to mitigate these challenges. Nevertheless, they neglect LLMs' in-context learning and execution records as feedback for continuous evolution. In this paper, we present SEFRQO, a Self-Evolving Fine-tuned RAG-based Query Optimizer. SEFRQO mitigates the cold-start problem of LQOs by continuously learning from execution feedback via a Retrieval-Augmented Generation (RAG) framework. We employ both supervised fine-tuning and reinforcement fine-tuning to prepare the LLM to produce syntactically correct and performance-efficient query hints. Moreover, SEFRQO leverages the LLM's in-context learning capabilities by dynamically constructing prompts with references to similar queries and the historical execution record of the same query. This self-evolving paradigm iteratively optimizes the prompt to minimize query execution latency. Evaluations show that SEFRQO outperforms state-of-the-art LQOs, achieving up to 65.05% and 93.57% reductions in query latency on the CEB and Stack workloads, respectively, compared to PostgreSQL.

Problem

• Addresses cold-start issues in learned query optimizers
• Mitigates workload shift and schema change retraining needs
• Leverages execution feedback for continuous optimizer evolution

Innovation

• Self-evolving RAG framework for continuous learning
• Combines supervised and reinforcement fine-tuning for optimization
• Dynamic in-context learning with historical execution references