Bootstrapping Learned Cost Models with Synthetic SQL Queries
This work addresses the need for realistic workloads in database testing and optimization, offering an incremental improvement in efficiency for training cost models.
The paper tackles the problem of training learned cost models for SQL query performance prediction by using synthetic data generation techniques to create high-quality datasets, resulting in a 45% reduction in the number of queries needed compared to competitive approaches.
Having access to realistic workloads for a given database instance is extremely important to enable stress and vulnerability testing, as well as to optimize for cost and performance. Recent advances in learned cost models have shown that when enough diverse SQL queries are available, one can effectively and efficiently predict the cost of running a given query against a specific database engine. In this paper, we describe our experience in exploiting modern synthetic data generation techniques, inspired by the generative AI and LLM community, to create high-quality datasets enabling the effective training of such learned cost models. Initial results show that we can improve a learned cost model's predictive accuracy by training it with 45% fewer queries than when using competitive generation approaches.