Relation-Aware Bayesian Optimization of DBMS Configurations Guided by Affinity Scores

TL;DR

RelTune tackles the challenge of DBMS parameter tuning by explicitly modeling inter-parameter dependencies through a Relational Graph learned via a GNN, and by guiding Bayesian optimization with an Affinity Score that leverages proximity to high-performing configurations. It uses an LLM to extract parameter descriptions, constructs instance-level graphs, and optimizes in a compact latent space with a hybrid objective that blends surrogate predictions and relational proximity. Empirical results on MySQL and PostgreSQL across diverse workloads show faster convergence and superior throughput/latency trade-offs compared to state-of-the-art BO-based and RL-based baselines, with validation of the affinity signal and robust ablations. The approach offers a principled, scalable pathway for holistic DBMS configuration tuning and paves the way for real-time adaptive tuning in heterogeneous systems.

Abstract

Database Management Systems (DBMSs) are fundamental for managing large-scale and heterogeneous data, and their performance is critically influenced by configuration parameters. Effective tuning of these parameters is essential for adapting to diverse workloads and maximizing throughput while minimizing latency. Recent research has focused on automated configuration optimization using machine learning; however, existing approaches still exhibit several key limitations. Most tuning frameworks disregard the dependencies among parameters, assuming that each operates independently. This simplification prevents optimizers from leveraging relational effects across parameters, limiting their capacity to capture performancesensitive interactions. Moreover, to reduce the complexity of the high-dimensional search space, prior work often selects only the top few parameters for optimization, overlooking others that contribute meaningfully to performance. Bayesian Optimization (BO), the most common method for automatic tuning, is also constrained by its reliance on surrogate models, which can lead to unstable predictions and inefficient exploration. To overcome these limitations, we propose RelTune, a novel framework that represents parameter dependencies as a Relational Graph and learns GNN-based latent embeddings that encode performancerelevant semantics. RelTune further introduces Hybrid-Score-Guided Bayesian Optimization (HBO), which combines surrogate predictions with an Affinity Score measuring proximity to previously high-performing configurations. Experimental results on multiple DBMSs and workloads demonstrate that RelTune achieves faster convergence and higher optimization efficiency than conventional BO-based methods, achieving state-of-the-art performance across all evaluated scenarios.

Figures (7)

• Figure 1: Throughput comparison of correlated (left) and independent (right) parameter pairs.
• Figure 2: Overview Architecture of RelTune
• Figure 3: Performance improvement for MySQL YCSB A,B,E and F.
• Figure 4: Performance improvement for PostgreSQL TPC-C and TPC-H.
• Figure 5: Correlation between Affinity Score and Performance Metrics for MySQL.