Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

λ-Tune: Harnessing Large Language Models for Automated Database System Tuning

Victor Giannankouris, Immanuel Trummer

Abstract

We introduce λ-Tune, a framework that leverages Large Language Models (LLMs) for automated database system tuning. The design of λ-Tune is motivated by the capabilities of the latest generation of LLMs. Different from prior work, leveraging LLMs to extract tuning hints for single parameters, λ-Tune generates entire configuration scripts, based on a large input document, describing the tuning context. λ-Tune generates alternative configurations, using a principled approach to identify the best configuration, out of a small set of candidates. In doing so, it minimizes reconfiguration overheads and ensures that evaluation costs are bounded as a function of the optimal run time. By treating prompt generation as a cost-based optimization problem, λ-Tune conveys the most relevant context to the LLM while bounding the number of input tokens and, therefore, monetary fees for LLM invocations. We compare λ-Tune to various baselines, using multiple benchmarks and PostgreSQL and MySQL as target systems for tuning, showing that λ-Tune is significantly more robust than prior approaches.

λ-Tune: Harnessing Large Language Models for Automated Database System Tuning

Abstract

We introduce λ-Tune, a framework that leverages Large Language Models (LLMs) for automated database system tuning. The design of λ-Tune is motivated by the capabilities of the latest generation of LLMs. Different from prior work, leveraging LLMs to extract tuning hints for single parameters, λ-Tune generates entire configuration scripts, based on a large input document, describing the tuning context. λ-Tune generates alternative configurations, using a principled approach to identify the best configuration, out of a small set of candidates. In doing so, it minimizes reconfiguration overheads and ensures that evaluation costs are bounded as a function of the optimal run time. By treating prompt generation as a cost-based optimization problem, λ-Tune conveys the most relevant context to the LLM while bounding the number of input tokens and, therefore, monetary fees for LLM invocations. We compare λ-Tune to various baselines, using multiple benchmarks and PostgreSQL and MySQL as target systems for tuning, showing that λ-Tune is significantly more robust than prior approaches.

Paper Structure

This paper contains 25 sections, 3 theorems, 2 equations, 8 figures, 5 tables, 4 algorithms.

Table of Contents

  1. Introduction
  2. $\lambda$-Tune
  3. Prompt Generation
  4. Prompt Template
  5. Workload Compression
  6. ILP Formulation
  7. Configuration Selector
  8. Configuration Evaluator
  9. Evaluating Configurations
  10. Cost Model for Query Scheduling
  11. Optimizing Query Order
  12. Query Clustering
  13. Experimental Evaluation
  14. Experimental Setup
  15. Comparison to Baselines
  16. ...and 10 more sections

Key Result

Theorem 4.1

The total tuning time (excluding reconfiguration overheads) is in $O(k\cdot \alpha\cdot C_{best})$, where $C_{best}$ is the execution time of the best configuration returned by the LLM, for $\alpha \ge 2$.

Figures (8)

  • Figure 1: $\lambda$-Tune Architecture
  • Figure 2: $\lambda$-Tune Configuration Evaluation
  • Figure 3: Scenario 1: Baselines do not Create Indexes (Pure Parameter Tuning), Default Indexes Available
  • Figure 4: Scenario 2: Baselines Create Indexes, no Indexes are Created by Default
  • Figure 5: Query Execution Times (TPC-H 1GB, Postgres): $\lambda$-Tune vs Default Configuration
  • ...and 3 more figures

Theorems & Definitions (3)

  • Theorem 4.1
  • Theorem 5.1
  • Theorem 5.2