Extend Andromeda beyond configuration debugging to SQL rewriting and index tuning

Investigate methods to extend Andromeda, the retrieval-augmented large language model framework for automatic DBMS configuration debugging, to support additional database optimization tasks, specifically SQL rewriting and index tuning, and to generate actionable recommendations for these tasks.

Background

Andromeda is introduced as a retrieval-augmented LLM framework that diagnoses misconfigured DBMS knobs and recommends corrective settings by leveraging heterogeneous sources such as historical questions, DBMS manuals, and telemetry data.

The authors note that while Andromeda effectively addresses configuration debugging, broader database optimization tasks like SQL rewriting and index tuning remain outside its current scope, motivating an explicit extension to cover these additional problem types.