QueryBrew introduces a system-agnostic approach to SQL-to-SQL query optimization, enabling automatic query rewriting across different database systems without requiring system-specific rules. The framework uses a novel intermediate representation called the QueryBrew IR to capture semantic equivalence between queries, allowing optimization techniques to be applied universally. Unlike traditional optimizers tied to specific engines like PostgreSQL or MySQL, QueryBrew operates independently of the target database, making it portable and adaptable.

The system leverages a combination of equivalence rules and cost-based search to explore the space of semantically equivalent queries, selecting the most efficient version for execution. Evaluation on standard benchmarks including TPC-H and JOB demonstrates significant performance improvements across multiple database backends. QueryBrew's architecture separates logical optimization from physical execution, enabling reuse of optimization logic.

This work represents a step toward portable, cross-platform query optimization, reducing the need for per-system tuning.