Reproduce “[Experiment, Analysis, and Benchmark] Systematic Evaluation of Plan-based Adaptive Query Processing” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Pei Mu, Anderson Chaves Carniel, Antonio Barbalace, Amir Shaikhha
A systematic evaluation of plan-based Adaptive Query Processing (AQP) on both on-disk (PostgreSQL) and in-memory (DuckDB) database systems identifies that performance benefits for on-disk systems are driven by join reordering, whereas in-memory systems gain from dynamic cardinality refinement. The research provides a module-level analysis of AQP components, quantifying their specific contributions and overheads in diverse environments.
Key insights
- The primary drivers of performance gains for plan-based AQP are architecture-dependent; in-memory systems benefit from dynamic cardinality refinement, while on-disk systems gain from heuristic join reordering.
- Cardinality updating mechanisms (monitor module) in on-disk DBMSs introduce I/O overhead that can negate the benefits of improved cardinality estimates, unlike in in-memory systems.
- The process of plan splitting itself introduces measurable performance overhead in both on-disk and in-memory systems, suggesting potential areas for optimization in splitting strategies.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.